Forest Ecology
A.G. Van der Valk
Editor
Forest Ecology
Recent Advances in Plant Ecology
Previously published in Plant Ecology Volume 201, Issue 1, 2009
123
Editor
A.G. Van der Valk
Iowa State University
Department of Ecology,
Evolution and Organismal Biology
141 Bessey Hall
Ames IA 50011-1020
USA
Cover illustration: Cover photo image: Courtesy of Photos.com
All rights reserved.
Library of Congress Control Number: 2009927489
DOI: 10.1007/978-90-481-2795-5
ISBN: 978-90-481-2794-8
e-ISBN: 978-90-481-2795-5
Printed on acid-free paper.
© 2009 Springer Science+Business Media, B.V.
No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical,
photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material
supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work.
springer.com
Contents
Quantitative classification and carbon density of the forest vegetation in Lüliang Mountains of
China
X. Zhang, M. Wang & X. Liang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1–9
Effects of introduced ungulates on forest understory communities in northern Patagonia are modified
by timing and severity of stand mortality
M.A. Relva, C.L. Westerholm & T. Kitzberger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11–22
Tree species richness and composition 15 years after strip clear-cutting in the Peruvian Amazon
X.J. Rondon, D.L. Gorchov & F. Cornejo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23–37
Changing relationships between tree growth and climate in Northwest China
Y. Zhang, M. Wilmking & X. Gou . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
39–50
Does leaf-level nutrient-use efficiency explain Nothofagus-dominance of some tropical rain forests
in New Caledonia?
A. Chatain, J. Read & T. Jaffré . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51–66
Dendroecological study of a subalpine fir (Abies fargesii) forest in the Qinling Mountains, China
H. Dang, M. Jiang, Y. Zhang, G. Dang & Q. Zhang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
67–75
A conceptual model of sprouting responses in relation to fire damage: an example with cork oak
(Quercus suber L.) trees in Southern Portugal
F. Moreira, F. Catry, I. Duarte, V. Acácio & J.S. Silva . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
77–85
Non-woody life-form contribution to vascular plant species richness in a tropical American forest
R. Linares-Palomino, V. Cardona, E.I. Hennig, I. Hensen, D. Hoffmann, J. Lendzion, D. Soto,
S.K. Herzog & M. Kessler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
87–99
Relationships between spatial configuration of tropical forest patches and woody plant diversity in
northeastern Puerto Rico
I.T. Galanes & J.R. Thomlinson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
101–113
Vascular diversity patterns of forest ecosystem before and after a 43-year interval under changing
climate conditions in the Changbaishan Nature Reserve, northeastern China
W. Sang & F. Bai . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
115–130
Gap-scale disturbance processes in secondary hardwood stands on the Cumberland Plateau,
Tennessee, USA
J.L. Hart & H.D. Grissino-Mayer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
131–146
Plurality of tree species responses to drought perturbation in Bornean tropical rain forest
D.M. Newbery & M. Lingenfelder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
147–167
Red spruce forest regeneration dynamics across a gradient from Acadian forest to old field in
Greenwich, Prince Edward Island National Park, Canada
N. Cavallin & L. Vasseur . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
169–180
Distance- and density-dependent seedling mortality caused by several diseases in eight tree species
co-occurring in a temperate forest
M. Yamazaki, S. Iwamoto & K. Seiwa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
181–196
Response of native Hawaiian woody species to lava-ignited wildfires in tropical forests and shrublands
A. Ainsworth & J. Boone Kauffman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
197–209
Evaluating different harvest intensities over understory plant diversity and pine seedlings, in a Pinus
pinaster Ait. natural stand of Spain
J. González-Alday, C. Martínez-Ruiz & F. Bravo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
211–220
Land-use history affects understorey plant species distributions in a large temperate-forest complex,
Denmark
J.-C. Svenning, K.H. Baktoft & H. Balslev . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
221–234
Short-term responses of the understory to the removal of plant functional groups in the cold-temperate
deciduous forest
A. Lenière & G. Houle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
235–245
Host trait preferences and distribution of vascular epiphytes in a warm-temperate forest
A. Hirata, T. Kamijo & S. Saito . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
247–254
Seed bank composition and above-ground vegetation in response to grazing in sub-Mediterranean
oak forests (NW Greece)
E. Chaideftou, C.A. Thanos, E. Bergmeier, A. Kallimanis & P. Dimopoulos . . . . . . . . . . . . . . . . . .
255–265
On the detection of dynamic responses in a drought-perturbed tropical rainforest in Borneo
M. Lingenfelder & D.M. Newbery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
267–290
Changes in tree and liana communities along a successional gradient in a tropical dry forest in
south-eastern Brazil
B.G. Madeira, M.M. Espírito-Santo, S. D’Ângelo Neto, Y.R.F. Nunes, G. Arturo Sánchez Azofeifa,
G. Wilson Fernandes & M. Quesada . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
291–304
Woody plant composition of forest layers: the importance of environmental conditions and spatial
configuration
M. Gonzalez, M. Deconchat & G. Balent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
305–318
The importance of clonal growth to the recovery of Gaultheria procumbens L. (Ericaceae) after
forest disturbance
F.M. Moola & L. Vasseur . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
319–337
Species richness and resilience of forest communities: combined effects of short-term disturbance
and long-term pollution
M.R. Trubina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
339–350
Hurricane disturbance in a temperate deciduous forest: patch dynamics, tree mortality, and coarse
woody detritus
R.T. Busing, R.D. White, M.E. Harmon & P.S. White . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
351–363
Quantitative classification and carbon density of the forest
vegetation in Lüliang Mountains of China
Xianping Zhang Æ Mengben Wang Æ
Xiaoming Liang
Originally published in the journal Plant Ecology, Volume 201, No. 1, 1–9.
DOI: 10.1007/s11258-008-9507-x Springer Science+Business Media B.V. 2008
Abstract Forests play a major role in global carbon
(C) cycle, and the carbon density (CD) could reflect
its ecological function of C sequestration. Study on
the CD of different forest types on a community scale
is crucial to characterize in depth the capacity of
forest C sequestration. In this study, based on the
forest inventory data of 168 field plots in the study
area (E 111300 –113500 , N 37300 –39400 ), the
forest vegetation was classified by using quantitative
method (TWINSPAN); the living biomass of trees
was estimated using the volume-derived method; the
CD of different forest types was estimated from the
biomass of their tree species; and the effects of biotic
and abiotic factors on CD were studied using a
multiple linear regression analysis. The results show
that the forest vegetation in this region could be
classified into 9 forest formations. The average CD of
the 9 forest formations was 32.09 Mg ha-1 in 2000
and 33.86 Mg ha-1 in 2005. Form. Picea meyeri had
the highest CD (56.48 Mg ha-1), and Form. Quercus
liaotungensis ? Acer mono had the lowest CD
(16.14 Mg ha-1). Pre-mature forests and mature
forests were very important stages in C sequestration
among four age classes in these formations. Forest
densities, average age of forest stand, and elevation
had positive relationships with forest CD, while slope
location had negative correlation with forest CD.
Keywords TWINSPAN Carbon density
Volume-derived method Forest vegetation
China
Introduction
X. Zhang M. Wang (&)
Institute of Loess Plateau, Shanxi University,
580 Wucheng Road, Taiyuan 030006,
People’s Republic of China
e-mail: mbwang@sxu.edu.cn
X. Zhang
Shanxi Forestry Vocational Technological College,
Taiyuan 030009, People’s Republic of China
X. Liang
Guandi Mountain State-Owned Forest Management
Bureau of Shanxi Province, Jiaocheng, Lishi 032104,
People’s Republic of China
Forests play a major role in global carbon (C) cycle
(Dixon et al. 1994; Wang 1999) because they store
80% of the global aboveground C of the vegetation
and about 40% of the soil C and interact with
atmospheric processes through the absorption and
respiration of CO2 (Brown et al. 1999; Houghton
et al. 2001a, b; Goodale and Apps 2002). Enhancing
C sequestration by increasing forestland area has
been suggested as an effective measure to mitigate
elevated atmospheric carbon dioxide (CO2) concentration and hence contribute toward the prevention of
global warming (Watson 2000). Recent researches
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_1
1
2
focus mainly on carbon storage of forest ecosystem
on landscape or regional scale (Fang et al. 2001;
Hiura 2005; Zhao and Zhou 2006). Many studies
have shown that the C sequestration abilities of
different forests change considerably, which can be
well explained by their CD values (Wei et al. 2007;
Hu and Liu 2006). Meanwhile the C storage of forests
may change substantially with forest ecosystems on a
community scale. This type of moderate-scale
research into the C storage of forests, however, has
been rarely conducted.
Many methods have been used to estimate the
biomass of forest vegetation (Houghton et al. 2001a,
b). Among them, the volume-derived method has
been commonly used (Brown and Lugo 1984; Fang
et al. 1996; Fang and Wang 2001). Forest volume
production reflects the effects of the influencing
factors, such as the forest type, age, density, soil
condition, and location. The forest CD estimated
from forest biomass will also indicate these effects.
Zhou et al. (2002) and Zhao and Zhou (2005)
improved the volume-derived method by hyperbolic
function, but the method has not been used to
estimate forest CD on the moderate scale.
The Lüliang Mountains is located in the eastern
part of the Loess Plateau in China, where soil and
water losses are serious. To improve ecological
environment there, the Chinese government has been
increasing forestland by carrying out ‘‘The ThreeNorth Forest Shelterbelt Program,’’ ‘‘The Natural
Forest Protection Project,’’ and ‘‘The Conversion of
Cropland to Forest Program’’ since 1970s. Previous
studies on the forest vegetation in this region focus
mainly on the qualitative description of its distribution pattern (The Editing Committee of Shanxi Forest
1984). The objectives of this study were (1) to
classify the forest vegetation on Lüliang Mountains
using quantitative classification method (TWINSPAN)
(Zhang et al. 2003; Zhang 2004); (2) to estimate the
CD of different forest types through biomass based
on the modified volume-derived method (Zhou et al.
2002) and to clarify the distribution pattern of forest
CD in this region; and (3) to quantify the contribution
of biotic and abiotic factors (including average forest
age, density, soil thickness, elevation, aspect, and
slope) to forest CD based on a multiple linear
regression analysis. The results would provide basic
data for further study of forest C storage pattern in
this region.
A.G. Van der Valk (ed.)
Methods
Study region
The study was conducted in the middle-north of
Lüliang Mountains (E 111300 –113500 , N 37300
–39400 ) with its peak (Xiaowen Mountain) 2831 m
above sea level (asl). The temperate terrestrial climate
is characterized by a warm summer, a cold winter, and
a short growing season (90–130 days) with a mean
annual precipitation of 330–650 mm and a mean
annual temperature of 8.5C (min. monthly mean of
-7.6C in January and max. monthly mean of 22.5C
in July). The soils from mountain top to foot are
mountain meadow soil, mountain brown soil, mountain alfisol cinnamon soil, and mountain cinnamon
soil (The Editing Committee of Shanxi Forest 1984).
There are two national natural reserves in this
region with Luya Mountain National Nature Reserve
in the north and Pangquangou National Nature
Reserve in the south, in which Crossoptlon mantchuricum (an endangered bird species), Larix
principis-rupprechtii forest, and Picea spp. (P. meyeri and P. wilsonii) forest are the key protective
targets.
Based on the system of national vegetation
regionalization, this area was classified into the
warm-temperate deciduous broad-leaved forest zone.
With the elevation rising, vegetation zone are,
respectively, deciduous broad-leaved forest, needlebroad-leaved mixed forest, cold-temperate coniferous
forest, and subalpine scrub-meadow.
Data collection
The forest inventory data from a total of 168 field
plots in 2000 and 2005 were used in this study. These
permanent plots (each with an area of 0.0667 ha)
were established systematically based on the grid of
4 km 9 4 km across the forestland of 2698.85 km2
in 1980s under the project of the forest survey of the
Ministry of Forestry of P. R. China (1982), in which
the data, such as tree species, diameter at breath
height of 1.3 m (DBH), the average height of the
forest stand, and the average age of the forest stand
had been recorded along with the data of location,
elevation, aspect, slope degree, slope location, and
soil depth. For trees with C5 cm DBH, the values of
their DBH were included in the inventory.
Forest Ecology
3
TWINSPAN classification
Table 1 Parameters of biomass calculation for dominant
species in this study
A total of 26 tree species had been recorded in the
168 plots. The importance values (IV) for every tree
species in each plot were calculated using the
following formula:
Species
IV ¼ ðRelative density þ Relative dominance
þ Relative frequencyÞ=300
where relative density is the ratio of the individual
number for a tree species over the total number for
all tree species in a plot, relative dominance is the
ratio of the sum of the basal area for a tree species
over the total basal area of all tree species in a plot,
and the relative frequency is the percentage of the
plot number containing a tree species over the total
plot number (168) in this inventory. Based on the
matrix of IVs of 26 9 168 (species 9 plots), the
forest vegetation can be classified into different
formations using the two-way indicator-species
analysis (TWINSPAN) (Hill 1979).
The volume production of an individual tree could be
obtained in the volume table (Science and Technology Department of Shanxi Forestry Bureau 1986)
according to its DBH. The volume of a species (V)
was the sum of its individual tree’s volume in a plot.
The total living biomass (B) (Mg ha-1) of a species
in a plot was calculated as:
V
a þ bV
a
b
n
R2
Larix principis-rupprechtii
0.94
0.0026
34
0.94
Pinus tabulaeformis
0.32
0.0085
32
0.86
Picea meyeri
0.56
0.0035
26
0.85
Platycladus orientalis
1.125
0.0002
21
0.97
Pinus armandii
0.542
0.0077
17
0.73
Populus davidiana
0.587
0.0071
21
0.92
Betula platyphylla
0.975
0.001
14
0.91
Quercus liaotungensis
0.824
0.0007
48
0.92
CD ¼ B Cc
ð2Þ
where B is the total living biomass of tree species in a
plot; CC is the average carbon content of dry matter,
which is assumed to be 0.5, though it varies slightly
for different vegetation (Johnson and Sharpe 1983;
Zhao and Zhou 2006).
Effects of influencing factors
Estimation of biomass and CD
B¼
Parameters in equation
ð1Þ
where V represents the total volume (m3 ha-1) of a
species in a plot, a (0.32–1.125) and b (0.0002–0.001)
are constants (Zhou et al. 2002). The constants for
most of the tree species in this study were developed
by Zhao and Zhou in 2006 (Table 1).
In regard to companion tree species in this study,
their biomass estimation was based on the parameters
of above known species according to their morphological similarity, i.e., Pinus bungeana is referred to
the parameters of Pinus armandii; Ulmus pumilla and
Tilia chinensis to those of Quercus liaotungensis; and
Acer mono and the rest of broad-leaved species to
those of Populus davidiana.
Forest CD (Mg ha-1) was calculated as:
The qualitative data of the aspect and slope location
were first transformed into quantitative data to
quantify their effects on forest CD. According to
the regulations of the forest resources inventory by
the Ministry of Forestry (1982), the aspect data were
transformed to eight classes starting from north (from
338 to 360 plus from 0 to 22), turning clockwise,
and taking every 45 as a class: 1 (338–22, north
aspect), 2 (23–67, northeast aspect), 3 (68–112,
east aspect), 4 (113–157, southeast aspect), 5
(158–202, south aspect), 6 (203–247, southwest),
7 (248–292, west aspect), and 8 (293–337,
northwest aspect). The slope locations in the mountains were transformed to 6 grades: 1 (the ridge), 2
(the upper part), 3 (the middle part), 4 (the lower
part), 5 (the valley), and 6 (the flat).
A multiple linear regression model was used to
analyze the effects of biotic and abiotic factors on
forest CD, assuming a significant effect if the
probability level (P) is \0.05:
Y^ ¼ a þ b1 X1 þ b2 X2 þ b3 X3 þ. . . þ bk Xk h
ð3Þ
where a is a constant, b1, b2, b3, and bk are regression
coefficients. Y^ represents CD and X1, X2, X3, X4, X5,
4
A.G. Van der Valk (ed.)
X6, and X7 represent forest density (X1), average age
(X2), elevation (X3), slope location (X4), aspect (X5),
slope degree (X6), and soil depth (X7) in each plot,
respectively. Here forest density is the individual
number of all tree species per area in a plot, and
forest age is the average age of dominant trees in the
plot.
168 plots
2nd level
3rd level
4th level
Results
1
(12)
Forest formations from TWINSPAN
According to the 4th level results of TWINSPAN
classification, the 168 plots were classified into 9
formations (Table 2), which were named according
to Chinese Vegetation Classification system (Wu
1980). The dendrogram derived from TWINSPAN
analysis is shown in Fig. 1. The basic characteristics
of species composition, structure along with its
environment for each formation are described as
follows:
1.
Form. Larix principis-rupprechtii (Form. 1 for
short, the same thereafter): L. principisrupprechtii was the dominant tree species of
the cold-temperate coniferous forest in north
China. It grew relatively faster with fine timber.
Therefore it was a very important silvicultural
tree species at middle-high mountains in this
region. This type of forest distributed vertically
from 1610 m to 2445 m above sea level, and
2
(20)
3
(17)
4
(24)
5 6
(35)(26)
7
(11)
8
(5)
9
(18)
Fig. 1 Dendrogram derived from TWINSPAN analysis. Note:
1. Form. Larix principis-rupprechtii; 2. Form. Picea meyeri; 3.
Form. Betula platyphylla; 4. Form. Populus davidiana; 5. Form.
Pinus tabulaeformis; 6. Form. Pinus tabulaeformis ? Quercus
liaotungensis; 7. Form. Quercus liaotungensis; 8. Form. Pinus
bungeana ? Platycladus orientalis, and 9. Form. Quercus
liaotungensis ? Acer mono. The number of plots for each
formation is shown between the brackets
2.
3.
common companion species were Picea meyeri
and P. wilsonii in the tree layer.
Form. Picea meyeri (Form. 2): P. meyeri forest
belonged to cold-temperate evergreen coniferous
forest. Its ecological amplitude was relatively
narrow with a range of vertical distribution from
1860 m to 2520 m. Betula platyphylla and Picea
wilsonii appeared commonly in this forest.
Form. Betula platyphylla (Form. 3): B. platyphylla was one of main tree species in this region
and occupied the land at moderate elevation
(1700–2200 m). In the tree layer, Populus
Table 2 The structure characteristics of 9 forest formations and their environmental factors
Form
Density (No./ha)
Age (Year)
Coverage (%)
Slope location
Elevation (m)
Slope ()
Aspect
Soil depth (cm)
1
849.3 ± 121.8
40.0 ± 5.4
54 ± 8.7
2.7 ± 0.1
1610–2445
19.1 ± 1.1 4.1 ± .6
2
869.6 ± 179.1
55.4 ± 4.8
62 ± 8.3
2.3 ± 0.2
1860–2520
19.6 ± 2.2 4.7 ± 0.6 50.6 ± 5.9
56.4 ± 5.1
3
774.3 ± 57.8
45.5 ± 5.3
45 ± 4.1
2.6 ± 0.2
1700–2200
21.6 ± 1.9 4.2 ± 0.8 48.7 ± 3.3
4
1071.9 ± 124.4
31.6 ± 2.6
41 ± 6.3
3.5 ± 0.2
1350–1997
23.0 ± 1.6 4.1 ± 0.6 49.2 ± 6.2
5
770.9 ± 139.7
54.7 ± 2.6
49 ± 5.7
2.9 ± 0.2
1360–2010
23.9 ± 2.2 2.9 ± 0.5 41.0 ± 4.1
6
756.2 ± 87.7
60.9 ± 3.7
46 ± 4.2
2.6 ± 0.2
1235–1820
29.4 ± 2.3 3.7 ± 0.4 34.2 ± 4.1
7
731.3 ± 154.7
56.8 ± 6.2
46 ± 7.4
3.0 ± 0.3
1452–2010
25.9 ± 2.1 3.4 ± 0.8 53.2 ± 3.7
8
1589.2 ± 616.2
53.8 ± 3.8
41 ± 2.5
2.6 ± 0.5
1250–1270
26.6 ± 3.5 3.6 ± 0.7 34.0 ± 7.1
9
910.3 ± 136.8
51.3 ± 4.6
51 ± 7.3
3.4 ± 0.2
1350–1660
23.2 ± 2.5 4.8 ± 0.5 39.4 ± 4.4
Note: 1. Form. Larix principis-rupprechtii; 2. Form. Picea meyeri; 3. Form. Betula Platyphylla; 4. Form. Populus davidiana; 5. Form.
Pinus tabulaeformis; 6. Form. Pinus tabulaeformis ? Quercus liaotungensis; 7. Form. Quercus liaotungensis; 8. Form. Pinus
bungeana ? Platycladus orientalis; 9. Form. Quercus liaotungensis ? Acer mono
Forest Ecology
5.
6.
7.
8.
9.
Biomass
According to the national guidelines for forest
resource survey (The Ministry of Forestry 1982),
each forest formation can be divided into five age
classes (young, mid-aged, pre-mature, mature, and
post-mature). Since there was only one plot where the
120
2000
2005
100
-1
Mean biomass ( Mgha )
4.
davidiana and Larix principis-rupprechtii were
the companion species.
Form. Populus davidiana (Form. 4): P. davidiana
was a pioneer tree species in the north secondary
forest. This forest appeared at moderate elevation
(1350–1997 m) and on southerly aspect. Tree
species were plentiful in it, including Pinus
tabulaeformis, Quercus liaotungensis, and so on.
Form. Pinus tabulaeformis (Form. 5): P. tabulaeformis (Chinese pine) was a main dominant
tree species of the warm-temperate coniferous
forest in north China. The Chinese pine forest
was a dominant forest type in Shanxi Province
(The Editing Committee of Shanxi Forest 1984).
In the study region, it occupied the land at
moderate elevation (1360–2010 m).
Form. Pinus tabulaeformis ? Quercus liaotungensis (Form. 6): this forest was present at low to
moderate elevation (1200–1800 m) on southfaced aspect.
Form. Quercus liaotungensis (Form. 7): the
Q. liaotungensis forest was a typical warmtemperate deciduous broad-leaved forest and a
main broad-leaved forest type in north China.
Q. liaotungensis mainly distributed at middlelow elevation (1400–2000 m) in the middlenorth of Lüliang Mountains.
Form. Pinus bungeana ? Platycladus orientalis
(Form. 8): there was relatively a few Pinus
bungeana ? Platycladus orientalis mixed forest
appearing at the lower elevation of 1200 m on
northerly aspect where environmental condition
was characterized by drought, infertility, and
cragginess.
Form. Quercus liaotungensis ? Acer mono (Form.
9): in the low elevation (1300–1660 m), Q. liaotungensis was always mixed with other broadleaved tree species, such as Acer mono, Prunus
armeniaca, and so on. Most of these trees were
light-demanding and drought-tolerant species.
5
80
60
40
20
0
1
2
3
4
5
6
7
8
9
Forest formation
Fig. 2 The mean biomass of each formation in 2000 and 2005
(Mg ha-1)
post-mature age class forest occurred, which
belonged to P. davidiana Form., the rest of plots fell
into four age classes (Fig. 3).
According to Eq. 1 and the parameters of each
species (Table 1; Zhao and Zhou 2006), the biomass
of each age class for 9 formations were calculated,
and the average biomasses of each formation are
shown in Fig. 2. The average biomass in 2005 was
slightly higher than that in 2000.
There was a wide range of change in the values of
mean biomass among the 9 formations. For instance, in
2005, the highest value of biomass (112.97 Mg ha-1)
was observed in Form. 2; next to Form. 2 were Form. 6
(85.51 Mg ha-1) and Form. 1 (83.49 Mg ha-1); in the
middle level were Form. 3 (60.64 Mg ha-1), Form. 5
(60.61 Mg ha-1), and Form. 7 (65.14 Mg ha-1); and
the lower values of biomass were found in Form. 4
(50.80 Mg ha-1), Form. 8 (43.69 Mg ha-1), and
Form. 9 (46.12 Mg ha-1).
Carbon density
The overall average values of carbon density (CD) for
the 9 formations were 32.09 Mg ha-1 in 2000 and
33.86 Mg ha-1 in 2005, respectively, and the average
values of CD for these formations ranged from
23.06 Mg ha-1 for Form. 9 to 56.48 Mg ha-1 for
Form. 2.
The CD among different age classes changed
considerably (Fig. 3), and showed an increased trend
6
A.G. Van der Valk (ed.)
Y^ ¼
Carbon density (Mg ha -1 )
100
Young
Middle-aged
Premature
Mature
80
60
40
20
0
1
2
3
4
5
6
7
8
9
Forest formation
Fig. 3 The carbon density of 9 forest formations in Lüliang
Mt. in 2005 (Mg ha-1). Note: There is no mature age class in
Form. 1, and there is only a single middle-aged class in Form. 8
from the young class to pre-mature or mature class in
most forest formations. The extremely low amount of
CD in the pre-mature forest of Form. 4 resulted from
the low biomass accumulation, which may be caused,
according to field observations, by (1) the insect
infestation which had occurred and led to the death of
some trees in plots 155 and 164, and (2) the droughty
habitats on southerly aspect where these two plots
were located, and the wilt of some tree species like
Populus davidiana was found.
In Form. 2, Form. 6, or Form. 7 the CD of mature
forest was lower than that of the pre-mature forest
due to the fact: Larix principis-rupprechtii, Picea
meyeri, and Pinus tabulaeformis were main timber
tree species in study region, and some of the mature
trees in these formations may have been illegally cut
down for timber use by some local residents.
Nevertheless, from the total percentage of the CD
of pre-mature and mature classes over the total CD of
all classes of each formation, it was found that the CD
in these two classes accounted for 74.9% in Form. 2,
70.6% in Form. 3, 60.8% in Form. 5, 63.2% in Form.
6, 58.3% in Form. 7, and 70.0% in Form. 9. This
indicated that pre-mature and mature forests were
very important C sequestration stages in most
formations.
Effects of biotic and abiotic factors on forest CD
Due to lack of some environmental data in some
plots, a total of 157 plot data was used for regression
analysis. Based on Eq. 3, a multiplelinear regression
equation between the forest CD Y^ and influencing
factors was established:
17:687 þ 0:17X1 þ 0:108X2 þ 0:019X3
1:182X4
ð4Þ
The partial correlation coefficients were 0.475
(P \ 0.01) for forest density (X1), 0.288 (P \ 0.01)
for average age (X2), 0.261(P \ 0.01) for elevation
(X3) and -0.178 (P \ 0.05) for slope location (X4),
respectively. It indicated that forest density, average
age of forest stand and altitude had positive correlation with CD; whereas slope location had negative
correlation with CD. And aspect (X5), slope degree
(X6), and soil depth (X7) had no significant relationship with the CD. This suggested that the CD rose
with the increase of forest density, average age, and
altitude; and it decreased with the slope location
change from 1 (the ridge) to 6 (the flat). The biggest
partial correlation coefficient for forest density indicated that forest density had a stronger effect on the
CD than the other factors.
Discussions
The results of quantitative classification (TWINSPAN) clearly reflected the vertical distribution
patterns of forest vegetation in Lüliang Mountains.
The warm-temperate deciduous broad-leaved forest
(Form. Quercus liaotungensis ? Acer mono) was
distributed in the low mountain area, and Pinus
bungeana ? Platycladus orientalis mixed forest was
located in this altitude range on the southern aspect
where the habitat was droughty and infertile. The
warm-temperate coniferous forest (Form. Pinus
tabulaeformis) and the warm-temperate needlebroad-leaved mixed forest (Form. Pinus tabulaeformis
? Quercus liaotungensis) were present in the lowerto-middle mountain area. And Quercus liaotungensis
forest also occupied this range. Deciduous broadleaved forests (Form. Populus davidiana and Form.
Betula platyphylla) occupied the middle-to-high
mountain range. Cold-temperate coniferous forests
(Form. Larix principis-rupprechtii and Form. Picea
meyeri) were distributed in the middle-to-high mountain area, in which the distribution range of Form. 1
was wider than Form. 2.
Considered together, the distribution patterns and
biomass estimates of the forests in Lüliang Mountains
revealed that the biomass tended to increase with the
Forest Ecology
altitude rising. Of the 5 coniferous formations (including coniferous and broad-leaved mixed formations),
the biomass increased from 43.69 Mg ha-1 for Form.
8 (1200 m asl), 60.61 Mg ha-1 for Form. 5 (1360–
2010 m asl), 85.52 Mg ha-1 for Form. 6 (1200–
1800 m asl), 83.49 Mg ha-1 for Form. 1 (1610–
2445 m asl) to 112.97 Mg ha-1 for Form. 2 (1860–
2520 m asl). Of the 4 broad-leaved formations, the
biomass increased from 46.12 Mg ha-1 for Form. 9
(1300–1660 m asl) and 50.80 Mg ha-1 for Form. 4
(1350–1997 m asl) to 65.14 Mg ha-1 for Form. 7
(1400–2000 m asl) and 60.64 Mg ha-1 for Form. 3
(1700–2200 m asl). In addition, the average biomass
(79.12 Mg ha-1) of the 5 coniferous formations was
greater than that (53.91 Mg ha-1) of the 4 broadleaved formations.
The average CD of forest vegetation of Lüliang
Mountains was 33.86 Mg ha-1 in 2005. It was lower
than the average level of 41.938 Mg ha-1 (Wang
et al. 2001a, b), 44.91 Mg ha-1 (Fang et al. 2001), or
41.32 Mg ha-1 (Zhao and Zhou 2006) estimated for
all forests in China. The lower CD in Lüliang
Mountains can be explained by (1) low annual
precipitation of 330–650 mm in this area (The
Editing Committee of Shanxi Forest 1984) and (2)
large proportion of young, middle-age, and premature forests (80%) and small proportion of mature
and post-mature forests (20%) (Liu et al. 2000).
Different forest formations had various ability of
carbon sequestration. In this study, the average CD
(56.48 Mg ha-1) of Form. Picea meyeri was higher
than those of other forest formations. This may result
from the higher average individual volume production
of Picea meyeri. According to The Editing Committee
of Shanxi Forest (1984), the average individual
volume production at the age of 60 were
0.0056 m3 year-1 for Picea meyeri, 0.0031 m3
year-1 for Larix principis-rupprechtii, and
0.0030 m3 year-1 for Pinus tabulaeformis, respectively. The average CD (42.76 Mg ha-1) of Form
Pinus tabulaeformis ? Quercus liaotungensis was
close to the average level in China, and this type of
mixed forest could be largely afforested in the lowerto-middle mountain of the Loess Plateau. Most of the
stands of Form. Larix principis-rupprechtii forest
were still at very young stage (at an average age of
40 years for all stands), so the CD (41.75 Mg ha-1) of
this Form. was relatively low. As Wang et al. (2001a,
b) and Zhou et al. (2000) suggested, in the middle-to-
7
higher mountain of the Loess Plateau, subalpine
coniferous tree species, such as Picea meyeri should
be primarily protected because they can sequestrate
more C than other tree species.
Under conditions of global climate change, the
impact of biotic and abiotic factors on forest carbon
density is complex. Many factors have synergistic
effect on forest carbon, and the influencing degree of
those factors is different (Houghton 2002). The
analysis of multiple linear regression showed that
forest density, average age, and elevation had
positive relations with forest CD, and slope location
had negative correlation with it.
In a single species population, the function relationship between mean biomass of individual trees
and density has long been an issue in dispute.
Recently, Enquist and Niklas (2002) put forward that
there is a power function relationship between
biomass (or C) of individual tree and forest density.
Therefore forest density is an important influencing
factor on forest carbon. In this research, the regression
analysis indicated that forest density had significantly
higher effect on carbon density than other factors.
The significant effects of altitude and slope
location on forest CD may be to some extent related
to human disturbance. Along with the elevation rise
or the slope location change from mountain foot to
top, the human activities decreased, and the carbon
accumulation of forest ecosystems increased. Therefore the forest CD tended to increase with elevation
rise or slope location rise.
Due to the fact that the volume-derived method
provides only the parameters of biomass calculation
for dominant species, and lacks the parameters for
companion species, the biomass estimation of companion species were based on the parameters of
known species according to the morphological similarity between the companion species and the known
species in this study (Table 1). This kind of approximation may result in inaccurate CD estimation.
Besides, only the living biomass of trees was
estimated, the biomass of shrubs, herbs, standing
dead wood, and litter on the ground were not taken
into account in this study. As Duvigneaued (1987)
noted that the total litter biomass accounts for 2–7%
of the total biomass of major biomes of the world, so
this study presents primarily the basic CD results of
the forest tree species in this area. Much detailed
work, especially that of the total biomass and carbon
8
storage of every forest formation, needs to be done in
the future.
Conclusion
The forest vegetation in this area was quantitatively
classified into 9 forest formations. They showed
distinctly the vertical distribution patterns along
elevation gradient in Lüliang Mountains. The average
CD was 32.09 Mg ha-1 in 2000 and 33.86 Mg ha-1
in 2005, with the highest CD (56.48 Mg ha-1) in
Form. Picea meyeri and the lowest CD
(16.14 Mg ha-1) in Form. Quercus liaotungensis ?
Acer mon. Pre-mature and mature forests generally
sequestrated more C than young and middle-aged
forests. Forest density, average age of forest stand, and
elevation had significantly positive relationships with
forest CD, and slope location showed negative correlation with forest CD. The forest density had a higher
effect on forest CD than other factors.
Acknowledgments This research was supported by the
National Natural Science Foundation of China (30170150).
We thank Professor Feng Zhang for reviewing earlier drafts of
this article; and anonymous reviewers for valuable comments
on the manuscript.
References
Brown S, Lugo AE (1984) Biomass of tropical forests: a new
estimate based on forest volumes. Science 223:1290–
1293. doi:10.1126/science.223.4642.1290
Brown SL, Schroeder P, Kern JS (1999) Spatial distribution of
biomass in forests of the eastern USA. For Ecol Manag
123:81–90
Dixon RK, Brown S, Houghton RA et al (1994) Carbon pools
and flux of global forest ecosystems. Science 263:185–
190. doi:10.1126/science.263.5144.185
Duvigneaued P (1987) La synthese ecologique (in Chinese,
trans: Li Y). Chinese Science Press, Beijing, China
Enquist BJ, Niklas KJ (2002) Global allocation rules for patterns of biomass portioning in seed plants. Science
295:1517–1520. doi:10.1126/science.1066360
Fang JY, Wang ZM (2001) Forest biomass estimates at
regional and global levels, with special reference to China’s forest biomass. Ecol Res 16:587–592. doi:10.1046/
j.1440-1703.2001.00419.x
Fang JY, Liu GH, Xu SL (1996) Biomass and net production of
forest vegetation in China (in Chinese with English
abstract). Acta Ecol Sin 16:497–508
Fang JY, Chen AP, Peng CH et al (2001) Changes in forest
biomass carbon storage in China between 1949 and 1998.
Science 292:2320–2322. doi:10.1126/science.1058629
A.G. Van der Valk (ed.)
Goodale CL, Apps MJ (2002) Forest sinks in the northern
hemisphere. Ecol Appl 12(3):891–899. doi:10.1890/10510761(2002)012[0891:FCSITN]2.0.CO;2
Hill MO (1979) TWINSPAN—a FORTRAN program for
arranging multivariate data in an ordered two-way table
by classification of the individuals and the attributes.
Department of Ecology and Systematics, Cornell University, Ithaca
Hiura T (2005) Estimation of aboveground biomass and net
biomass increment in a cool temperate forest on a landscape scale. Ecol Res 20:271–277. doi:10.1007/s11284005-0042-0
Houghton RA (2002) Terrestrial carbon sinks-uncertain
explanations. Biologist 49:155–160
Houghton JT, Ding Y, Griggs DJ (eds) (2001a) Climate change
2001: the scientific basis. Contribution of working group I
to the third assessment report of the intergovernmental
panel on climate change. Cambridge University Press,
Cambridge, UK
Houghton RA, Lawrence KT, Hackler JL et al (2001b) The
spatial distribution of forest biomass in the Brazilian
Amazon: a comparison of estimate. Glob Change Biol
7:731–746. doi:10.1046/j.1365-2486.2001.00426.x
Hu HF, Liu GH (2006) Roles of forest management in global
carbon dioxide mitigation (in Chinese with English
abstract). Chin J App Ecol 17(4):709–714
Johnson WC, Sharpe DM (1983) The ratio of total to merchantable forest biomass and its application to the global
carbon budget. Can J Res 13:372–383
Liu GH, Fu BJ, Fang JY (2000) Carbon dynamics of Chinese
forests and its contribution to global carbon balance (in
Chinese with English abstract). Acta Ecol Sin 20(5):733–740
Ministry of Forestry (1982) Standards for forest resources
survey. China Forestry Publishing House, Beijing, China
Science and Technology Department of Shanxi Forestry
Bureau (1986) The compilation of forestry standards in
Shanxi Province. Taiyuan, China
The Editing Committee of Shanxi Forest (1984) Shanxi Forest.
Chinese Forestry Press, Beijing, China, pp 135–136
Wang XK (1999) Study on regional carbon cycle of forest
ecosystem in China. Theses of committee of Synthesis
Investigation of Natural Resources. Chinese Academy of
Science, Beijing, China, p 123
Wang XK, Feng ZW, Ouyang ZY (2001a) The impact of human
disturbance on vegetation carbon storage in forest ecosystems in China (in Chinese with English abstract). For Ecol
Manag 148:117–123. doi:10.1016/S0378-1127(00)00482-5
Wang XK, Feng ZW, Ouyang ZY (2001b) Vegetation carbon
storage and density of forest ecosystems in China(in Chinese with English abstract). Chin J App Eco 12(1):13–16
Watson RT (2000) Land use, land-use change, and forestry: a
special report of the IPCC. Cambridge university Press,
Cambridge, p 377
Wei HD, Ma XQ, Liu AQ et al (2007) Review on carbon cycle
of forest ecosystem (in Chinese with English abstract).
Chin J of Eco-Agric 15(2):188–192
Wu ZY (1980) Chinese vegetation. Science Press, Beijing, China
Zhang JT (2004) Quantitative ecology. Science Press, Beijing,
China
Zhang F, Zhang JT, Zhang F (2003) Pattern of forest vegetation and its environmental interpretation in Zhuweigou,
Forest Ecology
Lishan mountain nature reserve (in Chinese with English
abstract). Acta Ecol Sin 23:421–427
Zhao M, Zhou GS (2005) Estimation of biomass and net primary productivity of major planted forests in China based
on forest inventory data. For Ecol Manag 207:295–313.
doi:10.1016/j.foreco.2004.10.049
Zhao M, Zhou GS (2006) Carbon storage of forest vegetation
in China and its relationship with climatic factors. Clim
Change 74:175–189. doi:10.1007/s10584-006-6775-0
9
Zhou YR, Yu ZL, Zhao SD (2000) Carbon storage and budget
of major Chinese formations (in Chinese with English
abstract). Acta Phytoecol Sin 24:518–522
Zhou GS, Wang YH, Jiang YL, Yang ZY (2002) Estimating
biomass and net primary production from forest inventory
data: a case study of China’s Larix forest. For Ecol Manag
169:149–157. doi:10.1016/S0378-1127(02)00305-5
Effects of introduced ungulates on forest understory
communities in northern Patagonia are modified
by timing and severity of stand mortality
Marı́a Andrea Relva Æ
Christian López Westerholm Æ
Thomas Kitzberger
Originally published in the journal Plant Ecology, Volume 201, No. 1, 11–22.
DOI: 10.1007/s11258-008-9528-5 Springer Science+Business Media B.V. 2008
M. A. Relva (&) T. Kitzberger
Laboratorio Ecotono, INIBIOMA-CONICET,
Universidad Nacional del Comahue, Quintral, 1250, 8400
Bariloche, Argentina
e-mail: arelva@crub.uncoma.edu.ar;
andrearelva@gmail.com
areas not subjected to such removal. Stepwise regression analyses showed that history and severity of tree
mortality strongly influence plant composition and
deer use of plants. For deer use (with pellet counts and
browsing index as response variables), results showed
a positive relationship with degree of stand mortality
and a negative relationship with cover of fallen logs.
Similarly, cover of unpalatable shrub species was
explained by canopy mortality history, whereas cover
of palatable shrub species was positively associated
with severity of canopy mortality. In areas where fallen
logs had been removed, pellet counts were six times
higher than those in control areas. Though total shrub
species cover was similar between log removal and
control areas, proportion of unpalatable shrubs
increased in areas where fallen logs had been removed.
In conclusion, deer use of plants was strongly limited
by tall fallen logs, allowing palatable species to
establish and grow. Fallen log removal accelerated
deer entrance and changed understory composition
toward more browse-resistant and unpalatable species.
These results underscore the importance of considering
the dynamics (timing, severity, and extent) of fallen
woody debris influencing understory herbivory and
post-disturbance succession. In addition, experimental
results underpin the importance of maintaining snags
and large woody debris in disturbed landscapes where
salvage logging is a routine procedure.
C. L. Westerholm
Plant Ecology and Systematics, Faculty of Science, Lund
University, Ecology Building, 223 62 Lund, Sweden
Keywords Austrocedrus chilensis Browsing
Disturbance Exotic deer Forest decline
Abstract Natural disturbances such as fires, windstorms, floods, and herbivory often act on plant
communities, affecting their structure and the abundance and composition of their species. Most research
has focused on the effects of single disturbances on
plant communities whereas the synergistic effects of
several disturbances have received less attention. In
this study, we evaluated how timing and severity of tree
mortality modified plant use by introduced deer and
early post-mortality successional trajectories in northern Patagonian conifer forests. We sampled understory
composition and deer use in Austrocedrus chilensis
(ciprés de la cordillera) forest stands undergoing
varying timing and severity of forest mortality as
reconstructed using dendroecological techniques. In
addition, we evaluated the effect of fallen logs on plant
composition and deer use of plants by monitoring areas
of massive dieback where fallen logs had been
removed for fire hazard reduction, and nearby control
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_2
11
12
Introduction
Coarse-scale disturbances such as fires, snow avalanches, windstorms, droughts, and insect defoliation
strongly influence the rate and direction of plant
succession. These disturbances release limiting
resources, triggering vegetation changes that attract
herbivores searching the landscape for patches of
high-quality forage (Jefferies et al. 1994). On the
other hand, the heterogeneous matrix of dead woody
debris left after forest disturbances can strongly limit
and control herbivore movement (Thomas et al.
1979; Hanley et al. 1989; Nyberg 1990). Thus, plant
communities will likely reflect a complex synergism
of disturbance characteristics that affect plant performance directly by releasing limiting resources
(Pickett and White 1985) and indirectly by modifying
herbivore foraging patterns (Stuth 1991).
Although forests are highly dynamic systems
subjected to natural disturbances of different scales,
relatively few studies have addressed how large
herbivores, such as ungulates, differentially use and
impact vegetation of sites affected by forest disturbances of varying severity and timing (Wisdom et al.
2006). Ungulates generally exert minor influences on
the structure and function of mature forest stands
(Russell et al. 2001). However, their effect following
a disturbance can determine the trajectory of the
system among alternative states (Hobbs 1996; Russell
et al. 2001). We hypothesized that depending on the
severity and timing of the disturbances, physical and
biotic conditions at disturbed sites may alter deer
behaviour, thus changing their role in modifying
plant succession. We predict that recent sudden,
massive forest dieback events such as windstorms
may create a mosaic of highly inaccessible microsites
composed of a tight network of fallen logs and
branches and will be dominated by palatable plants.
Older, less severe or more chronic patterns of tree
mortality, by contrast, may allow more accessibility,
will show signs of higher deer use and will be
dominated by unpalatable plants.
Forests of northern Patagonia, particularly on Isla
Victoria, are ideal for evaluating forest mortality–
herbivory interactions. Here, extensive stands of
Austrocedrus chilensis (D. Don.) Pic. Serm. & Bizarri
(ciprés de la cordillera) are being affected by ‘‘mal
del ciprés’’, a syndrome caused by a poorly known
agent (Filip and Rosso 1999; La Manna and
A.G. Van der Valk (ed.)
Rajchenberg 2004; Greslebin and Hansen 2006) that
causes root death and standing mortality followed by
mass canopy collapse owing to root weakening and
increased susceptibility to windthrow. At the landscape scale, poor soil drainage controls the
occurrence of patches of standing dead trees of
diverse sizes plus logs and fallen branches on the
forest floor that appear interspersed in a matrix of
healthy forest (La Manna et al. 2008; Fig. 1a).
Interacting with the understory and tree saplings in
these forests, there are also abundant introduced
cervids, mostly red deer (Cervus elaphus) and fallow
deer (Dama dama) (Simberloff et al. 2003). Austrocedrus forests are heavily used by introduced deer
owing to high forage availability and provision of
winter cover (Relva and Caldiz 1998; Barrios Garcia
Moar 2005). In addition, extensive removal of
downed slash and fallen logs along roads for fire
hazard mitigation (Fig. 1b) offers a unique largescale manipulative experimental setting in which to
test possible mechanisms involved in this interaction
between mortality and herbivory.
Here, we present results that combine dendroecological techniques for determining timing and
severity of past mortality with standard vegetation
and herbivore use assessments that preliminarily
underscore the importance of stand decline history
on understory vegetation structure and composition.
In addition, we experimentally demonstrate the
impact of fallen obstacles on herbivory by deer as a
key mechanism in modifying the strength of herbivory effects on vegetation.
Methods
Study site
The study was conducted in a 2 9 4 km area of
evergreen conifer Austrocedrus forest on northern
Isla Victoria, Nahuel Huapi National Park, Argentina
(40570 S; 71330 W; Fig. 2). Within the study area, we
sampled for tree mortality reconstructions, deer use
and vegetation censuses in four areas of ca. 1 ha each
representing forests with contrasting history and
severity of stand mortality (Criollos, Larga, Redonda,
Pseudotsuga, Fig. 2, Table 1).
Isla Victoria is an island running NW to SE that
comprises 3,710 ha, with a varied topography that
Forest Ecology
13
Fig. 1 Photographs showing massive mortality of Austrocedrus chilensis forests with standing dead trees, logs and fallen branches
(a) and adjacent areas where logs and fallen branches were removed (b) on Isla Victoria, northern Patagonia
includes flat, shallow valleys, and elevations of up to
1,025 m. Mean annual rainfall is 1,700 mm (Barros
et al. 1988), mostly occurring during winter (June to
September). Soils are allophanic (derived from volcanic ashes), sandy, permeable, and rich in organic
matter and acid pH (Koutché 1942). Isla Victoria is
covered mainly by southern beech pure Nothofagus
dombeyi forests, pure Austrocedrus forests, and
mixed N. dombeyi-Austrocedrus forests. Lomatia
hirsuta, Maytenus boaria, Nothofagus antarctica,
Luma apiculata, Myrceugenia exsucca, and Dasyphyllum diacanthoides are subdominant tree species
in these forests. The understory includes palatable
shrubs such as Aristotelia chilensis, Maytenus chubutensis, Ribes magellanicum, Schinus patagonicus,
and Chusquea culeou as well as unpalatable shrubs
such as Berberis spp. and Gaultheria spp. The
herbaceous layer includes native species such as
Uncinia sp. and exotics such as Cynoglosum creticum
and Digitalis purpurea. Species nomenclature follows Ezcurra and Brion (2005).
Historical disturbances consist of extensive fires
that occurred during European settlement resulting in
80- to 120-year-old postfire-cohorts (Veblen and
Lorenz 1987). These forests have scarce regeneration
because the dominant tree species are not shade
tolerant, although sporadic regeneration can occur in
small tree-fall gaps (Veblen et al. 1989). Since 1948,
when the first observation was recorded on Isla
Victoria (Havrylenko et al. 1989), and extending over
the island and the region with geographically varying
intensities, the main present disturbance pattern is
mal del cipre´s Austrocedrus mortality.
Superimposed on the pattern of disturbance by
fires and dieback are the effects of introduced
herbivores. In 1916, red deer (Cervus elaphus), axis
deer (Axis axis), and fallow deer (Dama dama) were
successfully introduced to the island. At present, red
deer and fallow deer are extremely abundant, while
axis deer is apparently extinct on the island. By
1959, exotic deer densities on the island were
estimated to be 40 individuals/km2 (Anziano 1962),
and recent estimates indicate densities of 26 individuals/km2 (Relva unpubl.). Average red deer
density throughout the present distributional range
in Patagonia has been estimated at about 2 individuals/km2 (Flueck et al. 2003); however, these
authors also state that in favourable conditions
densities may reach 100 deer/km2 (ecotonal habitat)
and 40–50 deer/km2 (steppe habitat). Exotic deer
have significantly modified the forests on Isla
Victoria, reducing cover by palatable species, such
as Aristotelia chilensis (Veblen et al. 1989), and
delaying the growth of Austrocedrus and Nothofagus
dombeyi seedlings and saplings to adult size (Veblen
et al. 1989; Relva and Veblen 1998).
14
A.G. Van der Valk (ed.)
using a Henson computer-compatible radial increment-measuring device. Disturbance dates were
determined on living trees by detecting growth
release events. In this study, we define release events
as occurring when the tree-ring width of five
contiguous years increased more than 150% compared to the preceding 5 years growth (Kitzberger
et al. 2000a). The growth release frequencies were
quantified in 10-year periods by calculating the
number of individuals that underwent growth release
in a period relative to total individuals present in that
period. Dates of death of dead-standing and downed
trees were established using the standard visual
skeleton plots method (Stokes and Smiley 1968) in
combination with the COFECHA cross-dating program (Holmes 1983). This program statistically
analyses the correlation between pieces of undated
(floating) tree-ring series and master series dated
independently. For cross-dating, Cerro Los Leones
(International Tree Ring Data Bank, http://
www.ngdc.noaa.gov/paleo/treering.html) was used
as the master tree ring chronology.
Vegetation and deer use
Fig. 2 Location of Austrocedrus chilensis forest stands
studied on Isla Victoria, Parque Nacional Nahuel Huapi,
Argentina. Closed circles denote control areas and open
squares denote log removal areas. See Table 1 for stand
characteristics
Field sampling
Mortality assessments
In each area we used dendroecological techniques
(Stokes and Smiley 1968) to reconstruct the timing
and duration of tree mortality events. In each area, in
fifteen 314 m2 plots we cored the closest live tree to
the centre of the plot at ca. 50 cm with increment
borers to determine dates of growth release related to
mal del cipre´s mortality and/or associated windthrow
from neighbour trees. Dead standing, wind-snapped,
and uprooted trees were sampled by cutting partial
cross-sections at the base of each individual to date
the year of death. All samples were sanded with
successive grades of sandpaper to obtain an optimal
view of annual rings. Ring widths in tree cores and
cross-sections were measured to the nearest 0.01 mm
In each area we sampled forest structure, understory
abundance and composition, and deer use with 15
concentric plots of variable sizes placed systematically every 20 m along three parallel lines that were
located in relatively homogeneous areas, each
approximately 50 m apart from adjacent lines. Forest
structure was sampled in fifteen 314 m2 circular
plots, in which we measured diameters of adult trees
([4 cm at breast height) in four categories: living,
uprooted dead, standing dead, and snapped dead tree.
Understory abundance and composition were surveyed in fifteen 100 m2 circular plots in which we
visually estimated cover by individual species of tree
saplings (height[10 cm and dbh\4 cm), shrubs, and
herbs. In each 100 m2 circular plot, we also counted
and measured tree sapling height and assessed
seedling abundances (height \10 cm) by counting
within four 1 m2 plots randomly distributed throughout the 100 m2 understory plots. We measured the
tallest shrub of each species and used a scale
according to Allen and McLennan (1983) to assess
the degree of browsing on saplings and shrubs. This
scale distinguishes: 0, no evidence of browsing; (1)
slightly browsed (one or two branches browsed); (2)
Forest Ecology
15
Table 1 Forest characteristics of Austrocedrus chilensis study areas and effects of mortality on Isla Victoria, Parque Nacional
Nahuel Huapi, Argentina
Area
Latitude (S)
Longitude (W)
Pseudotsuga
Criollos
Larga
40540
40530
40530
40530
0
0
0
71330
7132
7132
Redonda
7133
Annual precipitation (mm)
1600–1800
1600–1800
1600–1800
1600–1800
Elevation (m asl)
800
825
800
850
Aspect
NE
NE
NE
NE
a
Basal area live (%)
46.4
21.1
26.8
25.7
Basal area dead standing (%)
24.2
21.6
18.6
20.6
Basal area uprooted (%)
35.0
58.1
54.0
50.3
Basal area snapped (%)
6.2
5.4
7.4
5.3
Total basal area (m2/ha)
62.6 (9.5)a
125 (11.1)
133.4 (11.2)
173 (14.9)
Age of live trees (years)
Dead tree age (year)
b
116 (5.3) n = 9
104 (6.1) n = 12
75 (9.9) n = 13
116 (11.5) n = 14
103 (9.5) n = 13
134 (5.6) n = 15
56 (4.6) n = 14
52 (4.2) n = 11
Mortality initiation
1980
1980–1990
1970
1970–1980
Year of first death recorded
1972
1965
1933
1969
a
Values are means with standard errors in parentheses
b
Number of sampled trees
moderately browsed (more than two branches
browsed), and (3) heavily browsed (most branches
browsed). Pellet groups were counted using a 10 m2
circular plot placed in each study station. Degree of
browsing and pellet group counts were used as an
index of animal use (Mayle et al. 1999). The degree
of site accessibility to deer was estimated by measuring the maximum height of logs and fallen
branches, and by estimating their cover as was done
in the understory plots.
Fallen tree-removal experiment
To evaluate the effects of fallen trees on deer–
vegetation interactions, we performed a blocked
sampling design at control areas (Criollos, Redonda,
and Larga) and three nearby (\200 m away) areas
from which all downed dead trees had been removed
in 1994, 1997, and 1998, respectively (hereafter,
removal treatment). There were two different control
(non-removal) areas: (1) areas with more than two
downed trees (hereafter, non-removal treatment), and
(2) naturally open areas between fallen trees (hereafter, non-removal open treatment). Each treatment
was sampled in stratified manner using fifteen 20 m2
circular plots. Variables describing forest structure,
understory abundance and composition, and animal
use were recorded in a similar fashion to those
described at the beginning of this section.
Data analyses
We investigated the interaction among forest mortality, deer use, and understory traits through multiple
stepwise regression. One set of regression analyses
was performed to determine the minimum set of
variables related to forest mortality and understory
traits that allow us to predict deer use (pellet group
counts and degree of browsing as dependent variables). A second regression analysis determined the
variables related to forest mortality and deer use that
can explain the abundance of palatable and unpalatable shrubs species in the understory. Independent
variables related to forest mortality were: (i) history:
according to dendroecological data forest stands that
were categorized as recent (1, death dates peaking in
the 1980s) and old (2, death dates peaking in the
1970s), and (ii) severity: expressed as basal area of
live, uprooted, standing dead and snapped trees, and
cover of fallen branches. Variables related to
16
understory traits were herb cover, tree sapling cover,
and cover of unpalatable and palatable shrubs.
Effects of fallen tree removal on plant community
and deer use were evaluated by ANOVA using areas
as experimental units and triplets of log removal/log
non-removal/non-removal open treatments as blocks.
Differences in means between treatments were based
on post-hoc tests. In all statistical analyses, counts
(numbers of pellet groups) and measures (heights)
were log-transformed, and proportions (understory
cover) were arcsine-transformed when needed to
achieve normality and homoscedasticity.
A.G. Van der Valk (ed.)
del cipre´s mortality and subsequent windthrow
(Table 1). Around 25% of the basal area consisted
of live trees, whereas 50–60% of the basal area
consisted of downed, uprooted trees. By contrast,
Pseudotsuga, which was the youngest stand
(Table 1), suffered lower overall levels of mortality
and subsequent tree fall with ca. 45% of tree basal
area alive and ca. 35% of the basal area on the
ground. Percentages of wind-snapped and standing
dead trees were relatively uniform among stands
(Table 1).
Predictors of deer use and shrub composition
Results
Timing and severity of tree mortality
Growth release patterns in surviving trees and
frequency patterns of death dates suggest differences
in timing and severity of mortality occurred within
the study area. Larga showed the longest and most
uniform history of mortality with death dates and
releases starting in the 1950s, peaking in the 1970s,
and extending into the 1980s (Fig. 3). Redonda
showed evidence of mortality starting mainly in the
1960s and peaking in the 1970s, while at Criollos,
mortality started in the 1970s and peaked in the 1980s
and 1990s. Similar to Criollos, but with less severity,
Pseudotsuga had mortality starting in the 1970s and
peaking in the 1980s (Fig. 3). At Criollos, 80% of the
dated uprooting occurred in a relatively distinct
period during the 1980s and 1990s. By contrast,
uprooting during that same period accounted for 40%
and 50% of downed trees at Redonda and Larga,
respectively, thus suggesting a more gradual process
of canopy collapse. During the 1990s, dead trees in
massive mortality stands (Criollos, Redonda, and
Larga) were mostly uprooted (Fig. 3). Ring width
patterns of these uprooted trees indicated the existence of growth release events in a large percentage
of trees (50, 75, and 100% at Larga, Criollos, and
Redonda, respectively). This fact suggested that
wind-induced uprooting occurred after canopy opening owing to mortality of dead standing trees and/or
uprooting of neighbouring trees.
Criollos, Redonda, and Larga were on average the
stands affected the longest and most severely by mal
The multiple regression analyses showed that deer use
was positively related to the history of stand mortality
(stands with older mortality are used more heavily)
and negatively related to branch cover. Thirty-five
percent of the variance in the number of deer pellets
was explained by the history of stand mortality
(?, P \ 0.01) and fallen branch cover (-, P \ 0.01)
(model: F = 6.44; df = 4,48; P = 0.0031). Similarly, 32% of the variance in the degree of browsing
on plants was explained by the history of stand
mortality (?, P \ 0.05) and fallen branch cover
(-, P \ 0.05) (model: F = 2.99; df = 7,45;
P = 0.011). By contrast, no single vegetation variable
significantly explained deer use.
Composition of understory vegetation was also
explained mostly by history and severity of stand
mortality. Fifty percent of the variance in cover of
unpalatable shrub species was positively related to
history of stand mortality (stands with older mortality
have higher cover of unpalatable shrubs, P \ 0.001),
while the degree of browsing was negatively related to
cover of unpalatable shrubs (P \ 0.05) (model:
F = 9.38; df = 5, 47; P = 0.001). Cover of palatable
species was related only to basal area of uprooted
trees, a measure of mortality severity (P \ 0.05)
(model: F = 4.46; df = 3, 49; P = 0.00756),
explaining 21% of the variance in palatable species
cover.
Effects of fallen trees on deer use and vegetation
As expected, uprooted basal area (F = 112.8, df = 2,
P \ 0.001, Table 2) and branch cover on the ground
(F = 37.16, df = 2, P = 0.001) in the fallen tree
removal treatment were lower than those found in the
Forest Ecology
17
Fig. 3 Frequency of tree
death dates by cause (wide
bar) and frequency of live
tree releases (narrow bar) at
the study sites
non-removal treatment. In the treatment in which
fallen trees had been removed and in the naturally
open treatment, deer pellet number (F = 75.1,
df = 2, P \ 0.001) and browsing (F = 23, df = 2,
P = 0.002, Table 2) were higher than in the adjacent
treatment in which fallen trees had not been removed.
Total shrub cover was similar among removal and nonremoval treatments (F = 3.99, df = 2, P = 0.079).
However, the proportion of palatable shrub species—
such as Aristotelia chilensis, Ribes magellanicum,
Table 2 Mean (and SE) of different variables measured in fallen tree removal and non-removal treatments
Uprooted
basal area
(m2/20 m2)
Branch cover
(%)
Number of
pellet group
Browsing
index
Palatable shrub Non-palatable
cover (%)
shrub cover (%)
6.00 (3.50) a
Fallen tree
removal
treatment
0.03 (0.02) a 26.28 (2.82) a 6.31 (1.29) a 1.43 (0.11) a
Non-removal
treatment
0.63 (0.05) b 78.61 (4.94) b 0.11 (0.04) b 0.56 (0.13) b 50.84 (2.78) b 18.24 (5.5) b
19.56 (10.16) a
Naturally open
treatment
0.15 (0.02) c 21.07 (3.82) a 5.67 (0.81) a 1.54 (0.06) a
62.52 (19.33) a
3.12 (1.84) a
34.01 (7.85) a
Herb cover
(%)
30.52 (7.74) a
40.8 (19.9) a
Different lowercase letters indicate significant differences among different treatments at P \ 0.05 (ANOVA and post-hoc Tukey
Tests). Statistical analyses were conducted on the transformed values of variables, but original values are shown in the table
18
Maytenus boaria—was significantly higher in the
non-removal treatment compared with the removal
treatment and the naturally open treatment (F =
41.53, df = 2, P = 0.001). Conversely, cover by
unpalatable shrubs—such as Berberis spp.—was
15.8% and 12.3% higher in the removal treatment
and naturally open treatment, respectively, than in
non-removal treatment (F = 38.75, df = 2, P =
0.001, see Appendix). No significant differences
were found in total herb cover among the three
treatments (F = 1.73, df = 2, P = 0.25, Table 2).
Discussion
Timing and severity of tree mortality
Austrocedrus areas with moderate mortality (ca. 65%
of basal area dead) are relatively open, young, and
accessible forest with most trees alive or standing but
dead. In contrast, where mortality exceeds 75% of
basal area, many trees lie on the ground forming an
inaccessible tangled mass of logs and branches
several meters high. Mortality levels in this study
are similar to those found by Loguercio and Rajchenberg (2004) but higher than those found by La
Manna et al. (2006) for forests with similar stand
structure in southwestern areas of Rı́o Negro and in
the nearby province of Chubut.
Two temporal factors are important in the interaction between mortality and herbivores that may
affect plant communities: (1) the timing of canopy
opening (i.e., increase in light levels to understory
plants) and (2) the timing of canopy collapse (i.e.,
decreasing accessibility to herbivores). These stages
do not necessarily coincide. Dendroecological techniques allowed us to differentiate both processes. In
our system, most trees were attacked by root fungi,
lost foliage, and remained standing until root rot
made the trunk unstable and the tree fell. This was
evidenced in ring growth patterns of downed trees by
a strong suppression before and at the time of death.
Additional unattacked trees fell because the lack of
surrounding canopy trees made them susceptible to
wind-throw. This was evidenced in downed trees by
strong radial growth release (suggesting that trees
were not infected) before sudden death by snapping
or uprooting. In both cases, canopy opening may not
A.G. Van der Valk (ed.)
result in understory blocking for several years or even
one or two decades. This time lag between canopy
opening and understory physical blocking may have
an impact on understory composition. During early
phases of the decline process, the understory receives
light but there is also substantial herbivore pressure.
Therefore shade-intolerant plants that are resistant to
herbivores or are dispersed by them may benefit. In
our system, such as species may be Uncinia sp.,
which dominated recently dead forest, is lightdemanding, and is dispersed in deer fur. The initial
density of the stand may have been important
determinants of how fast the canopy collapsed after
mortality began. In our study, in all dense areas
(Criollos, Larga, and Redonda) uprooting has been
the main cause of mortality process for the past three
decades. The death dates in our study are similar to
those registered by Cali (1996), who worked in two
mainland Austrocedrus stands close to our study sites.
Interactive effects of forest mortality and deer use
on plant communities
Our results indicate that fallen logs with a high
density of branches strongly limited deer accessibility
to certain microsites and created natural exclosures
and safe sites for palatable plant establishment and
growth. Pulido et al. (2000) found a similar relationship between presence of a native camelid, Lama
guanicoe (guanaco), and slash in a managed Nothofagus pumilio forest in Tierra del Fuego (southern
Argentina). Rebertus et al. (1997) found that browsing by guanaco was negatively correlated to the
blowdown area of N. pumilio forest in Tierra del
Fuego. In blowdown areas above 5 hectares, guanaco
browsing was restricted to the periphery. Similarly,
Cavieres and Fajardo (2005) found in old-growth
stands of N. pumilio that guanaco damage was higher
in small gaps than in the bigger ones. On the other
hand, postfire coarse woody debris has been found to
provide Populus tremuloides refugia from red deer
browsing in Yellowstone National Park (Ripple and
Larsen 2001). On the contrary, Bergquist and
Örlander (1998) found that Picea abies browsed by
moose did not vary in sites with different amounts of
slash on the forest floor. Similarly, Kupferschmid and
Bugmann (2005) found that fallen trees do not
constitute a barrier to chamois (Rupicapra rupicapra)
browsing Picea abies saplings. According to Thomas
Forest Ecology
et al. (1979), a depth of dead and fallen material
higher than 0.6 m substantially limits deer use of the
area, and when the depth is high enough to make deer
jump, the energetic cost of locomotion increases
dramatically (Hanley et al. 1989; Nyberg 1990).
Another complementary explanation for deer to avoid
areas with deep slash is that they would not be able to
escape easily if a predator does attack (White et al.
2003).
In our study, the negative relationship between the
amount of fallen logs and the deer use was clearly
manifested when slash was removed. The number of
deer pellet groups found where slash had been
removed was six times the number found in control
areas. As a result of this heavier use, after only
4 years of the treatment, understory composition
changed dramatically toward more unpalatable and
browse-resistant species in the slash-removal
treatments.
The positive relationship between deer use and
time since peak mortality suggests that with time,
fallen trees lose decomposing branches, and accessibility increases. In the early stages, shrubs would be
not abundant except for Aristotelia chilensis, a shadeintolerant, tall shrub (Muñoz and González 2006) that
is highly palatable and consumed by deer (Anziano
1962; Veblen et al. 1989; Relva and Veblen 1998;
Relva and Caldiz 1998). In areas with recent or
severe mortality, A. chilensis was observed growing
between logs and fallen branches. This spatially
aggregated distribution in herbivore-free refuges (i.e.
safe sites where individuals grow and reproduce
successfully, far from the browsing range of the
herbivores) located in grazing areas was also
observed by Vázquez (2002a), who also found that
this type of distribution influenced the mechanisms of
pollination of this species. Positive association
between certain species of plants with coarse debris
has been noted in other forest systems in which
windstorms were generally predominant and produced great amounts of dead material on the forest
floor (Allan et al. 1997; Peterson and Pickett 2000; de
Chantal and Ganström 2007). However, the strong
positive relationship between A. chilensis and fallen
branches could additionally be a response to
improved recruitment conditions, as shown in other
species (Schreiner et al. 1996). In areas with the
oldest mortality (Redonda and Larga) and in microsites from which logs had been removed, deer use
19
increased, and shrub composition changed toward
less palatable species or browse-resistant ones such as
Berberis spp. Both B. buxifolia and B. darwinii,
which are common in Austrocedrus forests, are
dominant in intensely grazed areas (Rebertus et al.
1997; Vázquez 2002b; Gallopin et al. 2005). Berberis
spp. and other spiny shrubs may act as nurse plants of
other species, by physically protecting more palatable
plants from herbivores (De Pietri 1992) and/or
improving abiotic conditions to facilitate establishment and growth of tree seedlings (Kitzberger et al.
2000b). In our study site, we have found no saplings
of Austrocedrus in recently and severely disturbed
forest. This could be because of the high cover of
light-demanding herbs, Uncinia sp. and Digitalis
purpurea, in early post-disturbance stages that could
be negatively affecting tree seedling recruitment or
due to low seed production by overmature trees. By
contrast, in areas with less severe mortality, Austrocedrus seedlings and saplings are a dominant
component of the understory (see Appendix).
Because Austrocedrus is a shade-intolerant species,
the canopy opening produced by less severe mortality
probably explains this abundant tree regeneration
despite heavy use of canopy gaps by deer (Veblen
et al. 1989; Relva and Veblen 1998).
The spatially and temporally heterogeneous nature
of forest mortality interacting with large herbivores
may shape complex mosaics of vegetation. Prediction
of plant community composition and structure should
move forward from approaches that emphasize
disturbances modifying abiotic resources for plant
regeneration or plant–animal interactions toward
spatially explicit approaches that integrate plant
performance and animal behaviour within the context
of a dynamic forest landscape.
This study underpins the importance of maintaining snags and large woody debris for the role in
providing safe sites for tree and understory regeneration, a management policy that should also extend to
disturbed landscapes where salvage logging is a
routine procedure.
Acknowledgments We wish to thank Diego Vazquez for
valuable comments on the manuscript, park rangers of Isla
Victoria (Damián Mujica, Lidia Serantes, Domingo Nuñez, and
Carina Pedrozo) for helping us in many ways. Delegación
Técnica Regional and Intendencia del Parque Nacional Nahuel
Huapi assisted us with working permits, and Cau Cau and
Mares Sur with transportation. We are especially grateful to
20
A.G. Van der Valk (ed.)
Juan Gowda for helping on cross-section tree extractions, and
Eduardo Zattara for his field assistance. Daniel Simberloff
revised several versions of this manuscript improving the
language and clarity. This research was supported by a
postdoctoral fellowship to M.A.R from Consejo Nacional de
Ciencia y Técnica of Argentina CONICET and by funds from
Universidad Nacional del Comahue. Foundation LinnaeusPalme funded C.L.W scholarship.
Appendix
Mean cover (%) and standard error of vascular species recorded in fifteen 100 m2 plots in the study areas
Area
Criollos
Pseudotsuga
Larga
Redonda
Tree species
Austrocedrus chilensis
2.83 (1.31)
Lomatia hirsuta
0.68 (0.29)
Luma apiculata
1.5 (1)
1.69 (0.99)
1.69 (1.06)
1.87 (0.99)
16.79 (5.66)
Maytenus boaria
0.01 (0.01)
0.01 (0.01)
0.01 (0.01)
Nothofagus dombeyi
0.17 (0.17)
1.69 (0.99)
1.08 (1.07)
Pseudotsuga menziesiia
1.34 (0.33)
Shrub species
Aristotelia chilensis
6.34 (2.68)
0.35 (0.23)
3.21 (2.47)
0.01 (0.01)
Azara lanceolata
4.14 (1.60)
0.01 (0.01)
Berberis darwinii
2.51 (0.94)
3.2 (1.27)
Budleja globosa
Colletia hystrix
0.18 (0.17)
0.01 (0.01)
0.18 (0.17)
0.01 (0.01)
0.02 (0.01)
2.67 (2.49)
0.54 (0.28)
Gaultheria spp.
0.03 (0.01)
Maytenus chubutensis
Ribes magellanicum
39.5 (7.47)
0.36 (0.22)
0.21 (0.18)
0.01 (0.01)
4.36 (1.45)
0.19 (0.18)
0.17 (0.17)
Rosa rubiginosaa
Schinus patagonicus
20.36 (5.73)
0.57 (0.28)
0.01 (0.01)
5.56 (1.68)
Herb species
0.01 (0.01)
Acaena ovalifolia
Alstroemeria aurea
Blechnum spp.
0.01 (0.01)
0.01 (0.01)
0.01 (0.01)
0.01 (0.01)
0.17 (0.17)
Carex spp.
Cynanchum diemii
0.53 (0.26)
3.67 (1.22)
Cynoglossum creticuma
2.51 (1.33)
1.38 (1.00)
Digitalis purpureaa
Galium aparinea
30.17 (6.76)
Rumex acetosellaa
0.54 (0.28)
15.51 (4.32)
1.27 (1.07)
13.01 (3.67)
7.16 (2.78)
0.21 (0.18)
0.01 (0.01)
0.01 (0.01)
0.34 (0.23)
0.02 (0.01)
Vicia nigricans
Mutisia spp.
1 (1)
0.045 (0.01)
0.02 (0.01)
2.87 (2.67)
0.01 (0.01)
Uncinia spp.
a
Denotes exotic species
0.2 (0.18)
0.01 (0.01)
Rumohra adiantiformis
Grasses
0.01 (0.01)
0.01 (0.01)
Adiantum chilense
67.83 (6.8)
0.01 (0.01)
1.53 (1.00)
3.02 (1.29)
33.93 (6.74)
Forest Ecology
References
Allan C, Stewart G, Allen R (1997) Long term influences of
deer browsing on forest health and conservation values of
the Kaweka Range. Final Report Lincoln University and
WWF, Canterbury
Allen RB, McLennan MJ (1983) Indigenous forest survey
manual: two inventory methods. New Zealand For Serv
FRI Bulletin no. 48
Anziano AF (1962) Acción de los animales sobre la flora.
Anales de Parques Nacionales, Administración de Parques
Nacionales. Tomo IX:107–112
Barros VR, Cordon VH, Moyano C, Mendez R, Forquera JC,
Pizzio O (1988) Carta de precipitación de la zona oeste de
las provincias de Rı́o Negro y Neuquén. Primera contribución (1972–1981). Facultad de Ciencias Agrarias.
Universidad Nacional del Comahue, Cinco Saltos
Barrios Garcia Moar MN (2005) Patrones de daño, uso actual y
tendencia histórica de la abundancia de ciervos introducidos en bosques nativos de Isla Victoria, Patagonia,
Argentina. Licenciate Dissertation. Universidad Nacional
del Comahue, Bariloche, Argentina
Bergquist J, Örlander G (1998) Browsing damage by roe deer on
Norway spruce seedlings planted on clearcuts of different
ages. 1. Effect of slash removal, vegetation development,
and roe deer density. For Ecol Manage 105:283–293
Cali SG (1996) Austrocedrus: estudio de los anillos de crecimiento y su relación con la dinámica del ‘‘mal del ciprés’’
en el Parque Nacional Nahuel Huapi, Argentina. Licenciate Dissertation. Universidad Nacional del Comahue,
Bariloche, Argentina
Cavieres LA, Fajardo A (2005) Browsing by guanaco (Lama
guanicoe) on Nothofagus pumilio forest gaps in Tierra del
Fuego, Chile. For Ecol Manage 204:237–248
de Chantal M, Ganström A (2007) Aggregations of dead wood
after wildfire act as browsing refugia for seedlings of Populus tremula and Salix caprea. For Ecol Manage 250(1):3–8
De Pietri DE (1992) Alien shrub in a National Park: can they
help in the recovery of natural degraded forest? Biol
Conserv 62:27–130. doi:10.1016/0006-3207(92)90933-E
Ezcurra C, Brion C (2005) Plantas del Nahuel Huapi. Catálogo
de la Flora Vascular del Parque Nacional Nahuel Huapi,
Argentina
Filip GM, Rosso PH (1999) Cypress mortality (mal del ciprés)
in the Patagonia Andes: comparison with similar forest
disease and declines in North America. Eur J Forest
Pathol 29:89–96. doi:10.1046/j.1439-0329.1999.00133.x
Flueck WT, Smith-Flueck JA, Naumann CM (2003) The current
distribution of red deer (Cervus elaphus) in southern Latin
America. Z Jagdwiss 49:112–119. doi:10.1007/BF02190451
Gallopin GC, Martı́n CE, Mermoz MA (2005) Impacto de la
ganaderı́a en la cuenca del rı́o Manso Superior. Parte I:
Bosque de Ñire con Laura. Anales de Parques Nacionales,
Administración de Parques Nacionales. Tomo XVII:9–36
Greslebin A, Hansen E (2006) Novedades sobre el mal del
ciprés. Patagonia For 12(1):11–14
Hanley TAR, Robbins CT, Spalinger DE (1989) Forest habitats
and the nutritional ecology of Sitka black-tailed deer: a
research synthesis with implications for forest management. USDA For Serv Res Pap PNW-GTR-230
21
Havrylenko M, Rosso PH, Fontenla S (1989) Austrocedrus
chilensis: contribución al estudio de su mortalidad en
Argentina. Bosque 10(1):29–36
Hobbs NT (1996) Modification of ecosystems by ungulates.
J Wildl Manage 60(4):695–713. doi:10.2307/3802368
Holmes R (1983) Computer-assisted quality control in tree-ring
dating and measurement. Tree-ring Bull 43:69–75
Jefferies RL, Klein D, Shaver GR (1994) Vertebrate herbivores
and northern plant communities: reciprocal influences and
responses. Oikos 71:193–206. doi:10.2307/3546267
Kitzberger T, Veblen T, Villalba R (2000a) Métodos dendroecológicos y sus aplicaciones en estudios de dinámica
de bosques templados de Sudamérica. In: Roig FA (ed)
Dendrocronologı́a en América Latina. EDIUNC, Universidad de Cuyo, Mendoza, Argentina, pp 17–78
Kitzberger T, Steinaker D, Veblen T (2000b) Effects of climatic variability on facilitation of tree establishment in
northern Patagonia. Ecology 81(7):1914–1924
Koutché V (1942) Estación Forestal de Puerto Anchorena, Isla
Victoria; su organización y trabajos. Ministerio de Agricultura,
Dirección de Parques Nacionales, Buenos Aires, Argentina
Kupferschmid A, Bugmann H (2005) Effect of microsites, logs
and ungulate browsing on Picea abies regeneration in a
mountain forest. For Ecol Manage 205:251–265
La Manna L, Rajchenberg M (2004) The decline of Austrocedrus chilensis in Patagonia, Argentina: soil features as
predisposing factors. For Ecol Manage 190:345–357
La Manna L, Bava J, Collantes M, Rajchenberg M (2006)
Caracterı́sticas estructurales de los bosques de Austrocedrus chilensis afectados por ‘‘mal del ciprés’’ en
Patagonia, Argentina. Bosque 27(2):135–145
La Manna L, Matteucci SD, Kitzberger T (2008) Abiotic factors related to the incidence of the Austrocedrus chilensis
disease syndrome at a landscape scale. For Ecol Manage
256:1087–1095
Loguercio G, Rajchenberg M (2004) El ‘‘mal del ciprés’’ y la
silvicultura del ciprés de la cordillera. In: Arturi MF,
Frangi JL, Goya JF (eds) Ecologı́a y Manejo de los
Bosques de Argentina. EDULP, Universidad Nacional de
La Plata, La Plata, Argentina
Mayle BA, Peace AJ, Gill RMA (1999) How many deer? A
field guide to estimating deer population size. Field Book
18. Forestry Commission, Edinburgh
Muñoz A, González M (2006) Aristotelia chilensis (Mol)
Stuntz. In: Donoso C (ed) Las especies arbóreas de los
bosques templados de Chile y Argentina. Autoecologı́a,
Cuneo Ediciones, pp 166–172
Nyberg JB (1990) Interaction of timber management with deer
and elk. In: Nyberg JB, Janz DW (eds) Deer and elk
habitat as in coastal forests of Southern British Columbia.
BC Ministry of Forests and Ministry of Environment,
Special Report 5, pp 99–131
Peterson CJ, Pickett STA (2000) Patch type influences on
regeneration in a western Pennsylvania, USA, catastrophic windthrow. Oikos 90:489–500. doi:10.1034/
j.1600-0706.2000.900307.x
Pickett STA, White PS (1985) The ecology of natural disturbance and patch dynamics. Academic Press, Orlando
Pulido FJ, Dı́az B, Martı́nez Pastur G (2000) Incidencia del
ramoneo del guanaco (Lama guanicoe Müller) sobre la
22
regeneración temprana en bosques de lenga (Nothofagus
pumilio (Poepp et Endl) Krasser) de Tierra del Fuego,
Argentina. Investigación Agraria. Sistemas Recursos
Forestales 9(2):381–394
Rebertus AJ, Kitzberger T, Veblen T, Roovers LM (1997)
Blowdown history and landscape patterns in the Andes of
Tierra del Fuego, Argentina. Ecology 78:678–692
Relva MA, Caldiz M (1998) Composición estacional de la
dieta de ciervos exóticos en Isla Victoria, P.N. Nahuel
Huapi, Argentina. Gayana (Zool) 62(2):101–108
Relva MA, Veblen T (1998) Impacts of introduced large herbivores on Austrocedrus chilensis forests in northern
Patagonia, Argentina. For Ecol Manage 108:27–40
Ripple WJ, Larsen EJ (2001) The role of postfire coarse woody
debris in Aspen regeneration. West J Appl For 16(2):61–
64
Russell FL, Zippin DB, Fowler NL (2001) Effects of Whitetailed deer (Odocoileus virginianus) on plants, plant populations and communities: a review. Am Midl Nat 146:
1–26. doi:10.1674/0003-0031(2001)146[0001:EOWTDO]
2.0.CO;2
Schreiner EG, Krueger KA, Happe PJ, Houston DB (1996)
Understory patch dynamics and ungulate herbivory in oldgrowth forest of Olympic National Park, Washington. Can
J Res 26:255–265. doi:10.1139/x26-029
Simberloff D, Relva MA, Nuñez M (2003) Introduced species
and management of a Nothofagus/Austrocedrus forest.
Environ Manage 31:263–275. doi:10.1007/s00267-0022794-4
Stokes MA, Smiley TL (1968) An introduction to tree-ring
dating. The University of Chicago Press, Chicago
A.G. Van der Valk (ed.)
Stuth JW (1991) Foraging behaviour. In: Heitschmidt RK,
Stuth JW (eds) Grazing management. An ecological perspective. Timber Press, Oregon, pp 65–83
Thomas JW, Black H, Scherzinger RJ, Pedersen RJ (1979)
Deer and Elk. In: Thomas JW (ed) Wildlife habitats in
managed forest. The Blue Mountains of Oregon and
Washington. USDA For Serv Agriculture Handbook 553,
pp 104–147
Vázquez D (2002a) Interaction among introduced ungulate,
plants and pollinators: a field study in the temperate forest
of Southern Andes. PhD Dissertation, University of Tennessee, Knoxville, Usa
Vázquez D (2002b) Multiple effects of introduced mammalian
herbivores in a temperate forest. Biol Invasions 4:175–
191. doi:10.1023/A:1020522923905
Veblen TT, Lorenz DC (1987) Post-fire stand development of
Austrocedrus -Nothofagus forest in northern Patagonia.
Vegetatio 71:113–126
Veblen T, Mermoz M, Martı́n C, Ramilo E (1989) Effects of
exotic deer on forest regeneration and composition in
northern Patagonia. J Appl Ecol 26:711–724. doi:10.2307/
2404095
White CA, Feller MC, Bayley S (2003) Predation risk and the
functional response of elk-aspen herbivory. For Ecol
Manage 181:77–97
Wisdom MJ, Vavra M, Boyd JM, Hemstrom MA, Ager AA,
Johnson BK (2006) Understanding ungulate herbivoryepisodic disturbances effects on vegetation dynamics:
knowledge gaps and management needs. Wildl Soc
Bull 34(2):283–292. doi:10.2193/0091-7648(2006)34
[283:UUHDEO]2.0.CO;2
Tree species richness and composition 15 years after strip
clear-cutting in the Peruvian Amazon
Xanic J. Rondon Æ David L. Gorchov Æ Fernando Cornejo
Originally published in the journal Plant Ecology, Volume 201, No. 1, 23–37.
DOI: 10.1007/s11258-008-9479-x Springer Science+Business Media B.V. 2008
Abstract Although strip clear-cutting has a long
history of use in the temperate zone, it was only
recently introduced for timber extraction in tropical
rain forests, where it is known as the Palcazú Forest
Management System. In this system heterogeneous
tropical forests are managed for native gap-dependent
timber species by simulating gap dynamics through
clear-cutting long, narrow strips every 40 years. As
part of an assessment of the sustainability of this
system, we evaluated the recovery of tree basal area,
species richness, and composition after 15 years of
regeneration on two strips (30 9 150 m) clear-cut in
1989 in Jenaro Herrera, Peru. Timber stocking and the
effects of silvicultural thinning were assessed in both
strips. The strips recovered 58–73% of their original
basal area and 45–68% of their original tree species
richness. Although both strips recovered more than
50% of their original composition, commercial species
had lower basal areas and lower densities than in the
forest before the clearing. Pioneer species with high
basal areas remained dominant 15 years after the
cutting. Silvicultural thinning in 1996 reduced the
X. J. Rondon (&) D. L. Gorchov
Department of Botany, Miami University, Pearson Hall,
Oxford, OH 45056, USA
e-mail: rondonxj@gmail.com
D. L. Gorchov
e-mail: gorchodl@muohio.edu
F. Cornejo
Proyecto Castañales, Puerto Maldonado, Peru
abundance of pioneer species in both strips, and
increased the abundance of commercial species in
one of the strips. Half of one strip was harvested by
deferment-cut (only commercial trees [30 cm dbh
and ‘‘other’’ species [5 cm dbh were cut); regeneration here had greater abundance of commercial species
and lower abundance of pioneer species. The low
stocking of commercial trees challenges the sustainability claims for this forest management system.
Keywords Natural forest management
Palcazú forest management model Rarefaction
Sustainable management Tropical rain forest
Introduction
Strip-clear cutting has extensively been used in the
temperate zone for forest management (Thornton
1957; Smith 1986; Heitzman et al. 1999; Allison
et al. 2003); Tosi (1982) and Hartshorn (1989a, 1995)
introduced this system to manage tropical rainforests
for timber extraction. The first implementation was in
the Palcazú Valley in Peru, as part of a joint United
States Agency for International Development (AID)
and Peru Instituto Nacional de Desarrollo (INADE)
development project (Tosi 1982; Hartshorn 1989a).
As a result, Tosi’s (1982) and Hartshorn’s (1989a,
1995) strip clear-cutting system is also known as the
Palcazú Forest Management System. In the Palcazú
Forest Management System heterogeneous tropical
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_3
23
24
forests are managed for native gap-dependent timber
species by simulating gap dynamics through clearcutting long, narrow strips (Hartshorn 1989a, 1995).
In this system, upland forest is clear-cut into 30–40 m
wide strips with a rotation of 30 to 40 years. The
length of the strip varies and depends upon topography (Hartshorn 1989a).
In the Palcazú system, timber, regardless of species,
is harvested and used locally (sawnwood, preserved
roundwood, and charcoal) or sold to attain maximum
value from the strips (Hartshorn 1989a; Gorchov et al.
1993). Animal traction is used to reduce soil compaction (Hartshorn 1989a; Gorchov et al. 1993). Natural
regeneration of seeds and stump sprouts is permitted
(Gorchov et al. 1993). Silvicultural treatments may
also be applied in the regenerating strips to promote
growth of desired species (Dolanc et al. 2003).
Initially, the Palcazú system was thought to be a
sustainable alternative for timber extraction compared to uncontrolled logging or selective logging.
Tosi (1982) and Hartshorn (1989a) predicted that
non-commercial pioneer species would not regenerate well in this system because the strips were too
narrow to allow sufficient sunlight, and commercial
species would be well represented in the regeneration. Many tropical timber species are gap-dependent
(Swaine and Whitmore 1988), and such gap-dependent species have rapid height and diameter growth
(Lieberman et al. 1985).
Several studies, however, have questioned the
sustainability of the Palcazú system (Simeone 1990;
Cornejo and Gorchov 1993; Gram 1997; Southgate
1998). Rapid early regeneration with high tree species
richness suggested that this system is ecologically
sustainable (Hartshorn 1989a), but Gorchov et al.
(1993) found that after one year of regeneration the
composition of strips was mainly dominated by pioneer
species of low commercial value. Thinning enhanced
the growth rates of commercial stems 11 years after the
cutting, but they still averaged \0.3 cm/year in diameter growth (Dolanc et al. 2003). Clearly, data are still
needed for later stages of regeneration.
We studied tree regeneration after 15 years on two
strips clear-cut in 1989 in the Peruvian Amazon in
order to generate the first assessment of the ecological
sustainability of the strip clear-cutting system. To
assess the ecological sustainability of a forest management system one ought to assess the structural
characteristics of a developing forest (basal area and
A.G. Van der Valk (ed.)
biomass), community characteristics (species richness
and composition), and functional characteristics
(nutrient cycling and primary productivity). In this
study, we focused on the recovery of tree basal area,
species richness, and species composition 15 years
after the cutting with values prior to the cutting. The
criterion used to assess the ecological sustainability of
this system was to evaluate whether these community
descriptors had recovered to approximate pre-clearing
levels. This criterion is based on the assumptions of
sustainability for natural forest management; i.e.,
sustained timber yields can be produced while maintaining a high diversity (Bawa and Seidler 1998). A
second objective was to determine stocking of commercial species in the strips 15 years after the cutting
to assess timber regeneration in this system. A third
objective was to determine if silvicultural thinning and
harvesting by deferment-cut improved the recovery of
structural and community descriptors in the strips.
Clear-cutting is the least severe anthropogenic disturbance when compared to cutting and burning for
pasture or plantation establishment, and bulldozing for
road building or development (Uhl et al. 1982). Thus,
clear-cut stands tend to have a rapid increase in species
richness a few years after logging (Hartshorn 1989a;
Faber-Landgendoen 1992) and a faster richness recovery than stands cut and burned for pasture or bulldozed
(Uhl et al. 1982). However, composition usually takes
longer to recover (Finegan 1996; Guariguata and
Ostetarg 2001). Thus, we expected greater recovery of
basal area and species richness than of species composition. We also expected silvicultural thinning and
deferment-cutting in the strips to improve the recovery
of all of these structural and community descriptors.
Methods
Study site
This study took place at the Centro de Investigaciones
Jenaro Herrera (CIJH S 453.950 W 7339.040 ), 200 km
south of Iquitos, Loreto, Peru. Mean annual temperature is 26.5C and mean annual precipitation is
2521 mm (Spichiger et al. 1989). A relatively dry
period occurs from June to August, but rainfall highly
varies each month of the year (Ascorra et al. 1993;
Rondon 2008). Soils are sandy-loam and the vegetation
is considered lowland tropical rainforest on high terrace
Forest Ecology
25
(Spichiger et al. 1989). The families with highest
densities on high terrace at CIJH are Sapotaceae,
Leguminosae, Lecythidaceae, Chrysobalanaceae,
Lauraceae, and Myristicaceae (Spichiger et al. 1996).
History of clear-cut strips in CIJH
Two 30 9 150 m strips (Fig. 1), 150 m apart, were
clear-cut in 1989 in primary high terrace tropical rain
forest at CIJH. The area had been selectively logged
19
20
17*
18*
15
16
13*
14*
11
12
9
10
150 m
N
7*
8*
5
6
3*
4*
15–20 years prior, but the forest maintained an intact
canopy. The long axis of each strip was oriented
north–south. Strip 1 was cleared in April–May, 1989
and strip 2 in October–November, 1989. Lianas and
shrubs were cut before tree felling. Most trees [5 cm
in diameter at breast height (dbh) were felled in each
strip using directional felling to ensure that the trees
cut landed in the strips (Gorchov et al. 1993). A few
large trees ([28 cm dbh, N = 5 in strip 1 and N = 13
in strip 2) leaning out of the strips were not cut to
avoid damage to the surrounding forest (Cornejo and
Gorchov 1993). An experimental deferment-cut
treatment cut was implemented in the south half of
strip 2 (plots 1–10). In the deferment-cut treatment, only
commercial trees C30 cm dbh and ‘‘other’’ species
[5 cm dbh were harvested in 1989; the smaller trees
of commercial species were left uncut (n = 56, 5–
28 cm dbh) to grow for the next harvest (Cornejo and
Gorchov 1993). All timber harvested was locally
used or carried off site. A complete survey of the
trees (C5 cm dbh) was made during the 1989 felling
for both strips (Cornejo and Gorchov 1993).
Each strip was divided into 20 15 9 15 m plots
(Fig. 1), in which all stump sprouts and survivors
(saplings not cut \5 cm dbh in 1989) were identified
and tagged. Recruits (trees [2-m tall) were identified
and censused on 8 out of the 20 plots in each strip.
Censuses took place once a year during 1990–1994,
1996, and 2000. In addition, an experimental silvicultural thinning treatment took place in March 1996;
pioneer trees (all Cecropia and trees\10-m tall of the
genus Alchornea and the family Melastomataceae)
were girdled by machete in portions of each strip
(Fig. 1). Censuses carried out May–June, 2004 in
strip 1 and June–July, 2005 in strip 2 provide the
‘post-clearing’ data analyzed here.
Tree identification
1
2
30 m
Fig. 1 Schematic of each of the two strips (30 9 150 m) at
Centro de Investigaciones Jenaro Herrera, Peru. Twenty plots
were marked in each strip (15 9 15 m). Plots thinned in 1996
are shaded. Plots with asterisk (*) were censused regularly for
all saplings C2 m. Advanced regeneration and stump sprouts
were censused throughout the strip. In strip 2, in the south half
(plots 1 to 10), 56 commercial tree species (5–28 cm dbh) were
left uncut as part of a deferment-cut treatment. Figure modified
from Dolanc et al. (2003)
Tree identification was done in the field using Gentry
(1993) and Spichiger et al. (1989, 1990). Voucher
specimens were deposited at the CIJH herbarium,
AMAZ, and MU. Voucher specimens of difficult taxa
were brought for comparison to Missouri Botanical
Garden (MOBOT). Several taxa were not identified to
the species level in the pre-clearing (1989) period;
identification for these taxa was only done to genus
or family level. For analysis purposes, trees identified to the same genus or family, without species
26
determination, were considered as one morphospecies.
Some Cecropia species were difficult to identify to the
species level, and they were grouped as one morphospecies for all richness comparisons.
Data analysis
Comparisons of tree basal area (BA), species richness,
composition, and timber stocking were evaluated in
strip 1 in 1989 (prior to cutting) vs. 2004, 15 years after
cutting, and in strip 2, in 1989 (prior to the cutting) vs.
2005, 15 years after cutting. In strip 2, all comparisons of community descriptors between the pre- and
post-clearing period were carried out separately for the
clear-cut and deferment-cut portions. We are aware that
forests are not stable and community descriptors vary
over time. In this study, we used the pre-clearing level
(1989) as a reference of mature growth. All tree species
richness and composition comparisons were done for
trees [7.5 cm dbh since both strips had complete
datasets per plot for these trees. Additional comparisons of richness and composition of trees C5 cm dbh
between the post- and pre-clearing censuses were
carried out for strip 2 (Rondon 2008), but these did not
differ qualitatively from trees[7.5 cm dbh.
The effect of thinning and deferment-cut on
structural and community descriptors
Before comparing structural and community descriptors in the pre- versus the post-clearing period, we
tested the effect of silvicultural thinning in the postclearing period in order to determine whether it was
appropriate to pool thinned and unthinned plots. In
strip 1, we used SAS proc GLM with thinning as a fixed
factor and plots as replicates. For strip 2, we used a twoway ANOVA with two fixed factors, thinning and
felling treatment (clear-cut versus deferment-cut), and
their interaction. All analysis were done using SAS
version 9.1, with a = 0.05; ANOVA tables are
reported in Rondon (2008). Statistical findings should
be interpreted with caution since the 15 9 15 m plots
within each strip were not independent.
Structural and community descriptors
Basal area (BA, m2/ha) was calculated for trees
[10 cm dbh for each strip at pre-clearing, one year
after the clearing (1990), and 15 years post-clearing.
A.G. Van der Valk (ed.)
The effect of thinning and deferment-cut was tested
on per plot BA (m2/plot). Calculations of BA are in
Rondon (2008).
To compare tree species richness between the preand post-clearing censuses at equal sample sizes,
sample-based rarefaction curves were obtained from
EstimateS 7.5 (Colwell 2005). The 15 9 15 m plots
were used as subsamples in each strip. Separate
rarefaction curves were constructed for the clear-cut
and deferment-cut portions in strip 2. Before constructing the rarefactions for the two different
censuses, the effect of thinning and deferment-cut
on tree species density (no. of species/plot) was tested
using the post-clearing censuses of the strips.
Tree composition comparisons were done at the
genus level because species identification may not
have been consistent between censuses. Since the
classic Sorensen index is sensitive to sample size and
assemblages with numerous rare species (Chao et al.
2005), the abundance-based Sorensen index (L) was
used to assess compositional similarity between
censuses in the strips. Using EstimateS 7.5 (Colwell
2005), we calculated L, L = 2UV/(U?V), where U
and V are the total relative abundances of the shared
species in samples 1 and 2 (Chao et al. 2005).
After determining if thinning and deferment-cut
had an effect on L calculated between pre- and postclearing censuses for each 15 9 15 m plots in the
strips, we pooled the data for each strip (keeping
clear-cut and deferment-cut halves of strip 2 separate)
to assess the compositional change of the strips
between censuses. In strip 1, L was recalculated for
the entire strip between pre- and post- censuses
(N = 1). In strip 2, L was recalculated separately for
the deferment-cut (N = 1) and clear-cut (N = 1)
portions of the strip. These values were compared
with L between two mature forest stands: strip 1 and
strip 2, both before the clearing (1989).
To calculate the relative abundances and basal area
of commercial and pioneer species, trees[7.5 cm dbh
in the strips were classified as commercial, pioneer,
and ‘‘other’’ species (Table 1). Commercial species
were those in genera valued for sawnwood at international and local markets based on data from the
International Tropical Timber Organization (ITTO)
from 1997 to 2005 (ITTO 1997–2005) and studies in
the Peruvian Amazon (Peters et al. 1989; PinedoVasquez et al. 1990). The list did not include species
valued for roundwood or non-timber forest products.
Forest Ecology
27
Table 1 Commercial and pioneer taxa occurring in censused
plots at CIJH with sources for commercial taxa
Commercial
Source
Annonaceae
Commercial
Meliaceae
Duguetia
2
Guarea
Guatteria
2
Trichilia
Xylopia
2
Apocynaceae
Aspidosperma
Macoubea
2
Bignoniaceae
Tabebuia
Calophyllum
Combretaceae
Terminalia
Olacaceae
2
Sapotaceae
Heisteria
Chrysophyllum
2
1, 2
Fabaceae
Dialium
Osteophloeum
1
Clusiaceae
Manilkara
Pouteria
Diplotropis
1
Hymenaea
1, 2
2
1, 2, 3
2
2
1, 2
2
1, 3
1
Erisma
1
Pioneer
Parkia
2
Cecropiaceae
Swartzia
2
Cecropia
Euphorbiaceae
Aniba
2, 3
Alchornea
Endlicheria
2, 3
Melastomataceae
Licaria
3
Mezilaurus
2
Persea
1, 2, 3
Vochysia
1
Nectandra
Ocotea
3
Vochysiaceae
Ormosia
Lauraceae
1, 2
Simaroubaceae
Simarouba
2
2
Myristicaceae
Virola
Caryocaraceae
Caryocar
Clarisia
Iryanthera
1
2, 3
Moraceae
Brosimum
1, 2
Boraginaceae
Cordia
Source
All genera
1, 3
1, 2, 3
3
Lecythidaceae
Cariniana
Eschweilera
1, 2
(Dolanc et al. 2003). ‘‘Other’’ species were taxa that
were not classified into one of the other two groups and
taxa that were only identified to the family level
(N = 3 morphospecies in strip 1 and N = 8 morphospecies in strip 2, both in 1989). ‘‘Other’’ species were a
combination of fast growing species (e.g., Inga),
successional species (e.g., Protium), and old growth
species (e.g., Mabea). Since ‘‘other’’ species, grouped
taxa of several life histories, this group was not
statistically analyzed.
The relative abundance of commercial and pioneer
species was calculated for each 15 9 15 m plot in
both strips in the pre- and post-clearing censuses. The
effect of thinning and deferment-cut was tested on the
relative abundance of commercial species and pioneer species in the strips. Due to unequal variance of
samples in testing the effect of thinning on commercial species in strip 1, additional analysis was done
using Kruskal–Wallis test, a non-parametric test. This
test did not differ qualitatively from the parametric
analysis; thus, only the latter was reported here. For
each strip, we used paired t-tests to determine
whether the relative abundance of commercial and
pioneer species for 15 9 15 m plots differed between
censuses. We also calculated basal area of commercial, pioneer, and ‘‘other’’ species in both strips in the
pre- and post-clearing censuses of each strip.
Stocking (no. of trees/ha) of commercial species
was calculated for (1) small trees between 5 to
10 cm dbh, and (2) large trees [10 cm dbh, in the
post- and pre-clearing censuses of each strip. We
tested the effect of thinning and deferment-cut on the
number of commercial stems per plot for each size
class in the strips. To make timber stocking comparisons between censuses, for each size class the total
number of stems/ha in the post-clearing period was
calculated and compared to the pre-clearing period of
each strip.
2
Source: (1) ITTO 1997–2005; (2) Peters et al. 1989; (3)
Pinedo-Vasquez et al. 1990
Results
Taxa not appearing in commercial or pioneers were considered
‘‘others’’. This table was modified from Dolanc et al. (2003)
After 15 years of regeneration, the advance regeneration (trees that survived the clearing in 1989)
comprised 16 and 18% of the total tree regeneration
(trees [5 cm dbh) of strips 1 and 2, respectively;
stump sprouts comprised 3 to 6%, and recruits
(apparently regenerating from seed) 81 to 76%
(Table 2).
We classified those taxa that made up the vast majority
of pioneers in this system as ‘‘pioneer’’ species: the
genera Cecropia (Cecropiaceae), Alchornea (Euphorbiaceae), and all genera in the Melastomataceae family
28
A.G. Van der Valk (ed.)
Table 2 Number of trees C5 cm dbh censused in both strips
before the clearing (from Cornejo and Gorchov 1993) and after
the clearing in 2004 for strip 1 and 2005 for strip 2 at CIJH,
Peru
Categories
Strip
1
Strip
2
662
248
619
228
strip 2: F1,16 = 0.55, P = 0.467). In strip 2, neither
felling (deferment-cut versus clear-cut, F1,16 = 2.97,
P = 0.104) nor the interaction of thinning and felling
(F1,16 = 0.57, P = 0.461) affected 2005 BA.
Tree species richness
Pre-clearing (1989)
No. of trees
No. of morphospecies
Post-clearing (2004–2005)
Trees [5 cm dbh and not cut in 1989
3
52
131
108
25
36
Recruits C5 cm dbh in 2004/2005
662
457
Total
No. of not identified taxa in the postclearing
821
1
653
2
172
176
Survivors (\5 cm dbh but [2 m tall in
1989)
Sprouts from stumps of trees C5 cm dbh in
1989
No. of species
Number of taxa is given in italics
40
Pre-clearing (1989)
1 year post-clearing (1990)
15 years post-clearing (2004-2005)
35
2
Basal Area (m /ha)
30
25
20
15
Before clearing (1989), strip 1 had 422 trees
[7.5 cm dbh, comprising 187 morphospecies (not
all trees were identified to the species level in the preclearing censuses), whereas in 2004 there were 494
trees and 97 species. For strip 2, in 1989 there were
391 trees comprising 192 morphospecies compared to
410 trees and 109 species in 2005. Total number of
trees and species C5 cm dbh found in 1989 and in the
post-clearing censuses (2004/2005) of each strip are
reported in Table 2.
In both strips silvicultural thinning did not affect the
2004/2005 tree species density (strip 1: F1,18 = 1.85,
P = 0.191; strip 2: F1,16 = 0.01, P = 0.926); similarly, neither felling (F1,16 = 0.32, P = 0.580), nor the
interaction of thinning and felling (F1,16 = 0.08,
P = 0.781) affected the 2005 species density in strip
2. Fifteen years into the second rotation, strip 1 and the
clear-cut portion of strip 2 recovered 47 and 45% of
their pre-clearing richness, at equal sample sizes. The
deferment-cut portions of strip 2 recovered 68% of its
pre-clearing richness. Rarefaction curves for strip 1
and the clear-cut portion of strip 2 showed that species
10
Pre-clearing 1989
Post-clearing 2004
5
220
0
Deferment-cut
Strip 2
200
Clear-cut
Strip 2
2
Fig. 2 Stand basal area (m /ha) of strip 1 and strip 2 before the
clearing (1989), a year after the clearing (1990), and 15 years
(2004) after the clearing (strip 1—2004, strip 2—2005)
180
No. Species (S)
Strip 1
160
140
120
100
80
60
Stand basal area
40
20
After 15 years of the first cutting, strip 1 and strip 2
recovered 73% (21 m2/ha) and 58% (17 m2/ha) of
their original BA (Fig. 2), whereas the deferment-cut
portion of strip 2 recovered 75% (26 m2/ha) of its
pre-clearing BA (Fig. 2). Silvicultural thinning did
not affect 2004/2005 BA of trees [10 cm dbh in the
strips (in strip 1, F1,18 = 3.44, P = 0.080, and in
0
0
50
100
150
200
250
300
350
400
450
500
550
No. Trees (N)
Fig. 3 Sample based rarefaction curves for 1989 (N = 417)
and 2004 (N = 494) for trees [7.5 cm dbh in strip 1. Dotted
lines are 95% CI. Number of samples was rescaled to number
of individuals. Vertical line indicates species richness at equal
sample sizes
Forest Ecology
29
A Clear-cut
Pre-clearing 1989
Post-clearing, 2005
220
200
180
No. Species (S)
160
140
120
100
80
60
40
20
0
0
50
100
150
200
250
compositional similarity of two mature stands
(L = 0.855, Fig. 5). In strip 2, compositional similarity of 1989 vs. 2005 in the clear-cut (L = 0.592)
and the deferment-cut portion (L = 0.656) was lower
than the compositional similarity of two mature
stands (Fig. 5). Thinning did not affect the compositional similarity of trees [7.5 cm dbh between
1989 and 2004 in strip 1 (F1,18 = 3.78, P = 0.068)
or in strip 2 (F1,16 = 0.39, P = 0.542). In strip 2,
neither felling treatment (F1,16 = 1.03, P = 0.324)
nor the interaction of felling and thinning
(F1,16 = 0.72, P = 0.408) significantly affected the
compositional similarity between 1989 and 2005.
B Deferment-cut
220
Commercial species
200
180
No. Species (S)
160
140
120
100
80
60
40
20
0
0
50
100
150
200
250
No. Trees
Fig. 4 Sample based rarefaction curves for the (a) clear-cut
portion and (b) deferment-cut portion of strip 2 in 1989 and
2005 for trees[7.5 cm dbh. In the clear-cut portion, there were
196 trees in 1989 and 221 trees in 2005. In the deferment-cut
portion, there were 195 trees in 1989 and 189 trees in 2005.
Dotted lines are 95% CI. Number of samples was rescaled to
number of individuals. Vertical line indicates species richness
at equal sample sizes
richness was significantly lower in 2004/2005 than in
1989, as these curves diverged clearly and confidence
intervals did not overlap (Figs. 3, 4a). In the deferment-cut portion of strip 2, rarefaction curves showed
overlapping confidence intervals of species richness at
smaller sample sizes (N \ 75, Fig. 4b), but clearly
diverged at greater sample sizes. Thus, species richness
in the deferment-cut portion was also lower in 2005.
Tree composition
In 2004/2005, the strips had recovered more than
50% of the compositional similarity with the preclearing censuses. In strip 1, compositional similarity
of 1989 vs. 2004 (L = 0.828) was slightly lower than
The relative abundance of commercial species was
lower in 2004/2005 than in 1989 in strip 1 (thinned
plots: t = 6.44, P \ 0.01; unthinned plots: t = 7.99,
P \ 0.01), the clear-cut portions of strip 2 (t = 5.83,
P \ 0.001), and deferment-cut portions of strip 2
(t = 3.56, P \ 0.01) (Fig. 6a). Strip 1 and the clearcut portion of strip 2 recovered 25 and 43%,
respectively, of the relative abundance of commercial
species in the pre-clearing censuses, whereas the
deferment cut portions of strip 2 recovered 67%.
Silvicultural thinning tripled the relative abundance
of commercial species in one of the strips in 2004
(F1,18 = 6.29, P = 0.022). However, thinning did not
significantly affect the relative abundance of commercial species in strip 2 (F1,16 = 2.52, P = 0.132). In
strip 2, deferment-cut plots almost doubled the relative
abundance of commercial species found in clear-cut
plots (F1,16 = 6.52, P = 0.021), but the interaction of
thinning and felling treatment (F1,16 = 0.40,
P = 0.534) did not have an effect. In 1989, the BA
of commercial species in strip 1 and the clear-cut
portion of strip 2 were both about 14 m2/ha, and in the
deferment-cut portion of strip 2 was 18 m2/ha. In 2004/
2005 the BA of commercial species was 2 m2/ha in
strip 1 and 3 m2/ha in the clear-cut portion of strip 2, 14
to 21% of their 1989 BA, whereas in the deferment-cut
portion of strip 2 BA for these species was 6 m2/ha,
33% of its 1989 BA (Fig. 7).
Pioneer species
Pioneer species were still abundant in 2004/2005, 65
and 62% of all trees ([7.5 cm dbh) belonged to
30
A.G. Van der Valk (ed.)
Fig. 5 Compositional
similarity (L) of strip 1
between 1989 and 2004;
clear-cut and deferment-cut
halves of strip 2 between
1989 and 2005, and mature
forest (strip 1 versus strip 2)
in 1989 at the genus level,
using the abundance-based
Sorensen index
1.0
Compositional Similarity (L)
0.8
0.6
0.4
0.2
0.0
1989 vs. 2004
Strip 1
A 1.0 Commercial Species
1989
Post-clearing
Relative Abundance per plot
0.9
0.8
0.7
0.6
0.5
0.4
*
*
*
*
0.3
0.2
0.1
0.0
Thinned
Strip 1
Unthinned
Strip 1
Clear-Cut
Strip 2
Deferment-cut
Strip 2
Relative Abundance per plot
B 1.0 Pioneer Species
0.9
0.8
0.7
0.6
0.5
*
*
*
*
*
*
0.4
0.3
0.2
0.1
0.0
Thinned Unthinned Thinned Unthinned Thinned Unthinned
Strip 1
Strip 1 Clear-Cut Clear-Cut Deferm.- Deferm.Strip 1
Strip 1
Cut,
Cut,
Strip 2
Strip 2
Fig. 6 (a) Mean (?SE) relative abundance of commercial
species in thinned and unthinned 15 9 15 m plots of strip 1,
and the clear-cut and deferment-cut plots of strip 2 in 1989 and
post-clearing (2004, 2005) censuses. (b) Mean (?SE) relative
abundance of pioneer species in thinned and unthinned plots of
strip 1, and thinned and unthinned clear-cut, and deferment-cut
plots of strip 2, in 1989 and post-clearing censuses. Asterisks
(*) indicate significant difference between 1989 and postclearing census
Clear-cut 1989 vs.
2005 Strip 2
Deferment-cut 1989 vs.
2005 Strip 2
Strips 1-2 1989
pioneer species in strip 1 and the clear-cut portion of
strip 2, respectively. In the deferment-cut portion of
strip 2 only 42% of the trees belonged to pioneer
species. As expected the relative abundance of
pioneer species was higher in 2004/2005 than in
1989 regardless of thinning treatment in strip 1
(thinned plots of strip 1: t = 13.37, P \ 0.001;
unthinned plots of strip 1: t = 27.93, P \ 0.001),
the clear-cut portion of strip 2 (thinned plots:
t = 8.32,
P \ 0.01;
unthinned:
t = 13.80,
P \ 0.01), and the deferment-cut portion of strip 2
(thinned: t = 3.45, P = 0.018; unthinned: t = 6.03,
P \ 0.01, Fig. 6b). Silvicultural thinning and deferment-cutting both reduced the relative abundance of
pioneer species. In 2004/2005 unthinned plots had 19
to 36% greater relative abundance of pioneer species
than thinned plots in strip 1 and the clear-cut portion
of strip 2; in the deferment-cut, unthinned plots
doubled thinned plots in relative abundance of
pioneer species (strip 1: F1,18 = 21.10, P \ 0.001;
strip 2: F1,16 = 11.37, P \ 0.01). In strip 2, clear-cut
plots had greater abundance of pioneers than the
deferment-cut plots (F1,16 = 10.41, P \ 0.01), and in
some case doubled the amount of pioneers. However,
the interaction of felling treatment and thinning
(F1,16 = 2.73, P = 0.118) did not have an effect on
pioneer species. In 1989 the BA of pioneer species in
both strip 1 and the clear-cut portion of strip 2 were
about 1 m2/ha, compared to the deferment-cut
portion of strip 2 which was about 0.2 m2/ha. In
2004/2005, the BA of commercial species was 19 m2/
ha in strip 1 and 12 m2/ha in the clear-cut portion of
strip 2. The BA of commercial species of the
Forest Ecology
Commercial
Other
Pioneer
100
Percent of Stand Basal Area (%)
Fig. 7 Percent basal area
of commercial, pioneer, and
‘‘other’’ species
[7.5 cm dbh for strip 1 in
1989 and 2004, and in the
clear-cut and deferment-cut
portions of strip 2 in 1989
and 2005
31
80
60
40
20
0
1989 Strip 1
1989 Clear-cut
Strip 2
2005 Clear-cut 1989 Deferment- 2005 DefermentStrip 2
Cut Strip 2
Cut Strip 2
700
Trees 5 to 10 cm dbh
Trees > 10 cm dbh
600
Timber Stocking (no. stems/ha)
Fig. 8 Timber stocking
(no. of commercial stems/
ha) of small trees (5–
10 cm dbh) and large trees
([10 cm dbh) in the preclearing (1989) and postclearing period (2004/2005)
for strip 1, and the
deferment-cut and clear-cut
portions of strip 2
2004 Strip 1
500
400
300
200
100
0
1989 Strip 1
deferment-cut portion of strip 2 was 10 m2/ha in
2005. Figure 7 shows the percent BA of commercial,
pioneer, and ‘‘other’’ species in 1989 and 2004/2005.
Stocking of commercial stems
In both strips, timber stocking of large stems
([10 cm dbh) was lower in 2004/2005 than in 1989
(Fig. 8). In strip 1, stocking of large commercial
stems recovered 11% of its pre-clearing value (33 vs.
304 stems/ha). The clear-cut and deferment-cut portions of strip 2 recovered 27% (76 vs. 280 stems/ha)
and 59% (178 vs. 302 stems/ha) of their pre-clearing
stocking, respectively. Stocking of small stems (5 to
10 cm dbh) in 2004/2005 was similar to pre-clearing
2004 Strip 1
1989 ClearCut, Strip 2
2005 ClearCut, Strip 2
1989 Deferm.- 2005 Deferm.Cut, Strip 2
Cut, Strip 2
levels, and greater than stocking of large stems
(Fig. 8). In both strips, the 1996 silvicultural thinning
treatment did not affect the stocking of small (strip 1:
F1,18 = 2.50, P = 0.131; strip 2: F1,16 = 0.30,
P = 0.590) and large commercial stems (strip 1:
F1,18 = 1.68, P = 0.211; strip 2: F1,16 = 0.91,
P = 0.355) in 2004/2005. In 2005, the defermentcut plots of strip 2 had greater than twice as much
stocking of large commercial stems than the clear-cut
plots (F1,16 = 7.60, P = 0.014), but similar stocking
of small commercial stems (F1,16 = 0.23, P = 0.637,
Fig. 8). The interaction of thinning and felling
affected neither the stocking of small (F1,16 = 0.80,
P = 0.385)
nor
large
commercial
stems
(F1,16 = 0.00, P = 0.961).
32
Discussion
Basal area recovery
The recovery of a high percentage of stand BA
15 years after clear-cutting (73% in strip 1 and 58%
in the clear-cut portion of strip 2) is consistent with
rapid BA growth in the early years of secondary
succession (Saldarriaga et al. 1988; Moran et al.
1996; Denslow and Guzman 2000), although this
strongly depends on land use history and site
productivity. BA of forest stands 12 to 18 years after
clear-cutting for pulp in Colombia did not exceed
50% of old growth values (Faber-Landgendoen
1992). In Brazil, BA recovery 11 to 12 years after
clear-cutting treatment was 50% of undisturbed forest
and 60% of its pre-clearing value (Parrotta et al.
2002). Parrota et al. (2002) also compared BA
recovery of different systems 11 to 12 years after
harvesting. They found that high intensity harvesting
or clear-cut (removal of 373 m3, all above-ground
biomass) had a lower BA recovery (50%) than
moderate harvesting (trees B20 cm and C60 cm dbh
for a total removal of 219 m3) (68%), and low
harvesting (trees C45 cm dbh for a total of 201 m3)
treatments (68%). Thus, the recovery of BA in this
study was comparable to that reported for moderate
harvest in Brazil (Parrotta et al. 2002) and somewhat
higher than clear-cutting in Colombia (Faber-Landgendoen 1992).
Species richness recovery
The strips in the pre-clearing stage had high species
richness: estimates reported in Table 2 underestimate
the true richness since identification of some trees
was done to morphospecies. Therefore, the extent to
which species richness recovered after 15 years to
pre-clearing values (47% in strip 1 and 45% in the
clear-cut portion of strip 2) is probably slightly
overestimated. Nevertheless, this was similar to the
recovery 18 years after clear-cutting for pulp in
Colombia, \50% of mature forest (trees
C10 cm dbh, Faber-Landgendoen 1992). Less intensive harvesting systems, however, have greater
species-richness recovery. Parrotta et al. (2002)
reported lower species richness recovery of trees
C15 cm dbh following clear-cut treatment (32%)
versus moderate (59%) and low harvesting (94%)
A.G. Van der Valk (ed.)
treatments after 11 to 12 years. In a dipterocarp forest
in Borneo, Cannon et al. (1998) found that samples
8 years after selective logging (removal of 43% of
stand BA) had as many tree species as unlogged
forest.
Several studies have found that species richness
tends to be more similar in secondary growth and old
growth when smaller tree size classes are compared
(Saldarriaga et al. 1988; Faber-Landgendoen 1992;
Aide et al. 1996; Guariguata et al. 1997; Magnusson
et al. 1999; Denslow and Guzman 2000; Parrotta et al.
2002; Peña-Claros 2003). We were not able to make
such comparisons in our study due to incomplete preclearing datasets for smaller trees in both strips.
Composition recovery
While species richness increases in the early years of
secondary succession, and takes only a few decades
to reach old growth values when land use has not
been severe and seed sources are close, composition
of these forests remains different from old growth and
may take longer to become similar to old growth
stands (Finegan 1996; Guariguata and Ostetarg
2001). In our study, the strips recovered more than
50% of their pre-clearing composition at the genus
level. If the analysis had been done at the species
level, compositional similarities would have been
lower, but genus-level analysis was conservative in
the face of possible inconsistencies between censuses
in some species identification, and is often done in
studies of diverse tropical rainforests (e.g., Laurance
et al. 2004). Despite this high composition recovery
in the strips, the relative abundance and basal area of
commercial and pioneer species were far from
reaching pre-clearing levels.
Recruitment of commercial species after harvesting is difficult due to the different environmental
conditions required by different species for regeneration. Although Swaine and Whitmore (1988)
considered most commercial species gap-dependent,
commercial seedlings have a broad range of shadetolerances (Martini et al. 1994; Pinard et al. 1999).
Out of 31 timber species (of high and low commercial value) studied by Pinard et al. (1999), 45% were
shade intolerant and regenerated in forest edges and
large gaps, 36% were shade-tolerant and regenerated
in the understory, and 19% were in between the latter
groups and regenerated under partial shade or small
Forest Ecology
gaps. Similarly, Martini et al. (1994) classified timber
species of the Brazilian Amazon.
Recruitment from seed was more important in our
system than stump sprouts or advance regeneration.
Sprouting of timber species in Amazonia is common;
out of 305 timber species saplings, 87% of them
produced sprouts following the breaking and crushing
injuries associated with logging (Martini et al. 1994).
In the strips, however, stump sprouts and the advance
regeneration had a minor role in tree regeneration
(Table 2) and the regeneration of commercial species. Although 41% of the stumps ([7.5 cm dbh) had
one or more living sprouts, 10 months after cutting
one of the strips (Gorchov et al. 1993), only four
sprouting stumps in each strip (unpublished data)
were of commercial value after 15 years. A high
density of saplings (903/ha), belonging to mature
forest trees, including many of commercial value,
survived the clearing operation in 1989 (Gorchov
et al. 1993), but 15 years later these only comprised a
small percentage (16–18%) of the total regeneration
in the strips, a little higher than the sprouting stumps.
Low seed input and/or high seed predation of
commercial species could have lowered the recruitment of commercial species into the strips, resulting
in low stocking and relative abundance of commercial trees in the strips 15 years later. Using seed traps
aboveground, Gorchov et al. (1993) showed that very
few large seeds, characteristic of timber species, were
dispersed into the strips by birds or bats, one year
after clear-cutting. Also, seeds of a valuable timber
species, Hymenaea courbaril, were rarely moved by
rodents into the interior of a strip, 10 to 30 months
after the clearing (Gorchov et al. 2004). Predation of
timber seeds (Pouteria sp.), was also greater in the
strips than in the surrounding forest, 3 years after
strip clear-cutting (Notman et al. 1996). Once established, commercial species compete for light with
vines, lianas, and short-lived pioneer species that
quickly colonize logged areas (Buschbacher 1990;
Fredericksen and Mostacedo 2000; Pariona et al.
2003). As a result, growth and BA of commercial
species often respond to logging less favorably than
faster growing species of low commercial values
(Silva et al. 1995; Kammessheidt 1998).
After 15 years of regeneration, timber stocking of
small stems (5–10 cm dbh) in both strips was similar to
pre-clearing levels. However, stocking of larger stems
([10 cm dbh) was low (33.3–75.5 stems/ha) and far
33
from reaching pre-clearing levels (300 stems/ha,
Fig. 8), and mature forest levels (233 stems/ha in
Peters et al. (1989)), and lower than in a 50 year-old
communal forest near Iquitos (125.5/ha for trees
[25 cm dbh in Pinedo-Vasquez et al. 1990). This low
stocking of large commercial stems in this system
negatively affects the economic value projected for a
potential second harvest after 25 years (Rondon 2008).
On the other hand, pioneers with large basal areas
were still abundant in 2004/2005, 8 to 9 years after the
thinning treatment. In the study of clear-cutting for
pulp, pioneer species in a 12-year old forest comprised
more than 50 to 60% of basal area and biomass (FaberLandgendoen 1992); Parrotta et al. (2002) found that
although tree floras within low, moderate, and intensive (clear-cut) harvesting treatments were broadly
similar to those of undisturbed plots after 11 years; the
clear-cut treatment was dominated by a higher proportion of short-lived early successional tree species,
including Cecropia and Vismia.
One year after the clearing, the majority of the
seedlings in the strip were a few bat (Cecropia)- and
bird-dispersed (Melastomataceae and Alchornea triplinervia) pioneer tree species (Gorchov et al. 1993).
Cecropia membranacea, one of the species with the
most seedlings in the strips, was also present in the
seed bank; other tree seedlings, not represented in the
seed bank, were attributed to the seed rain (Gorchov
et al. 1993). Seeds from the seed bank as well as
recently dispersed seeds contribute to the development of secondary forest. In a tropical forest of
Mexico, all viable seeds of Cecropia obtusifolia were
renewed from the soil almost every year; seed loss
was mainly due to pathogen attack and high predation
rates, but the seed bank was continually replenished
by seed rain (Alvarez-Buylla and Martı́nez-Ramos
1990). It is very likely that the pioneer trees that
currently dominate the strips depended on seed
dispersal events that followed the clearing of the
strips. One year after clearing one of the strips, batand wind-dispersed seeds accounted for more seed
dispersal in the strip interior than bird-dispersed
seeds, which arrived at high density within the forest
or strip edge (Gorchov et al. 1993). Fifteen years
after the felling, pioneer species comprised 65 and
62% of the trees in strip 1 and the clear-cut portion of
strip 2, respectively.
Germination and establishment of short-lived
pioneer species (such as Cecropia) can be reduced when
34
residual vegetation and litter are present (Uhl et al.
1981; Putz 1983; Molofsky and Augspurger 1992). In
this study, only slash\2.5 cm was left on site (Cornejo
and Gorchov 1993). Although substantial, this amount
of litter was apparently not sufficient to suppress
germination and establishment of pioneer species.
In Jenaro Herrera, pioneer species such as Cecropia, Alchornea, Miconia, and Vismia spp. have been
found to be dominant in 14 and 17-year old fallows
(Baluarte Vásquez 1998). Dominance of few pioneers
that established early in succession tends to ‘‘break
up’’ within \25 years (Denslow and Guzman 2000).
Senescence and mortality of these species will have a
strong impact on the future biomass and stem density
of secondary stands (Feldpausch et al. 2007). Thus,
BA recovery in the strips is not likely to increase
continuously over the next years unless there is
higher growth of commercial and ‘‘other’’ species
into larger size classes.
Silvicultural thinning
Liberation treatments such as thinning of lianas and
pioneer species are commonly used to improve
recruitment and tree growth (de Graaf et al. 1999;
Guariguata 1999, 1997; Dolanc et al. 2003; Pariona
et al. 2003). In this study, silvicultural thinning in 1996
was sufficient to significantly increase the 1996–2000
growth of commercial species (Dolanc et al. 2003),
and to reduce 2004/2005 relative abundance of pioneer
species of both strips, although pioneers were still
abundant in the post-clearing censuses of both strips.
Thinning also increased the relative abundance of
commercial species significantly in one of the strips.
However, thinning did not have an effect on basal area,
compositional similarity, or timber stocking 8 to
9 years after the treatment application. The lack of
effects of thinning on these community parameters
might be because large Alchornea and melastomes that
were not thinned, because some of the girdled pioneer
trees did not die, and/or due to increased growth of the
trees remaining in the thinned plots.
Deferment-cut
Deferment-cutting appeared to be more sustainable
than clear-cutting. The deferment-cut portion of strip
2 had greater BA, species richness, and composition
recovery than the clear-cut portion. The deferment-
A.G. Van der Valk (ed.)
cut portion also had higher representation, stocking,
and BA of commercial species, and a lower percentage of pioneers, than the clear-cut portion. This better
recovery of the deferment-cut is consistent with the
well documented role of remnant or residual vegetation in promoting recovery of species richness, tree
density, and aboveground biomass (Guariguata and
Ostetarg 2001; Parrotta et al. 2002; Chazdon 2003).
The Palcazú forest management system
Tosi (1982) and Hartshorn (1989a) proposed harvesting cycles of 30 to 40 years for the strip clear-cutting
system. Tree regeneration in the two clear-cut strips,
15 years into the second harvesting, suggests that this
system may not be ecologically sustainable, but this
conclusion is tempered by replication constraints at
the plot and site scale of this study.
Both strips showed some inherent variability in the
pre- and the post-clearing censuses, especially in the
recovery of commercial species. Predicting species
richness and composition of the strips in the next 15
to 25 years would be difficult because this system
would still be affected by variability in recruitment,
growth, and mortality rates of commercial, pioneer,
and ‘‘other’’ species due to biotic and abiotic factors.
Thus far, 15 years into the regeneration, our results
reveal that in this system regeneration of pioneer
species exceed that of commercial species, even
when the strips are surrounded by a matrix of old
growth forest. In a forest managed by the strip clearcutting system as it was originally proposed for the
Palcazú, 44,000 ha would be under management for
timber production (Hartshorn 1989b), and about half
of the area would be cleared (Hartshorn 1989b); thus,
eventually the surrounding matrix for many of the
strips would be that of young growth. Therefore, the
species that would thrive in these strips would be the
ones that can reproduce within the cutting cycle of
30–40 year; i.e., pioneers. Contrary to predictions of
Tosi (1982) and Hartshorn (1989a), pioneer species
dominate the composition of the strips 15 years into
the regeneration. Unless pioneer species have a high
mortality rate in the coming years, and there is more
recruitment of commercial and ‘‘other’’ species into
the larger size classes, this system is not sustainable.
Two approaches could be taken to reduce the
number of pioneer species in the strips. It is possible
that cutting narrower strips (\30 m) in this system
Forest Ecology
may reduce the amount of light entering the strip and
thus, the germination and establishment of pioneer
species. Periodic silvicultural thinning treatments
may further reduce the abundance of pioneer species
and further increase the establishment and growth of
more commercial and ‘‘other’’ species in the strips.
We are aware that in the future high quality timber
species will become scarce due to their high demand
and strong extraction pressures. International markets
will start accepting a broader range of lower quality
timber species that are also gap-dependent, but this
market will take some time to develop. In this study
we were interested in studying the regeneration of
timber species that already have an established
market in order to assess the value of the strips in a
potential second harvest.
From the economic perspective, composition in a
forest management system has a great influence on
the financial value of the next harvest. Relative
abundance, stocking, and growth of commercial
species will determine whether the second harvest
(which is in the next 15–25 years) will be financially
profitable. In order to fully assess the economic
viability of this system, we have also investigated
whether those few large commercial trees in the strips
would reach marketable size in the next 25 years, in
time for a second cutting (Rondon 2008).
Acknowledgements We thank Dr. Dennis del Castillo, Ing.
Euridice Honorio, Ing. Gustavo Torres, and the Instituto de
Investigaciones de la Amazonı́a Peruana (IIAP) for allowing us
to conduct this study at Centro de Investigaciones Jenaro
Herrera (CIJH). We thank the Instituto de Recursos Naturales
(INRENA) for providing collecting and exportation permits as
well as Zunilda Rondón for help in the application process. We
also thank Italo Melendez and Margarita Jaramillo for
assistance in the field. Identification was performed with the
help of Rodolfo Vásquez at Missouri Botanical Garden
(MOBOT), Nállaret Dávila at CIJH herbarium, and César
Grandés at Herbario Amazonense (AMAZ). We thank Tom
Crist, Hank Stevens, and anonymous reviewers for comments
on earlier drafts of this manuscript. This study was funded by
USAID Program in Science and Technology Cooperation,
Grant no. 7228 to J. Terborgh, D. Gorchov and F. Cornejo and
by Academic Challenge Grant (Botany, Miami University),
Garden Club of Ohio, Sigma Xi, and Hispanic Scholarship
Fund grants awarded to X. J. Rondon.
References
Aide TM, Zimmerman JK, Rosario M, Marcano H (1996)
Forest recovery in abandoned cattle pastures along
35
elevational gradient in Northeastern Puerto Rico. Biotropica 28:537–548. doi:10.2307/2389095
Allison TD, Art HW, Cunningham FE, Teed R (2003) Fortytwo years of succession following strip clearcutting in a
northern hardwoods forest in northwestern Massachusetts.
For Ecol Manag 182:285–301. doi:10.1016/S0378-1127
(03)00066-5
Alvarez-Buylla ER, Martı́nez-Ramos M (1990) Seed bank
versus seed rain in the regeneration of a tropical pioneer
tree. Oecologia 84:314–325
Ascorra CF, Gorchov DL, Cornejo F (1993) The bats from
Jenaro Herrera, Loreto, Peru. Mammalia 57:533–552
Baluarte Vásquez J (1998) Composición floristica de los bosques secundarios de Jenaro Herrera, Loreto. Folia
Amazonica 9:195–218
Bawa KS, Seidler R (1998) Natural forest management and
conservation of biodiversity in tropical forests. Conserv
Biol 12:46–55. doi:10.1046/j.1523-1739.1998.96480.x
Buschbacher RJ (1990) Natural forest management in the
humid tropics: ecological, social, and economic considerations. Ambio 19:253–258
Cannon CH, Peart DR, Leighton M (1998) Tree species diversity
in commercially logged Bornean rainforest. Science
281:1366–1368. doi:10.1126/science.281.5381.1366
Chao A, Chazdon LR, Colwell RK, Shen T (2005) A new
statistical approach for assessing similarity of species
composition with incidence and abundance data. Ecol Lett
8:148–159. doi:10.1111/j.1461-0248.2004.00707.x
Chazdon RL (2003) Tropical forest recovery: legacies of
human impact and natural disturbances. Perspect Plant
Ecol Evol Syst 6:51–71. doi:10.1078/1433-8319-00042
Colwell RK (2005) EstimateS: Statistical estimation of species
richness and shared species from samples. Version 7.5.
Persistent URL www.purl.oclc.org/estimates. Cited 24
Oct 2007
Cornejo F, Gorchov DL (1993) La tala rasa en fajas, el aprovechamiento integral del recurso y rendimiento
económico: El caso de Jenaro Herrera, Loreto, Peru. Rev
Forestal Peru 20:21–37
de Graaf NR, Poels RLH, Van Rompaey RSAR (1999) Effects
of silvicultural treatment on growth and mortality of
rainforest in Surinam over long periods. For Ecol Manag
124:123–135. doi:10.1016/S0378-1127(99)00057-2
Denslow JS, Guzman S (2000) Variation in stand structure,
light, and seedling abundance across a tropical moist
forest chronosequence, Panama. J Veg Sci 11:201–212.
doi:10.2307/3236800
Dolanc C, Gorchov DL, Cornejo F (2003) The effects of silvicultural thinning on diversity and demography of trees
regenerating in strip cuts in the Peruvian Amazon. For Ecol
Manag 182:1–14. doi:10.1016/S0378-1127(03)00012-4
Faber-Landgendoen D (1992) Ecological constraints on rain
forest management at Bajo Calima, western Colombia.
For Ecol Manag 53:213–244. doi:10.1016/0378-1127(92)
90044-A
Feldpausch TR, da Conceicao Prates-Clark C, Fernandes ECM,
Riha SJ (2007) Secondary forest growth deviation from
chronosequence predictions in Central Amazonia. Glob
Change Biol 13:967–979
Finegan B (1996) Patterns and process in neotropical secondary rain forests: the first 100 years of succession. Trends
36
Ecol Evol 11:119–124. doi:10.1016/0169-5347(96)
81090-1
Fredericksen TS, Mostacedo B (2000) Regeneration of sawtimber species following selective logging in a Bolivian
tropical forest. For Ecol Manag 131:47–55. doi:
10.1016/S0378-1127(99)00199-1
Gentry AH (1993) A field guide to the families and genera of
woody plants of Northwest South America (Colombia,
Ecuador, Peru). Conservation International, Washington,
DC, 895 pp
Gorchov DL, Cornejo F, Ascorra C, Jaramillo M (1993) The
role of seed dispersal in the natural regeneration of rain
forest after strip-cutting in the Peruvian Amazon. Vegetatio 107(108):339–349
Gorchov DL, Palmeirim JM, Jaramillo M, Ascorra C (2004)
Dispersal of seeds of Hymenaea courbaril (Fabaceae) in a
logged rain forest in the Peruvian Amazon. Acta Amazon
34:265–273. doi:10.1590/S0044-59672004000200014
Gram S (1997) Indian forestry in Latin American rain forests.
Sustain Dev 5:21–29. doi :10.1002/(SICI)1099-1719
(199703)5:1\21::AID-SD54[3.0.CO;2-I
Guariguata MR (1999) Early response of selected tree species
to liberation thinning in young secondary forest in
Northeastern Costa Rica. For Ecol Manag 124:255–261.
doi:10.1016/S0378-1127(99)00072-9
Guariguata MR, Ostetarg R (2001) Neotropical secondary
forest succession: changes in structural and functional
characteristics. For Ecol Manag 148:185–206. doi:
10.1016/S0378-1127(00)00535-1
Guariguata MR, Chazdon RL, Denslow JS, Dupuy JM,
Anderson L (1997) Structure and floristics of secondary
and old-growth forest stands in lowland Costa Rica. Plant
Ecol 132:107–120. doi:10.1023/A:1009726421352
Hartshorn GS (1989a) Application of gap theory to tropical
forest management: natural regeneration on strip clearcuts in the Peruvian Amazon. Ecology 70:567–569. doi:
10.2307/1940208
Hartshorn GS (1989b) Sustained yield management of natural
forests. In: Browder JO (ed) Fragile lands of Latin
America. Westview Press, London
Hartshorn GS (1995) Ecological basis for sustainable development in tropical forests. Annu Rev Ecol Syst 26:155–
175. doi:10.1146/annurev.es.26.110195.001103
Heitzman E, Pregitzer KS, Miller RO, Lanasac M, Zuidema M
(1999) Establishment and development of northern whitecedar following strip clearcutting. For Ecol Manag
123:97–104. doi:10.1016/S0378-1127(99)00025-0
ITTO (1997–2005) Annual review and assessment of the world
timber situation. International Tropical Timber Organization, Yokohama
Kammessheidt L (1998) Stand structure and spatial pattern of
commercial species in logged and unlogged Venezuelan
forest. For Ecol Manag 109:163–174. doi:10.1016/S03781127(98)00245-X
Laurance WF, Oliveira AA, Laurance SG, Condit R, Nascimento HEM, Sanchez-Thorin AC et al (2004) Pervasive
alteration of tree communities in undisturbed Amazonian
forests. Science 248:171–175
Lieberman D, Lieberman M, Hartshorn GS, Peralta R (1985)
Growth rates and age-size relationships of tropical wet
forest trees in Costa Rica. J Trop Ecol 1:97–109
A.G. Van der Valk (ed.)
Magnusson WE, de Lima OP, Quintiliano-Reis F, Higuchi N,
Ferreira-Ramos J (1999) Logging activity and tree
regeneration in an Amazonian forest. For Ecol Manag
113:67–74. doi:10.1016/S0378-1127(98)00418-6
Martini AMZ, de Rosa AN, Uhl C (1994) An attempt to predict
which Amazonian tree species may be threatened by
logging activities. Environ Conserv 21:152–161
Molofsky J, Augspurger CK (1992) The effect of leaf litter on
early seedling establishment in a tropical forest. Ecology
73:68–77. doi:10.2307/1938721
Moran EF, Packer A, Brondizio E, Tucker J (1996) Restoration
of vegetation cover in eastern Amazonia. Ecol Econ
18:41–54. doi:10.1016/0921-8009(96)00034-1
Notman E, Gorchov DL, Cornejo F (1996) Effect of distance,
aggregation, and habitat on levels of seed predation for
two mammal-dispersed neotropical rain forest tree species. Oecologia 107:221–227. doi:10.1007/BF00328602
Pariona W, Fredeicksen TS, Licona JC (2003) Natural regeneration and liberation of timber species in logging gaps in
two Bolivian tropical forests. For Ecol Manag 181:313–
322. doi:10.1016/S0378-1127(03)00002-1
Parrotta JA, Francis JK, Knowles OH (2002) Harvesting
intensity affects forest structure and composition in an
upland Amazonian forest. For Ecol Manag 169:243–255.
doi:10.1016/S0378-1127(01)00758-7
Peña-Claros M (2003) Changes in forest structure and species
composition during secondary forest succession in the
Bolivian Amazon. Biotropica 35:450–461
Peters CM, Gentry AH, Mendelsohn O (1989) Valuation of an
Amazonian rainforest. Nature 339:655–656. doi:10.1038/
339655a0
Pinard MA, Putz FE, Rumı́z D, Guzmán R, Jardim A (1999)
Ecological characterization of tree species for guiding
forest management decisions in seasonally dry forests in
Lomerı́o, Bolivia. For Ecol Manag 113:201–213. doi:
10.1016/S0378-1127(98)00426-5
Pinedo-Vasquez M, Zarin D, Jipp P, Chota-Inuma J (1990)
Use-values of tree species in a communal forest reserve in
Northeast Peru. Conserv Biol 4:405–416. doi:10.1111/
j.1523-1739.1990.tb00315.x
Putz FE (1983) Treefall pits and mounds, buried seeds, and the
importance of soil disturbance to pioneer trees on Barro
Colorado Island, Panama. Ecology 64:1069–1074. doi:
10.2307/1937815
Rondon XJ (2008) Assessing the sustainability of strip clearcutting in the Peruvian Amazon. Dissertation, Miami
University, Oxford, OH
Saldarriaga JG, West DC, Tharp ML, Uhl C (1988) Long-term
chronosequence of forest succession in the Upper Rio
Negro of Colombia and Venezuela. J Ecol 76:938–958.
doi:10.2307/2260625
Silva JNM, de Carvalho JOP, do Lopes JCA, de Almeida BF,
Costa DHM, de Oliveira LC, Vanclay JK, Skovsgaard JP
(1995) Growth and yield of a tropical rain forest in the
Brazilian Amazon 13 years after logging. For Ecol Manag
71:267–274. doi:10.1016/0378-1127(94)06106-S
Simeone R (1990) Land use planning and forestry-based
economy: the case of the Amuesha Forestry Cooperative.
Tebiwa J Ida Mus Nat Hist 24:7–12
Smith DM (1986) The practice of silviculture, 8th edn. Wiley,
New York
Forest Ecology
Southgate D (1998) Tropical Forest Conservation: An economic assessment of the alternatives in Latin America.
Oxford University Press, Oxford, UK, 175 pp
Spichiger R, Meroz J, Loizeau PA, Stutz de Ortega L (1989)
Contribución a la Flora de la Amazonı́a Peruana. Los
Arboles del Arboretum Jenaro Herrera, vol 1. Conservatoire et Jardin Botaniques de Ginebra, Switzerland, 359 pp
Spichiger R, Meroz J, Loizeau PA, Stutz de Ortega L (1990)
Contribución a la Flora de la Amazonı́a Peruana. Los
Arboles del Arboretum Jenaro Herrera, vol 2. Conservatoire et Jardin Botaniques de Ginebra, Switzerland, 565 pp
Spichiger R, Loizeau PA, Latour C, Barriera G (1996) Tree
species richness of a South-Western Amazonian forest
(Jenaro Herrera, Peru, 73 400 W/4 540 S). Candollea
51:559–577
Swaine MD, Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests. Vegetatio
75:81–86. doi:10.1007/BF00044629
37
Thornton PL (1957) Effects of cutting methods on advance
reproduction in two mature mixed coniferous swamps in
Upper Michigan. J For 55:448–451
Tosi JA (1982) Sustained yield management of natural forests.
Consultant report for USAID/Peru. Tropical Science
Center. San Jose. Costa Rica, 53 pp
Uhl C, Clark K, Clark H, Murphy P (1981) Early plant succession after cutting and burning in the upper Rio Negro
region of the Amazon basin. J Ecol 69:631–649. doi:
10.2307/2259689
Uhl C, Jordan C, Clark K, Clark H, Herrera R (1982) Ecosystem recovery in Amazon caatinga forest after cutting,
cutting and burning, and bulldozer clearing treatments.
Oikos 38:313–320. doi:10.2307/3544671
Changing relationships between tree growth and climate
in Northwest China
Yongxiang Zhang Æ Martin Wilmking Æ
Xiaohua Gou
Originally published in the journal Plant Ecology, Volume 201, No. 1, 39–50.
DOI: 10.1007/s11258-008-9478-y Springer Science+Business Media B.V. 2008
Abstract Recently, several studies have shown
changing relationships between tree growth and climate
factors, mostly in the circumpolar north. There, changing relationships with climate seem to be linked to
emergent subpopulation behavior. Here, we test for
these phenomena in Northwest China using three tree
species (Pinus tabulaeformis, Picea crassifolia, and
Sabina przewalskii) that had been collected from six
sites at Qilian Mts. and Helan Mts. in Northwest China.
We first checked for growth divergence of individual
sites and then investigated the relationship between tree
growth and climate factors using moving correlation
functions (CF). Two species, Pinus and Sabina, from
two sites clearly showed growth divergence, not only in
the late twentieth century as reported in other studies,
but also over nearly the whole record. In divergent sites,
one chronology shows more stable relationships with
climate factors (usually precipitation). In non-divergent
sites, nearly all relationships either vary in strength or
become non-significant at one point. While this might
possibly be related to increased stress on some trees due
to increasing temperature, the exact causes for this shift
in sensitivity remain unclear. We would like to highlight
the necessity for additional studies investigating possible non-stationary growth responses of trees with
climate, especially at sites that are used for climate
reconstruction as our sites in Northwest China.
Keywords Growth divergence Northwest China
Paleoreconstruction Tree line Tree ring
Introduction
Y. Zhang X. Gou
Center for Arid Environment and Paleoclimate Research,
Key Laboratory of Western China’s Environmental
Systems, Ministry of Education, Lanzhou University,
Lanzhou 730000, People’s Republic of China
Y. Zhang
Institute of Tibetan Plateau Research, Chinese Academy
of Sciences, P.O. Box 2871, Beijing 100085,
People’s Republic of China
Y. Zhang (&) M. Wilmking
Ecosystem Dynamics, Institute for Botany and Landscape
Ecology, University Greifswald, Grimmer Strasse 88,
17487 Greifswald, Germany
e-mail: yz070767@uni-greifswald.de
Global warming is a great concern to human populations, as it has been shown to bring many threats, such
as heat waves and warmer weather, spreading disease,
earlier spring arrival, plant and animal range shifts and
population declines, sea level rise, and frequent
disaster (Greenough et al. 2001; Kahn 2005; Webster
et al. 2005; Shepherd and Wingham 2007; IPCC
2007). In order to evaluate whether current global
warming is unprecedented or not, it is essential to put
this warming into a long-term perspective. Due to the
limited time interval of instrumental climate records,
many natural proxies have been used to conduct
paleoclimatic reconstructions. Tree rings have been
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_4
39
40
widely used to reconstruct the variability of many
climate factors (e.g., temperature, precipitation), due to
their annual resolution, wide spatial distribution, and
the possibility of using simple linear models of
climate–tree growth relationships that seem to be
easily verified and calibrated (Hughes 2002).
It is generally assumed in dendroclimatological
studies that the approximate relationship between tree
growth and the limiting climate factor is stable over
time (Fritts 1976). However, many recent studies have
reported problems with this assumption. Either formerly temperature sensitive tree ring chronologies
have lost or decreased in temperature sensitivity
(Jacoby et al. 1996, 2000; Briffa et al. 1998a, b; Smith
et al. 1999; Solberg et al. 2002), increased in sensitivity (Knapp et al. 2001; Wilmking et al. 2005), or even
changed from positive to negative temperature sensitivity or vice versa (Wilmking et al. 2004, 2008).
In addition, temperature reconstructions in the
northern hemisphere based on tree rings are often not
able to follow the documented temperature increase
in recent decades, thus leading to a widening gap (a
divergence) between the temperature curve and the
tree ring-based temperature reconstruction. D’Arrigo
et al. (2008) has recently termed this phenomenon the
‘‘divergence problem’’ in northern forests.
Meanwhile, even trees growing at the same site
showed not only opposite response relationships with
climate factors, but also diverging long-term growth
trends in the late twentieth century, possibly diluting the
climate signal when averaged to site chronologies
(Wilmking et al. 2004, 2005; Driscoll et al. 2005;
Pisaric et al. 2007). These growth trend differences have
also been called ‘‘diverging.’’ In order to avoid possible
confusion, we will use the term ‘‘growth divergence’’ for
differences in growth trends between trees and ‘‘divergence problem’’ for the underestimation of recent
temperatures by tree ring-based climate reconstructions.
Neither the real reasons causing (1) the shift in tree
growth response to temperature (or possibly other
environmental factors), (2) the diverging between
recorded and reconstructed temperature, and (3) the
diverging growth trends between neighboring trees nor
the interaction between those three challenges are
known. Some possible mechanisms have been proposed to explain these shifts in tree growth–climate
relationships, such as temperature-induced drought
stress (Jacoby and D’Arrigo 1995; Barber et al. 2000;
Lloyd and Fastie 2002), non-linear thresholds, or time-
A.G. Van der Valk (ed.)
dependent responses to recent warming (D’Arrigo
et al. 2004; Wilmking et al. 2004; Sergio et al. 2007),
delayed snowmelt and related changes in seasonality
(Vaganov et al. 1999), air pollution (Wilson and Elling
2004; Yonenobu and Eckstein 2006), and differential
growth/climate relationships inferred for maximum,
minimum and mean temperatures (Wilson and Luckman 2002, 2003). In addition, there are also some other
potential causes, for example, end effects during
chronology development and biases in instrumental
target data and its modeling (Cook and Peters 1997;
Melvin 2004; Hoyt 2006; D’Arrigo et al. 2008).
Whatever the reasons for the growth divergence or
the divergence problem are, these phenomena seem
to be limited to the high latitudes of the northern
hemisphere (D’Arrigo et al. 2008). But do these
divergences and changes in climate sensitivity only
appear in the circumpolar northern latitudes or do
they exist worldwide? Here, we try to better understand the magnitude and extent of these phenomena
by testing several sites in northwest China.
In northwest China, tree rings (width, isotopes,
density) have been widely used in dendroclimatological studies at the alpine tree line (e.g., Yuan et al.
2003; Zhang and Wu 1997; Shao et al. 2004, 2005;
Gao et al. 2005). Several climate reconstructions
have been conducted in the Qilian Mts. (Zhang and
Wu 1996; Wang et al. 2001; Gou et al. 2001) and the
Helan Mts. (Liu et al. 2004, 2005), some extending
back over 1,000 years (Kang et al. 2002). However,
no study has yet considered testing for growth
divergence or the stability of the relationship between
tree growth and climate over time in this region. A
better understanding of the tree growth responses to
climate in northwest China during the last century,
however, is important not only for regional paleoclimatic studies but also for forest carbon uptake
simulations and future forest planning.
Materials and method
Study area
For this study, we sampled three regionally dominating tree species (two sites each) in the Qilian Mts. and
Helan Mts.: (1) Pinus tabulaeformis, (2) Sabina
przewalskii, and (3) Picea crassifolia. The Qilian
Mts. and Helan Mts. are two prominent mountains in
Forest Ecology
41
Fig. 1 Map of the
sampling sites (m) in
eastern Qilian Mts. and
Helan Mts., as well as of the
nearby grid data (d) from
CRUTS2.1 (37750 N,
101750 E; 36750 N,
102750 E; 38750 N,
105750 E). The gray dots
indicate the grid cells
around the sampling sites
that showed similar results
Northwest China (Fig. 1). Each of them has its own
typical topography and typical atmospheric systems.
The Qilian Mts., located on the northern edge of the
Tibet Plateau, have several peaks over 4,000 m,
which create a strong rain shadow effect for monsoons coming from the southeast. Our study area is
situated in the eastern part of the Qilian Mts., a
transitional area between temperate monsoons and
continental climate. There, Picea and Sabina are two
typical and widespread conifer species. The Helan
Mts. are located in north central China, where the arid
northwest areas meet the Loess Plateau. They extend
over 200 km from south to north, but only 15–60 km
from east to west with peak elevations between 2,000
and 3,000 m a.s.l. Located along the northwest
margin of the East Asian Summer Monsoon, the
Helan Mts. act as a barrier to the penetration of
monsoon rainfall into northwest China. Pinus and
Picea are two typical and widespread species in
Helan Mts.
Jones 2005) instead of the measured data for our
analysis, fully aware that relationships between tree
ring parameters and CRU data are usually weaker
than with nearby station data, mostly because of the
large scale smoothing applied in the CRU datasets
(Mitchell and Jones 2005). In order to evaluate the
quality of gridded data over time, a 10-year moving
standard deviation (SD) was employed. The moving
SD of precipitation data has an abrupt change around
1934, indicating a possible problem with the precipitation data prior to 1934. The moving SD of
temperature is stable over the whole time period.
Therefore, in this study we used only gridded data
from 1934 to 2000 (1934 to 1999 for sites of Helan
Mts.).
The climate data of the nearest CRU grid cells
were used in this study for the calculations of
climate–growth relationship, since the relationships
between chronologies and climate data of the four
grid cells around the sample sites were similar (data
not shown).
Climate data
Tree ring sampling and cross dating
Most of the meteorological stations in Northwest
China were set up after 1950 and thus provide only a
short climatic record. As a consequence, we used
gridded data from the high-resolution 0.5 9 0.5
gridded climate dataset CRUTS2.1 (Mitchell and
Increment cores of trees in the Qilian Mts. were
collected in October 2000. A total of 29 cores were
taken from 16 living Picea trees growing between
2,600 and 2,900 m a.s.l. near the lower tree line and
42
A.G. Van der Valk (ed.)
were herein named S1. All sampled trees were
healthy and growing on an east-facing slope with
moist soil (if compared to the other sampling sites).
From 23/22 Sabina trees 41/39 cores were taken at
two sites taken from the east Qilian Mts. and named
J1 and J2. All cores were taken from healthy trees
growing at the upper limit of the forest at an elevation
of about 2,930–3,100 m a.s.l. with thin gray cinnamonic soil.
The increment cores of Helan Mts. were collected
in October 1999. A total of 35 cores were taken from
18 living Picea trees growing above 2,500 m a.s.l. at
west slope of Helan Mts. and were named S2. From
40/25 Pinus trees growing 68/41 cores were taken
above 2,000–2,300 m a.s.l. at two sites from Helan
Mts. and named P1 and P2. P1 and P2 are at the east
and west slopes of the Helan Mts., respectively. The
soil at both sites was thin and rocky. The dominant
tree species in both forest sites was Pinus, typically
found growing at an elevation between 1,900 and
2,350 m a.s.l. (Table 1).
Tree rings were processed and cross-dated with
standard dendrochronological techniques (Cook and
Kairiukstis 1990). Ring width was measured on a
Velmex system with a precision of 0.001 mm. The
program COFECHA (Holmes 1983) was employed to
check the quality of visual cross-dating.
Tree ring data processing
Divergent growth trends over time and chronology
development
First, we used the raw data to calculate the growth
trend of each tree for the last 40 years using linear
regression. All series at each site were classified into
two groups: one with increasing growth trend (slope,
b [ 0) and the other with decreasing growth trend
(b \ 0) (Pisaric et al. 2007). All series were then
standardized with the program ARSTAN (Cook
1985) using conservative negative exponential or
linear regression. Due to the failure of conservative
detrending in a few series (Table 2), they were
standardized with the Hugershoff growth curve.
Chronologies based on the groups were built using
traditional methods. Since we used raw data to judge
the growth trend, the two chronologies (one with
increasing and one with decreasing growth trend)
were combined into one chronology, if the two
groups showed a similar trend after detrending.
Table 1 Site information
Site
Species
Latitude (N)
Longitude (E)
Elevation (m)
Slope
S1
Picea crassifolia
37.87
101.53
2600–2900
East-facing
38.63
105.78
2500
North-facing
39.08
106.08
2600
North west-facing
38.72
105.98
2400
North west-facing
37.93
101.53
2930–3100
South-facing
36.59
102.31
*3100
South-facing
S2
P1
Pinus tabulaeformis
P2
J1
Sabina przewalskii
J2
Table 2 Statistics of the eight chronologies. Hug., series detrended by Hugershoff growth curve
Site
Sub-chro.
Sample size (cores/trees)
Hug.
Time interval
MS
AC
Rbt
EPS
EPS [ 0.85
S1
S1
29/16
2
1840–2000
0.430
-0.124
0.601
0.945
1900
S2
S2
35/18
2
1869–1999
0.351
-0.006
0.496
0.811
1920
P1
P1I
8/6
0
1742–1999
0.648
-0.004
0.583
0.854
1850
P2
P1D
P2
60/38
41/25
6
7
1700–1999
1739–1999
0.566
0.696
-0.067
-0.013
0.615
0.694
0.988
0.983
1820
1819
J1
J1I
18/13
0
1288–2000
0.351
-0.007
0.256
0.857
1865
J1D
22/14
0
1590–2000
0.304
-0.034
0.248
0.868
1875
J2
J2
39/22
4
1740–2000
0.422
-0.043
0.341
0.952
1852
MS, Mean sensitivity; AC, first-order autocorrelation; Rbt, the mean interseries correlation; EPS, the expressed population signal and
the year from when EPS is consistently greater than 0.85
Forest Ecology
43
Several descriptive statistics, commonly adopted in
dendrochronology, were used to compare chronologies. These statistics include the mean sensitivity
(MS) and SD (to assess the high-frequency variations), the first-order serial autocorrelation (AC) (to
detect eventual persistence retained after the standardization), the mean correlation between trees
(Rbt), and the expressed population signal (EPS) (to
estimate the amount of year-to-year growth variations
shared among trees of the same chronology). In order
to visualize the apparent growth divergence over
time, we subtracted the detrended tree ring width
indices of the increasing chronology from the
decreasing chronology where applicable.
Biondi and Waikul 2004). A moving CF employs a
fixed number of years progressively slid across time to
compute the correlation coefficients (Biondi 1997).
Considering the length of recorded data and the
reliability of sample size, we chose 48 years as the
moving interval for each calculating analysis. Moving
CFs produce a temporal set of coefficients for each
predictor and coefficients not significant at the 95%
confidence level are changed to zero. Here, we just
present the results of the moving correlations in detail
because the moving CFs not only include the results
obtained from simple correlation analysis, but also
provide a dynamic perspective on the evolution of tree
responses to climate over time.
Climate–growth relationships and their stability over
time
Results
After dividing all series into groups of increasing and
decreasing growth trends and chronology building, we
tested each resulting chronology for its climate–growth
relationship. First, simple correlation functions (CF)
were employed. We then tested the stationary and
consistency of these climate–tree growth relationships
over time using moving CF (DENDROCLIM2002,
(a)
50
25
0
0
2
(b)
50
25
0
0
4
(c)
2
Ring width index
After calculating growth trends for all trees, we
found that sites J1 and P1 each contained one group
with increasing and one with decreasing growth
trends. Hence, they were separated for chronology
building (Fig. 2) and subsequently termed J1D and
60
30
0
2
0
4
50
25
0
(d)
2
50
0
25
(e)
2
0
50
0
2
(f)
25
0
0
1800
1850
1900
1950
2000
Sample depth
Fig. 2 The standard ringwidth chronologies, 48-year
smoothing (thicker line),
and their corresponding
sample depth, (a) S1; (b)
S2; (c) P1, dashed line P1I
and continuous line P1D;
(d) P2; (e) J1, dashed line
J1I and continuous line
J1D; (f) J2
Growth divergence and chronologies
44
P1D (for decreasing) and J1I and P1I (for increasing). Trees from the other four sites did not divide
into groups and thus showed no growth divergence
and were subsequently combined into one chronology per site, resulting in a total of eight
chronologies from the six sites. Time spans of the
chronologies were different and we chose a fixed
common period 1900–2000 to compare the quality
of different chronologies. MS and first-order serial
AC varied from 0.304 to 0.696 and from -0.004 to
-0.124, respectively. All sites (except S1) exhibited
low serial AC in their mean chronologies (in
Table 2), which was mostly removed after autoregressive modeling of single series. Two useful
parameters for evaluating the quality of a chronology are the mean interseries correlation (Rbt, varied
from 0.248 to 0.694) and the EPS (varied from
0.811 to 0.988). The EPS values of the chronologies
are greater than 0.85 except for S2, which has a
0.983 EPS value during 1920 and 1999, but low
value during 1900 and 1920, since most trees of this
site were younger than 100 years. The two Sabina
sites (J1 and J2) had lower Rbt and EPS values, but
all EPS values were above the accepted cut-off of
0.85 (Wigley et al. 1984).
Smoothing the chronologies with a 48-year spline
showed that the two sub-chronologies from a site
(J1D and J1I; P1D and P1I) had a very similar
short-term variation but different long-term trends
(Fig. 2).
Influence of climate and the stability of the tree
ring growth–climate relationships
Regional influences of climate on tree growth
Most chronologies show a strong, consistent, and
positive relationship with precipitation during the
current growth season (Fig. 3). June precipitation
was the key variable for tree growth in both
mountain ranges. The strength of the positive
correlations, however, varied over time except for
J1D and J2.
Climate–growth relationships of each species
First noticeable in the Pinus chronologies is the high
climate sensitivity of P1I chronology, which shows
significant correlation with several climate
A.G. Van der Valk (ed.)
parameters. P1D and P2 have quiet similar relationships but less significant than P1I. The Pinus trees are
generally limited by growth season (especially June)
precipitation, but this relationship weakened and is
not significant in recent years. September temperature
of the growing season and October precipitation of
the previous year had influence on all Pinus chronologies with a positive relationship, but the
relationships are not stable during the calculated
time period. During the growing season, temperature
seems to play a limited role for tree growth with
negative relationships, for example, in August and
March. These relationships are not stable over time.
In P1I, two stable positive relationships between tree
growth and prior October and December started from
early 1980 and became stronger and stable in recent
time. Also, in P1I, there is a noticeable phenomenon
that the precipitation seems to affect growth of P1I
moving forward from July to May over time.
The Sabina chronologies are not consistent with
each other. At the J1 site, both J1D and J1I have a
strong negative correlation with June temperature,
but while this relationship in both chronologies
weakened over time, it recently dropped to nonsignificant in J1I. This weakening of the negative
correlation with June temperature in J1I is concurrent
with an emergent positive correlation with June
precipitation. The positive correlations with prior
November and December temperature and negative
correlation with the precipitation of August and
September in J1I also became stable and significant
during the second half of the record. In J1D, the
negative relationships with temperature of January
and February gradually became stronger. At the J2
site, May temperature and September temperature
affected tree growth with negative and positive
relationships, previous December temperature started
to affect tree growth with a significant positive
relationship, which became stronger in the late
twentieth century.
In S1, the relationships between tree growth and
current September temperature and June precipitation
are positive, but the relationships dropped to nonsignificant at the end of the record. Instead, a negative
relationship between the S1 chronology and several
monthly temperatures became significant at the end
of the record. S2 has mainly stationary significant
positive correlations with prior October and current
June precipitation over time.
Forest Ecology
45
Fig. 3 Moving correlations between climate variables
(monthly mean temperature (T) and monthly total precipitation
(P)) and tree ring chronologies (a) S1; (b) S2; (c) P1I; (d) P1D;
(e) P2; (f) J1I; (g) J1D; (h) J2. Previous year October—current
year September were used in the analyses
Seasonalized pattern
relationships with precipitation. The response pattern
to precipitation was generally opposite to the correlations with temperature during growing season (May,
June, July) and September.
For different seasons, trees from different sites (except
J1D and S2) display similar relationships with temperature and precipitation (Table 3). In previous year’s
autumn and early winter (from October to December),
there is a positive relationship between tree growth and
both temperature and precipitation. During winter,
early spring, and summer, trees have similar negative
relationships with temperature but some positive
Discussion
Recent studies show several problems with the development of tree growth–climate relationships, e.g.,
46
A.G. Van der Valk (ed.)
Fig. 3 continued
growth divergence of subpopulations (Wilmking et al.
2004, 2008), the ‘‘divergence problem’’ or underestimation of current temperatures by tree ring-based
climate reconstructions (D’Arrigo et al. 2008) and
changing relationships between tree growth and
temperature variability (e.g., Carrer and Urbinati
2006), most of them occurring circumpolar at high
northern latitudes. Here, we present evidence of
growth divergence and instability of tree growth–
climate relationships in three conifer tree species
in northwest China. Two species (Sabina and Pinus)
at two different sites show subpopulation behavior
with growth divergence but the other sites do not.
The growth divergences at these sites appeared not
only in late twentieth century but also about 1920
(Fig. 4) in both species at similar times but with
differing amplitude. The largest divergences occurred
in Sabina in the late twentieth century but in Pinus
during the 1920s. The growth divergence here is
different from the growth divergences shown in other
studies (Wilmking et al. 2005; Pisaric et al. 2007),
which only reported growth divergence in the late
twentieth century. Unfortunately, our sample size of
different sites is inadequate to comprehensively
address the question, if the growth divergence
observed in this study is the result of a specific
combination of site factors, such as elevation, slope,
and exposure.
Forest Ecology
47
Table 3 The signs of significant moving correlation coefficients between chronologies and monthly climate factors (mean temperature T and total precipitation P)
T
P-Oct
P
P1I
P1D
+
+
P2
S2
S1
J2
P1I
P1D
P2
S2
+
+
+
+
S1
J1I
J1D
J2
+
+
+
+
+
+
Jan
Feb
Mar
J1D
-
P-Nov
P-Dec
J1I
-
-
+
+
-
-
-
-
+
+
-
+
+
+
-
+
+
Apr
May
Jun
-
-
-
-
-
-
-
Jul
Aug
Sep
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
+
+
+
+
-
-
+
-
All correlation coefficients were calculated based on a 48-year time interval and previous year October (P-Oct)—current year
September were used in the analysis
Fig. 4 An index of growth
divergence in P1 (a) and J1
(b) shows similar trends but
different amplitudes over
the period 1934–1999. The
thick lines are 11-year
smoothing lines
1.2
(a)
0.6
0.0
-0.6
1.2
(b)
0.6
0.0
-0.6
1880
1900
1920
1940
1960
1980
2000
Year
Although the growth responses to climate revealed
by most chronologies support the common fact that
growing season (especially June) precipitation is the
main limiting factor for tree growth, there are still other
phenomena: (1) Positive correlations between tree
growth and previous year autumn temperature suggest
that the warm October condition likely support trees to
keep carbohydrate storage and perhaps increased
foliage or wood production in the subsequent growing
season (Julian 2000; Schaberg et al. 2000). Previous
48
studies indicate that temperate conifers have a positive
carbon gain in warm winter days when their leaves are
not frozen (Chabot and Hicks 1982; Havranek and
Tranquillini 1995). (2) Negative relationship with
temperature and positive relationship with precipitation during winter and early spring (from January to
March) might indicate the protection of snow cover at
high elevation. At the alpine timberline, Oberhuber
(2004) found that trees show a tendency to suffer from
enhanced desiccation during winter and early spring
periods with insufficient snow depth because of
increased transpiration rates of needles and shoots,
photo inhibitory stress, and short-term fluctuations in
shoot temperatures. (3) The warm autumn of the
current year also has a strong effect on tree growth in
most species. Trees seem to grow better during the
warmer autumn. According to Shi et al. (2008),
temperature could play an important role on tree ring
formation at the end of the growing season in arid and
semi-arid areas by prolonging the growing season.
The instable relationships over time between tree
growth and climate factors might have been caused
by different combination of climate factors (e.g.,
temperature and precipitation). Recent studies indicate that there is an ongoing warming and drying
trend for all seasons in north central China (Wang
and Zhou 2005; Zhai et al. 2005; Ma and Fu 2006).
Individual trees (especially Pinus and Sabina) might
have become more sensitive to micro-site differences,
resulting in the breakdown of the uniform growth
behavior at the sites and subsequent differing growth
trends and climate sensitivity. One example is that the
chronologies with increasing trend show more stable
positive correlation with June precipitation than the
chronologies with decreasing growth trend. The high
correlations with prior winter and autumn of the
current year appeared in both increasing chronologies
during the calculated time period, indicating that
those trees could take advantage of the available
conditions better than trees of the decreasing chronology. However, the decreasing trees started to
become more sensitive to the temperature during the
late spring and early summer, which could be induced
by more desiccation through increasing evaporation
before the arrival of the summer monsoon (Ding
1994). Furthermore, the positive correlation with
precipitation, gradually moving forward from July to
May in the P1I chronology, might also indicate the
drying trend in the early growth season.
A.G. Van der Valk (ed.)
Conclusion
Recent studies discussed three major challenges to
the field of dendroclimatology, (1) changing relationships between tree growth with climate over time, (2)
emerging sub-chronology behavior at sites formerly
considered suitable to build one chronology, and as a
possible result, (3) the ‘‘divergence problem’’ or
underestimation of recent warming trends by tree
ring-based climate reconstructions. Many of these
studies were conducted either in the boreal zone or at
altitudinal tree limit in Europe. Here, we present
evidence of the first two phenomena in mid-latitude
NW China. Diverging growth trends were found in
Pinus and Sabina sites but not in Picea sites. The
correlations between tree growth with climate factors
at most sites are instable over time, as also indicated
by switches from significant to non-significant or vice
versa relationships with climate factors at different
periods of the record. Non-divergent sites have more
stationary relationships with climate factors than
chronologies from divergent sites. There, decreasing
chronologies show more stable relationships with
climate than increasing chronologies, which are more
sensitive to climate factors, indicated by higher
correlation scores.
For the future, we see a major need for additional
work at two fronts: (1) to test more regions and
species for the phenomena of changing climate–
growth relationships over time and (2) to better
understand the mechanisms affecting growth from the
level of the individual tree to the population through
time. Only then can we safely proceed to use tree
rings as robust source of paleoclimatic information.
Acknowledgment We thank Dr. Jayendra Singh and Jinbao
Li for discussing and help.
References
Barber V, Juday G, Finney B (2000) Reduced growth of Alaska
white spruce in the twentieth century from temperatureinduced drought stress. Nature 405:668–672. doi:10.1038/
35015049
Biondi F (1997) Evolutionary and moving response functions
in dendroclimatology. Dendrochronologia 15:139–150
Biondi F, Waikul K (2004) DENDROCLIM2002: a C++
program for statistical calibration of climate signals in
tree-ring chronologies. Comput Geosci 30:303–311. doi:
10.1016/j.cageo.2003.11.004
Forest Ecology
Briffa K, Schweingruber F, Jones P, Osborn T (1998a) Reduced
sensitivity of recent tree growth to temperature at high
northern latitudes. Nature 391:678–682. doi:10.1038/35596
Briffa K, Schweingruber F, Jones P, Osborn T, Harris I, Shiyatov
S et al (1998b) Trees tell of past change climates: but are
they speaking less clearly today? Philos Trans R Soc Lond
B Biol Sci 353:65–73. doi:10.1098/rstb.1998.0191
Carrer M, Urbinati C (2006) Long-term change in the sensitivity of tree-ring growth to climate forcing in Larix
decidua. New Phytol 170:861–872. doi:10.1111/j.14698137.2006.01703.x
Chabot BF, Hicks DJ (1982) The ecology of leaf life spans.
Annu Rev Ecol Syst 13:229–259. doi:10.1146/annurev.
es.13.110182.001305
Cook ER (1985) A time series approach to tree-ring standardization. PhD thesis, University of Arizona
Cook ER, Kairiukstis LA (1990) Methods of dendrochronology. Kluwer, Dordrecht
Cook ER, Peters K (1997) Calculating unbiased tree-ring indices
for the study of climatic and environmental change. Holocene 7:361–370. doi:10.1177/095968369700700314
D’Arrigo R, Kaufmann R, Davi N, Jacoby G, Laskowski C,
Myneni R et al (2004) Thresholds for warming-induced
growth decline at elevational treeline in the Yukon Territory. Global Biogeochem Cycles 18:GB3021. doi:
10.1029/2004GBO02249
D’Arrigo R, Wilson R, Liepert B, Cherubini P (2008) On the
‘‘divergence problem’’ in northern forests: a review of
tree-ring evidence and possible causes. Global Planet
Change 60:289–305. doi:10.1016/j.gloplacha.2007.03.004
Ding YH (1994) Monsoon over China. Kluwer Academic Publisher
Driscoll W, Wiles G, D’Arrigo R, Wilmking M (2005) Divergent
tree growth response to recent climatic warming, Lake
Clark National Park and Preserve, Alaska. Geophys Res
Lett 32:L20703. doi:10.1029/2005GL024258
Fritts HC (1976) Tree ring and climate. Academic Press, London
Gao SY, Lu RJ, Qiang MR (2005) Reconstruction of precipitation in the last 140 years from tree ring at south margin
of the Tengger Desert, China. Chin Sci Bull 50(21):2487–
2492. doi:10.1360/982005-363
Gou XH, Chen FH, Wang YJ, Shao XM (2001) Spring precipitation reconstructed in the east of the Qilian Mountain
during the last 280 a by tree ring width. J Glaciol Geocryol 23(3):292–296 (in Chinese)
Greenough G, McGeehin M, Bernard SM (2001) The potential
impacts of climate variability and change on health
impacts of extreme weather events in the United States.
Environ Health Perspect 109(Suppl. 2):191–198
Havranek M, Tranquillini W (1995) Physiological processes
during their winter dormancy and their ecological significance. In: Smith WK, Hinkley TM (eds) Ecophysiology of
coniferous forest. Academic Press, New York, pp 95–124
Holmes RL (1983) Computer-assisted quality control in treering dating and measurement. Tree Ring Bull 43:69–78
Hoyt D (2006) Urban heat islands and land use changes.
http://www.warwickhughes.com/hoyt/uhi.htm
Hughes MK (2002) Dendrochronology in climatology—the state
of the art. Dendrochronologia 20:95–116. doi:10.1078/
1125-7865-00011
IPCC Fourth AR (2007) Summary for policy makers In: Climate
change 2007 Cambridge University Press, Cambridge
49
Jacoby GC, D’Arrigo R (1995) Tree-ring width and density
evidence of climatic and potential forest change in Alaska.
Global Biogeochem Cycles 9:227–234. doi:10.1029/
95GB00321
Jacoby GC, D’Arrigo R, Davaajamts TS (1996) Mongolian tree
rings and twentieth century warming. Science 273:771–
773. doi:10.1126/science.273.5276.771
Jacoby GC, Lovelius N, Shumilov O, Raspopov O, Kurbainov
J, Frank D (2000) Long-term temperature trends and tree
growth in the Taymir region of northern Siberia. Quat Res
53:312–318. doi:10.1006/qres.2000.2130
Julian LH (2000) Effect of daily minimum temperature on
photosynthesis in eastern Hemlock (Tsuga canadensis L.)
in Autumn and Winter. Arct Antarct Alp Res 32(4):368–
374. doi:10.2307/1552384
Kahn ME (2005) The death toll from natural disasters: the role
of income, geography, and institutions. Rev Econ Stat
87(2):271–284. doi:10.1162/0034653053970339
Kang XC, Chen GD, Kang ES, Zhang QH (2002) Based tree
rings data reconstruction over 1000-year streamflow of
mountain pass in Heihe River. Sci China Ser Dokl Earth
Sci 32(8):675–685
Knapp PA, Soule PT, Grissino MHD (2001) Detecting potential regional effects of increased atmospheric CO2 on
growth rates of western Juniper. Glob Change Biol 7:903–
917. doi:10.1046/j.1365-2486.2001.00452.x
Liu Y, Shi JF, Shishov V (2004) Reconstruction of May–July
precipitation in the north Helan Mountain, Inner Mongolia
since AD 1726 from tree-ring late-wood widths. Chin Sci
Bull 49(4):405–409. doi:10.1360/03WD0409
Liu Y, Cai QF, Shi JF (2005) Seasonal precipitation in the
south-central Helan Mountain region, China, reconstructed from tree-ring width for the past 224 years. Can J
For Res 35(10):2403–2412. doi:10.1139/x05-168
Lloyd A, Fastie C (2002) Spatial and temporal variability in the
growth and climate response of treeline trees in Alaska.
Clim Change 52:481–509. doi:10.1023/A:1014278819094
Ma ZG, Fu CB (2006) Some evidence of drying trend over
northern China from 1951 to 2004. Chin Sci Bull
51:2913–2925. doi:10.1007/s11434-006-2159-0
Melvin T (2004) Historical growth rates and changing climatic
sensitivity of boreal conifers. Ph.D. thesis, Climatic
Research Unit, East Anglia
Mitchell T, Jones PD (2005) An improved method of constructing a database of monthly climate observations and
associated high-resolution grids. Int J Climatol 25:693–
712. doi:10.1002/joc.1181
Oberhuber W (2004) Influence of climate on radial growth of
Pinus cembra within the alpine timberline ecotone. Tree
Physiol 24:291–301
Pisaric M, Carey S, Kokelj S, Youngblut D (2007) Anomalous
20th century tree growth, MackenzieDelta, Northwest
Territories, Canada. Geophys Res Lett 34:L05714. doi:
10.1029/2006GL029139
Schaberg IG, Snyder MC, Shane JB, Donnelly JR (2000)
Seasonal patterns of carbohydrate reserves in red spruce
seedlings. Tree Physiol 20:549–555
Sergio R, Annie D, Tommaso A, Vinicio C (2007) Evidence of
threshold temperatures for xylogenesis in conifers at high
altitudes. Oecologia 152:1–12. doi:10.1007/s00442-0060625-7
50
Shao XM, Huang L, Liu HB (2004) Reconstruction of precipitation variation from tree rings in recent 1000 years in
Delingha, Qinghai. Sci China Ser Dokl Earth Sci 34(2):
145–153
Shao XM, Li EY, Huang L, Wang LL (2005) A 1437-year precipitation history from Qilian Juniper in the Northeastern
Qinghai-Tibetan Plateau. PAGES News 13(2):14–15
Shepherd A, Wingham D (2007) Recent sea-level contributions
of the Antarctic and Greenland ice sheets. Science
315(5818):1529–1532. doi:10.1126/science.1136776
Shi J, Liu Y, Vaganov EA, Li J, Cai Q (2008) Statistical and
process-based modeling analyses of tree growth response
to climate in semi-arid area of north central China: a case
study of Pinus tabulaeformis. J Geophys Res 113:G01026.
doi:10.1029/2007JG000547
Smith KT, Cufar K, Levanic T (1999) Temporal stability and
dendroclimatology in silver fir and red spruce. Phyton
Ann Rei Bot 39:117–122
Solberg BO, Hofgaard A, Hytteborn H (2002) Shifts in radial
growth responses of coastal Picea abies induced by climatic change during the 20th century, central Norway.
Ecoscience 9:79–88
Vaganov EA, Hughes MK, Kirdyanov AV, Schweingruber FH,
Silkin PP (1999) Influence of snowfall and melt timing on
tree growth in subarctic Eurasia. Nature 400:149–151.
doi:10.1038/22087
Wang Y, Zhou L (2005) Observed trends in extreme precipitation events in China during 1961–2001 and the
associated changes in large-scale circulation. Geophys
Res Lett 32:L09707. doi:10.1029/2005GL022574
Wang YJ, Chen FH, Gou XH, Du SY (2001) Study on response
relationship between tree-ring and climate factors and
climate reconstruction in middle region of Qilianshan
Mountains. J Desert Res 21(2):135–140 (in Chinese)
Webster PJ, Holland GJ, Curry JA, Chang HR (2005) Changes
in tropical cyclone number, duration, and intensity in a
warming environment. Science 309(5742):1844–1846.
doi:10.1126/science.1116448
Wigley TL, Briffa KR, Jones PD (1984) On the average value of
correlated time series, with applications in dendroclimatology and hydrometeorology. J Clim Appl Meteorol
23:201–213. doi :10.1175/1520-0450(1984)023\0201:OT
AVOC[2.0.CO;2
A.G. Van der Valk (ed.)
Wilmking M, Myers-Smith I (2008) Changing climate sensitivity of black spruce in a peatland-forest landscape in
interior Alaska. Dendrochronologia. doi:10.1016/j.dendro.
2007.04.003
Wilmking M, Juday GP, Barber VA, Zald HSJ (2004) Recent
climate warming forces contrasting growth responses of
white spruce at treeline in Alaska through temperature
thresholds. Global Change Biol 10:1724–1736. doi:
10.1111/j.1365-2486.2004.00826.x
Wilmking M, D’Arrigo R, Jacoby G, Juday G (2005) Divergent
growth responses in circumpolar boreal forests. Geophys
Res Lett 32:L15715. doi:10.1029/2005GLO23331
Wilson R, Elling W (2004) Temporal instability in tree-growth/
climate response in the Lower Bavarian Forest region:
implications for dendroclimatic reconstruction. Trees
Struct Funct 18:19–28. doi:10.1007/s00468-003-0273-z
Wilson R, Luckman BH (2002) Tree-ring reconstruction of
maximum and minimum temperatures and the diurnal
temperature range in British Columbia, Canada. Dendrochronologia 20(3):257–268. doi:10.1078/1125-7865-00023
Wilson R, Luckman BH (2003) Dendroclimatic reconstruction
of maximum summer temperatures from upper tree-line
sites in interior British Columbia. Holocene 13(6):853–
863. doi:10.1191/0959683603hl663rp
Yonenobu H, Eckstein D (2006) Reconstruction of early spring
temperature for central Japan from the tree-ring widths of
Hinoki cypress and its verification by other proxy records.
Geophys Res Lett 33:L10701. doi:10.1029/2006GL02
6170
Yuan YJ, Jin LY, Shao XM (2003) Variations of the spring
precipitation day numbers reconstructed from tree rings in
the Urumqi River drainage, Tianshan Mts. over the last
370 years. Chin Sci Bull 48(14):1507–1510. doi:10.1360/
02wd0251
Zhai P, Zhang X, Wan H, Pan X (2005) Trends in total precipitation and frequency of daily precipitation extremes over
China. J Clim 18:1096–1108. doi:10.1175/JCLI-3318.1
Zhang ZH, Wu XD (1996) The change and rapid shifts of
moisture index and its annual variability since 1310 A.D.
in Qilianshan area. Quat Sci 4:368–378 (in Chinese)
Zhang ZH, Wu XD (1997) Reconstruction of climate variation
in the last 700 years in the Qi Lian Mountain area using
tree-ring data. Chin Sci Bull 42(8):849–851 (in Chinese)
Does leaf-level nutrient-use efficiency explain
Nothofagus-dominance of some tropical rain forests
in New Caledonia?
Alex Chatain Æ Jennifer Read Æ Tanguy Jaffré
Originally published in the journal Plant Ecology, Volume 201, No. 1, 51–66.
DOI: 10.1007/s11258-008-9477-z Springer Science+Business Media B.V. 2008
Abstract Tropical rain forests generally have a
complex structure and a high diversity of species in
their canopy, but in some rain forests the upper
canopy is dominated by a single species. The factors
controlling this dominance are uncertain. In New
Caledonia, Nothofagus species dominate the upper
canopy of some rain forests on ultramafic soils. Here
we investigate whether leaf-level nutrient-use efficiency (NUE) could explain dominance by
Nothofagus. We found no evidence of a competitive
advantage in Nothofagus in terms of leaf-level NUE:
Nothofagus species did not have lower leaf macronutrient concentrations, nor did they resorb more
nutrients than co-occurring species on average. They
did, however, have lower foliar Ni concentrations on
average. Leaf decay rate across all species in a
glasshouse-based trial correlated positively with
foliar P and negatively with cell wall content,
lignin:P, C:P, lignin:N, leaf toughness and tannin
activity. Multivariate analysis suggested that total cell
wall concentration exerted the strongest independent
A. Chatain J. Read (&)
School of Biological Sciences, Monash University,
Melbourne, VIC 3800, Australia
e-mail: jenny.read@sci.monash.edu.au
T. Jaffré
IRD – Laboratoire de Botanique et d’Écologie Végétale
Appliquée, Institut de recherche pour le développement,
Centre de Nouméa, BP A5, Nouméa 98848,
New Caledonia
effect on variation among species in decomposition
rate. Slow decomposition of Nothofagus leaf litter
may facilitate continued dominance of the upper
canopy by suppressing establishment and growth of
co-occurring species or by promoting disturbance
through fire, since disturbance has been suggested as
necessary for regeneration and maintenance of dominance by Nothofagus species. However, the
biological mechanisms allowing Nothofagus to
achieve initial dominance of these post-disturbance
forests are uncertain, and may still include plant-level
NUE.
Keywords Decomposition Litter
Monodominance Resorption
Introduction
Tropical rain forests most often have a complex
structure and diverse canopy composition. However,
some have 50–100% of their upper canopy dominated
by a single species, referred to as monodominance
(Connell and Lowman 1989). These monodominant
forests are not rare, occurring in all continents that
support tropical rain forests (Hart et al. 1989), but are
more common in some regions than others (Torti
et al. 2001). The mechanisms promoting monodominance are still uncertain and may vary among forests.
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_5
51
52
For example, monodominant forests are common in
suboptimal environments, such as where poor substrate quality or severe climate prevail (Connell and
Lowman 1989; Hart et al. 1989), but can also occur
adjacent to mixed-canopy forest where there is no
obvious spatial determinant (Read et al. 2006).
Monodominant forests can be either early or late
successional stages (Connell and Lowman 1989; Hart
et al. 1989; Nascimento et al. 2007). Torti et al.
(2001), investigating the cause of monodominance of
Gilbertiodendron dewevrei in Congolese rain forests,
suggested that it was the result of traits that modified
the understorey, making it difficult for other species
to co-exist. These suggested traits included adult
plants casting a deep shade and litter layer, making it
difficult for seedlings of other species to establish,
with a leaf litter slow to decompose, potentially
reducing nitrogen availability.
In New Caledonia, Nothofagus species dominate
the upper canopy of some tropical rain forests on
ultramafic soils. These stands are often contiguous
with diverse mixed-canopy forests, also on ultramafic
soils (Read et al. 2000, 2006). Comparison of
Nothofagus forests and adjacent mixed rain forest
showed little evidence of soil-mediated boundaries
(Read et al. 2006) and the factors controlling the local
boundaries of these Nothofagus-dominated forests are
uncertain. It is likely that numerous factors contribute, but in apparent contrast to the Gilbertiodendron
forests, preliminary data suggest that the Nothofagusdominated stands, at least in the lowlands, are an
early successional stage following disturbance such
as fire or cyclone, potentially replaced by mixed rain
forest in the absence of further disturbance (Read and
Hope 1996; Read et al. 2006). Here, we focus on
traits that might allow Nothofagus to dominate the
upper canopy of these early (first generation) postdisturbance forests.
One of the key traits facilitating monodominance
in this case may be a high nutrient-use efficiency
(NUE). The ultramafic soils on which the Nothofagus
forests grow in New Caledonia have high contents of
heavy metals such as Ni, which can be toxic to plants,
and low concentrations of macronutrients, particularly Ca, P and K (Jaffré 1992; Read et al. 2006).
Plants have evolved characteristics to cope with low
levels of soil nutrients, including efficient acquisition
of nutrients, internal nutrient economy via redistribution and/or low nutrient requirements (Clarkson
A.G. Van der Valk (ed.)
and Hanson 1980; Aerts and Chapin 2000). Plants
that are efficient with respect to soil nutrients produce
more growth per absorbed nutrients than inefficient
ones, especially when nutrient supplies are low
(Clarkson and Hanson 1980). Hence, NUE may be
the key trait allowing Nothofagus species to dominate
these forests on infertile soils.
This study aims to determine if Nothofagus species
differ from co-occurring species in some components
of leaf-level NUE. Leaf-level NUE, a component of
whole-plant NUE, operates at three levels, the initial
green leaf NUE (INUE), the leaf life span (LNUE)
and post- or senesced leaf NUE (PNUE). Foliar
nutrient concentration and resorption during senescence contribute to the efficiency with which
nutrients are used in nutrient-limited environments
(Vitousek 1984; Aerts and Chapin 2000). Resorption
can also affect neighbouring plants by influencing
rates of nutrient input to the system through decomposition, with lower litter nutrient concentrations
commonly resulting in slower decomposition (Swift
et al. 1979; Aerts and Chapin 2000; Hobbie and
Vitousek 2000). Hence, resorption traits may influence the capacity of Nothofagus to dominate the
canopy of early successional forests by increasing
whole-plant NUE and growth rate, and prolong the
period of dominance by decreasing access to nutrients
for competing species with higher nutrient demands.
We address the following specific questions: (1) Do
Nothofagus species have lower foliar nutrient concentrations and higher resorption efficiency and
proficiency than co-occurring species, suggesting a
lower nutrient demand and high efficiency of use at
the leaf level? (2) Do the leaves of Nothofagus
species decompose more slowly than those of cooccurring species, thereby potentially affecting nutrient cycling and availability to other species? (3) What
leaf traits contribute to variation in decomposition
among these species? In particular, we examine the
roles of litter nutrient content, cell wall, phenolics
and leaf toughness.
Methods
Site selection and leaf collection
Four study sites were selected in Nothofagus forests
in the southeast of the main island of New Caledonia
Forest Ecology
53
species, each from a different family (Table 1). Sites
could generally not be replicated at each location and
so trees form the replicates for within-site analyses.
At maturity, these species occupy various levels of
the forest canopy. Most of the selected species occupy
the middle to upper canopy, with trunk diameters at
breast height commonly exceeding 20 cm. The green
leaves sampled were the most recent fully expanded
and hardened leaves with an age \1 year. Senesced
leaves were selected if they were yellow or red (not
brown), with no evidence of decay. Green and
senesced leaves were collected from the same tree
(Table 1), across a range of soils, altitudes and
canopy compositions (described in Read et al. 2002,
2006). These evergreen forests experience an annual
rainfall of ca. 1,800–3,500 mm, with a short drier
season from September to November of variable
severity and duration. All forests were on ultramafic
soils, but those at Col de Mouirange were influenced
by gabbro intrusions, having higher concentrations of
Ca and slightly lower concentrations of Ni (Read
et al. 2006). Leaves were collected in October–
November 2005 from the dominant Nothofagus
species at each site and 4–7 co-occurring canopy
Table 1 Species collected
at each study site
Site and species
Measurements
Col de Mouirange Haut (CDMh): 22120 S, 166400 E, 320 m asl
Nothofagus aequilateralis (Baum.-Bodenh.) Steenis (Nothofagaceae)
g, s, d
Agathis lanceolata Lindley ex Warb. (Araucariaceae)
g, s
Arillastrum gummiferum (Brongn. & Gris) Pancher ex Baill. (Myrtaceae)
g, s
Codia discolor (Brongn. & Gris) Guillaumin (Cunoniaceae)
g, s, d
Deplanchea speciosa Vieill. (Bignoniaceae)
g, s, d
Hibbertia lucens Brongn. & Gris ex Sebert & Pancher (Dilleniaceae)
g, s
Col de Mouirange Bas (CDMb): 22120 S, 166410 E, 250 m asl
Nothofagus discoidea (Baum.-Bodenh.) Steenis
g, s, d
Acropogon dzumacensis (Guillaumin) Morat (Malvaceae)
g, s, d
Cerberiopsis candelabra var. candelabra Vieill. (Apocynaceae)
g, s, d
Crossostylis grandiflora Pancher ex Brongn. & Gris (Rhizophoraceae)
g, s, d
Diospyros parviflora (Schltr.) Bakh. f. (Ebenaceae)
g, s, d
Ficus austrocaledonica Bureau (Moraceae)
g, s, d
Storthocalyx chryseus Radlk. (Sapindaceae)
g, s
Dzumac: 2230 S, 166280 E, 940 m asl
Nothofagus codonandra (Baill.) Steenis
Sites have been described in
Read et al. (2000, 2006).
The forest chosen at Col de
Yaté contains Nothofagus
but is not monodominant.
Species nomenclature is
taken from Jaffré et al.
(2004). The ‘measurements’
column indicates the
species included in
analyses: g, green leaf
traits; s, senesced leaf traits;
d, decomposition study
g, s, d
Alphitonia neocaledonica (Schltr.) Guillaumin (Rhamnaceae)
g, s, d
Cryptocarya guillauminii Kosterm. (Lauraceae)
g, s, d
Flindersia fournieri Pancher & Sebert (Flindersiaceae)
g, s, d
Gastrolepis austrocaledonica (Baill.) Howard (Stemonuraceae)
g, s, d
Myodocarpus fraxinifolius Brongn. & Gris (Araliaceae)
g, s, d
Strasburgeria robusta (Vieill. ex Pancher & Sebert) Guillaumin
(Strasburgeriaceae)
g, s, d
Styphelia pancheri (Brongn. & Gris) F. Muell (Ericaceae)
g, s, d
0
0
Col de Yaté: 2210 S, 16654 E, 340 m asl
Nothofagus balansae (Baill.) Steenis
g, s, d
Calophyllum caledonicum Vieill. (Clusiaceae)
g, s, d
Elaeocarpus yateensis Guillaumin (Elaeocarpaceae)
g, s, d
Guettarda eximia Baill. (Rubiaceae)
g, s, d
Neoguillauminia cleopatra (Baill.) Croizat (Euphorbiaceae)
g, s, d
Planchonella kuebiniensis Aubrév. (Sapotaceae)
g
Semecarpus neocaledonica Engl. (Anacardiaceae)
g, s, d
54
where possible, from sun-lit branches on forest edges
or in large canopy gaps. Leaves of 4 of the 28 species,
however, were collected from the forest floor (both
green and senesced leaves) because of difficulty in
collecting leaves from branches; therefore, their
growth light environment is uncertain. However,
these were trees of the uppermost forest canopy, and
leaves were not likely to have been very heavily
shaded. Leaves were only collected when they
appeared fresh, as judged by colour and glossiness.
Three to five replicate trees or collecting points were
used for each species. Leaves used in chemical
analyses were initially air-dried, then freeze-dried and
ground to a powder in a ball mill.
Leaf traits
Macronutrient concentrations (N, P, K, Ca and Mg)
were measured in green and senesced leaves, allowing determination of nutrient contents of functional
leaves and of resorption. Resorption was measured as
resorption efficiency and proficiency. Resorption
efficiency (the percentage reduction of nutrients
between green and senesced leaves) can indicate the
degree to which nutrients are conserved in foliage,
encompassing both nutritional demand and nutrient
withdrawal by the plant, and resorption proficiency
(the nutrient concentration of senesced leaves) indicates the absolute levels to which nutrients can be
reduced in a plant, providing a measure of the degree
to which selection has acted to minimise nutrient loss
(Killingbeck 1996). Use of nutrient concentration per
mass to estimate resorption efficiency may lead to
errors due to changes in leaf mass caused particularly
by resorption of compounds during senescence, but
these errors are relatively small (Aerts 1996). Some
error may also arise from the assumption of an
identical constitution of senesced leaves when young
to the green leaves sampled. We assume here that any
such errors will be consistent across species. In
addition to macronutrients, we measured Ni since it is
potentially toxic and species vary in their capacity to
exclude or tolerate this metal in their leaves (Baker
1981). Nitric acid digestion (USEPA Method 3050B)
was used to extract macronutrients and metals (P, K,
Ca, Mg and Ni are reported) from freeze-dried
ground leaves prior to measurement by ICP-OES. C
and N were determined by a Leco CHN-2000 autoanalyser. All were expressed per unit leaf dry mass.
A.G. Van der Valk (ed.)
We measured other traits of senesced leaves
suggested to affect decomposition rates (phenolics,
cell wall content and toughness). ‘‘Total phenolics’’
were extracted in acetone following Cork and Krockenberger (1991), and assayed by the Prussian-Blue
method (Price and Butler 1977) as modified by
Graham (1992), with concentration expressed as
gallic acid equivalents (GAE) per leaf dry mass.
Tannins were extracted as for total phenolics, and
tannin activity was estimated by the mass of protein
precipitated using the Blue BSA (bovine serum
albumen) method of Asquith and Butler (1985), with
a bovine gamma globulin standard and was expressed
per unit leaf dry mass. Cell wall content was
measured as neutral detergent fibre (NDF, a measure
of total cell wall, not including pectins), acid
detergent lignin (ADL, a measure of lignin plus
cutin) and cellulose, following Van Soest (1994).
Leaf toughness was measured on fresh hydrated
green leaves, since differential water content among
senesced leaves could confound interpretation.
Toughness, measured as work to fracture, was
determined for five leaves per species (within 24 h
of collection) using a shearing test on a 5-mm-wide
strip cut from one side of the leaf and sheared
transversely at a random position along its length
following Read and Sanson (2003). For Calophyllum
caledonicum, which has closely spaced secondary
veins perpendicular to the midrib, we cut across the
veins rather then parallel to them. The force–
displacement curve derived from each test was
analysed by Leaf2000 software (M. Logan, Monash
University). We calculated the work required to shear
the leaf strip as the area under the force–displacement
curve, expressed per unit strip width, and specific
work to shear as the work to shear per unit strip
thickness, with thickness measured by a digital
micrometer. We measured specific leaf area (SLA)
of senesced leaves, but for a few species for which
leaves were scarce, we used green leaves (1–12%
higher for species in which both leaf classes were
measured).
Leaf decomposition experiment
Rates of leaf decomposition were determined by a
glasshouse-based litter bag experiment, modified
from Cornelissen (1996). Air-dried leaves were dried
at 40C for 48 h and weighed. Then, one or more leaf
Forest Ecology
55
halves (depending on leaf size) were placed in a
polyester litterbag with a mesh size of 1 mm.
Individual bag size was determined by the size of
the leaves to keep the individual leaf in contact with
the substrate. Three to five replicate bags per species
were used, depending on leaf availability. We
compared leaf decomposition rates in packs of litter
weighing 1–2 g with litterbags containing an individual leaf for seven species. Plastic planter bags ca.
20 9 10 cm were prepared 1 week prior to litterbag
burial with 7-cm depth of commercial plant-based
compost, and a thin layer of mixed decomposing leaf
litter from the Monash University systems garden to
ensure a range of soil microbes was present. One
litter bag was placed in each pot in March 2006 and
covered with 5 cm of compost. Pots were randomised
in position in a glasshouse (mean daily maximum
temperature of 26.1C, with a minimum humidity of
75%) and watered twice daily to maintain moist
conditions typical of rainforest. After 7 weeks, leaves
in some additional ‘trial’ bags (checked every ca.
2 weeks to observe the rate of decomposition) were
already highly decomposed. Therefore, litter bags in
the main experiment were removed after 8 weeks so
that differences among species in decomposition rate
would be apparent. The leaves were lightly rinsed to
remove extraneous material and dried at 40C to
constant mass. Samples were then weighed to determine percentage mass loss.
in decomposition rates between leaf packs and
individual leaves. Principal components analysis
(PCA) was used to summarise traits of senesced
leaves suggested in earlier studies to affect decay
rates. Hierarchical partitioning was used to determine
independent contributions of traits to decomposition
rate using the hier.part package v. 1.0–2 (Mac Nally
and Walsh 2004) of R v. 2.5.1 (R Development Core
Team 2004): IHP indicates the percentage contribution of each trait (limited to 12 independent variables)
to the total explained variance; rand.hp uses a
randomisation test to compute Z-scores for tests of
statistical significance, the latter based on an upper
0.95 confidence limit (Z C 1.65). In order to reduce
the set of independent variables, we excluded SLA
and Ni, which were not significantly correlated with
decay rate, and cellulose, which was represented
within the NDF variable. Due to rounding errors that
can occur when more than nine independent variables
are included (Walsh and Mac Nally 2007), we
removed variables that did not consistently make a
significant contribution in repeated analyses with
variables entered in differing order (N, C:N and total
phenolics). SYSTAT v. 11 was used for all other
analyses. A critical value of a = 0.05 was used for
hypothesis tests.
Data analysis
Macronutrient and Ni concentrations
in green leaves
The data were analysed at two levels. First, Nothofagus species were compared with the average of other
species (the average of the 4–7 species’ means)
across all sites using a randomised block design (sites
as blocks). Any ‘site’ effects are potentially the result
of differences in site growth conditions and/or
differences in the suite of species. The data were
then analysed separately for each site using a planned
contrast between the dominant Nothofagus species
and co-occurring species, i.e. comparing the average
value of the Nothofagus species with the average of
the means of the co-occurring species. We were only
interested in large-scale trends, and so did not
compare Nothofagus species to individual species at
each site. Pearson correlation was used to investigate
associations between decomposition rate and other
leaf traits. T-tests were used to determine differences
Results
Within each site, there was ca. 2- to 8-fold variation in
macronutrient concentrations among species, with
most variation in K (3- to 4-fold), Ca (3- to 8-fold),
Mg (2- to 7-fold) (Fig. 1) and Ca:Mg (4- to 8-fold)
(Table 2). However, there were no significant differences in nutrient concentrations between Nothofagus
species and the mean of co-occurring species across
all sites (randomised block analysis, Fig. 1). High
foliar N:P ratios suggested P-limitation in Nothofagus
species and co-occurring species (Table 2). Foliar P
concentration varied 2- to 3-fold among species at
each site, but was lower in Nothofagus than the
average of co-occurring species only at Col de Yaté
(planned contrast of Nothofagus versus all other
species) (Fig. 1). Nothofagus had higher N concentrations at Col de Mouirange than the average of co-
56
A.G. Van der Valk (ed.)
Fig. 1 Concentrations of macronutrients and Ni in green
leaves at each site. Mean ± SE are given for each species, with
Nothofagus species shown by filled bars. Species are given in
the order shown in Table 1. Ch, Col de Mouirange Haut; Cb,
Col de Mouirange Bas; D, Dzumac; Y, Col de Yaté. The results
of randomised block ANOVA are given (L, data log-
transformed for analysis). Asterisks indicate where there is a
significant difference between Nothofagus and the mean of cooccurring species at a site (planned contrasts); a line below the
asterisk indicates no significant difference after removal of an
outlier
Table 2 Chemical ratios in green and senesced leaves of Nothofagus and other species
Nothofagus species
N:P green
33 ± 2 (30–39)
Other species
30 ± 3 (18–49)
F1,3, P
0.59, 0.499
N:P senesced
80 ± 30 (27–166)
63 ± 6 (25–128)
0.26, 0.647
Ca:Mg green
Ca:Mg senesced (L)
3.8 ± 1.4 (2.0–7.8)
5.7 ± 1.8 (3.4–11.1)
5.2 ± 1.2 (0.8–19.6)
7.7 ± 2.3 (3.6–14.2)
4.14, 0.135
4.86, 0.115
C:N sensesced
C:P sensesced (L)
Lignin:N sensesced
Lignin:P sensesced (L)
85 ± 4 (73–89)
6,876 ± 2,714 (2,443–14,793)
36 ± 7 (24–57)
3,491 ± 2,039 (933–9,583)
90 ± 8 (51–177)
5,479 ± 537 (2,540–12,221)
28 ± 2 (11–53)
1,835 ± 90 (355–4,832)
0.16, 0.715
0.01, 0.952
0.89, 0.414
0.16, 0.717
The data presented are mean ± SE (mass-based ratios) of Nothofagus (n = 4 species/sites) and other species (n = 4 sites, the value
for each site being the average of 4 to 7 species’ means). The range of species’ means is given in brackets. ADL is given as ‘lignin’.
The results of randomised block analysis are given. L, log-transformed for analysis
occurring species, but lower N at Yaté (planned
contrast, Fig. 1). At all sites except Col de Mouirange
Haut (CDMh), K concentrations were lower in
Nothofagus species than co-occurring species
(planned contrasts, Fig. 1). Ca was lower in Nothofagus at Col de Mouirange Bas (CDMb), but higher at
CDMh than co-occurring species, with no significant
differences at Yaté and Dzumac (planned contrasts,
Fig. 1). Mg concentrations and Ca:Mg were highly
variable, with Mg higher and Ca:Mg (Table 2) lower
in Nothofagus than co-occurring species at Dzumac,
but Mg lower at CDMb (planned contrasts, Fig. 1). Ni
levels varied 2- to 12-fold among species at each site,
and were lower in Nothofagus species than the mean
of co-occurring species across sites (Fig. 1, randomised block analysis), even after outliers were excluded
(log10 Ni: F = 32.8, P = 0.011). However, Ni concentration was significantly lower in Nothofagus than
Forest Ecology
co-occurring species only at CDMh and Yaté (but
near-significant at CDMb: P = 0.06) (planned contrast, Fig. 1), but not after outliers were excluded.
High Ni contents were recorded in Codia discolor and
Neoguillauminia cleopatra, but no species showed
evidence of hyperaccumulation (Ni [ 1 mg g-1
foliar dry mass: Brooks et al. 1977). Comparison of
sites for each element found differences only in foliar
P concentrations (CDMb [ CDMh and Dzumac)
(Fig. 1).
Macronutrient and Ni concentrations in senesced
leaves
There were no significant differences in macronutrient concentrations (resorption proficiency) and ratios
of senesced leaves between Nothofagus species and
the mean of co-occurring species across all sites
(randomised block analysis, Fig. 2, Table 2). All
species, except Crossostylis grandiflora, Alphitonia
neocaledonica and Guettarda eximia, had N proficiency below the 0.7% value suggested by
Killingbeck (1996) as indicating high N proficiency,
with only Arillastrum gummiferum having N concentrations at the 0.3% level suggested to represent
‘ultimate’ N-proficiency (the maximum level to
which nutrients can be reduced). At CDMh, higher
N was recorded in senesced N. aequilateralis leaves
57
than the mean of co-occurring species, with no
differences at other sites (Fig. 2). Trends in P
concentration varied among sites, with higher P in
Nothofagus than the mean of co-occurring species at
Dzumac, but lower at CDMb. Only N. discoidea
(Fig. 2) had high P proficiency (\0.04% P for
evergreens: Killingbeck 1996), with no species
reaching the ultimate proficiency of 0.01% (Killingbeck 1996). K in senesced leaves followed a similar
pattern to green leaves, but was significantly lower in
N. codonandra at Dzumac (Fig. 2). Ca and Mg were
lower in senesced leaves of N. discoidea than cooccurring species at CDMb (Fig. 2). Ni concentrations were lower on average in senesced Nothofagus
leaves across sites (Fig. 2).
Resorption efficiency ranged from 28% to 64% for
N, 23% to 90% for P and 12% to 89% for K across
species. There were no significant differences in
resorption efficiency of N, P and K between Nothofagus and the mean of co-occurring species across all
sites (Fig. 3), and no differences among sites
(P \ 0.05).
Foliar decomposition rates and leaf traits
There was 9-fold variation in mass loss among
species, and leaves of Nothofagus species decomposed at less than half the rate of co-occurring species
Fig. 2 Concentrations of macronutrients and Ni in senesced leaves at each site. Mean ± SE are given for each species, with
Nothofagus species shown by filled bars. Abbreviations and results are given as in Fig. 1
58
A.G. Van der Valk (ed.)
Table 3 Relationships between decomposition rate (logtransformed) and traits of senesced leaves
Leaf trait
SLA
Nitrogen
Fig. 3 Resorption efficiency of Nothofagus species and cooccurring species, averaging across the mean of species (±SE).
Randomised block ANOVA indicated no significant effect of
species (Nothofagus versus other species) or site. Species
effects: N, F1,3 = 0.2, P = 0.660; P, F1,3 = 0.2, P = 0.657;
K, F1,3 = 2.2, P = 0.236
on average (18% vs. 42% mass loss) (randomised
block analysis, Fig. 4), with slower decomposition
recorded in Nothofagus at CDMb and Dzumac (not at
Yaté when the outlier was removed) (planned contrasts) (Fig. 4). Comparisons of single leaves with
leaf packs showed either no difference (three species)
or faster decay in single leaves (four species,
P \ 0.05).
Of the leaf traits commonly considered to influence decomposition rates (Table 3), only NDF
Fig. 4 Decomposition rates (given as percentage mass loss
after 8 weeks) of Nothofagus species and co-occurring species
at each site. Mean ± SE are given for each species, with
Nothofagus species shown by filled bars. Abbreviations and
results are given as in Fig. 1
R
IHP
Z
0.27
-0.03
Phosphorus
0.50*
Nickel
0.02
5.6
-0.03
0.46
C:N
-0.23
C:P (L)
-0.56**
9.1
NDF
-0.78***
25.5
3.39*
ADL
-0.60**
8.7
0.52
Cellulose
-0.69***
Lignin:N
-0.59**
7.7
0.21
Lignin:P (L)
-0.64***
10.3
1.17
Work to shear (L)
-0.48*
Specific work to shear (L)
-0.65***
Total phenolics
-0.32
Tannin activity
PCA Component 1
-0.51*
-0.84***
PCA Component 2
0.01
PCA Component 3
0.16
8.0
0.14
13.9
1.26
11.1
0.90
Values given are Pearson correlation coefficients (R) in the first
column, with results of hierarchical partitioning analysis (IHP,
the independent contribution of each trait as a percentage of the
explained variance, and the Z-score) in the second and third
columns. Significance levels for R are given as follows:
* P \ 0.05; ** P \ 0.01; *** P \ 0.001. Significant Z-scores
(Z C 1.65) are indicated by a single asterisk. Only nine
variables were included in the final hierarchical partitioning
analysis. L, log-transformed for analysis
differed significantly between Nothofagus and cooccurring species across sites (randomised block
analysis, Figs. 2, 5, 6; Table 2). For other traits,
differences were recorded between Nothofagus and
the means of co-occurring species at some study sites,
but not always consistently across sites. For example,
leaf toughness (work to shear) was higher on average
in N. codonandra than co-occurring species at
Dzumac, but similar or lower in Nothofagus than
co-occurring species at other sites (planned contrasts,
Fig. 5). Leaf toughness per unit thickness (specific
work to shear) was higher on average in Nothofagus
than co-occurring species at Yaté and Dzumac, with
no difference at the Col de Mouirange sites (Fig. 5).
No difference was recorded for SLA (Fig. 5). Total
phenolics and tannin activity were higher in Nothofagus than the average of co-occurring species at
CDMb and Dzumac, but tannin activity was lower in
Forest Ecology
Fig. 5 Physical traits and cell wall components of leaves at
each site. Traits were measured on senesced leaves except for
mechanical traits, which were measured on green leaves, and
SLA, which was measured on a mixture of green and senesced
59
leaves. Mean ± SE are given for each species, with Nothofagus species shown by filled bars. Abbreviations and results are
given as in Fig. 1
Fig. 6 Phenolic
concentrations of senesced
leaves at each site. Protein
precipitation provides a
measure of tannin activity.
Mean ± SE are given for
each species, with
Nothofagus species shown
by filled bars. Abbreviations
and results are given as in
Fig. 1
N. balansae than co-occurring species on average at
Yaté (planned contrasts, Fig. 6). The C:N ratio was
on average lower for N. aequilateralis at CDMh than
co-occurring species, and N. discoidea had significantly higher C:P, lignin:N and lignin:P ratios at
CDMb than co-occurring species (planned contrasts,
data not presented). For all these traits, where there
was a significant difference across or within sites,
there was usually at least one other species with
similar values to those of Nothofagus.
Leaf decomposition rates correlated positively with
foliar P and negatively with measures of cell wall,
lignin:N, lignin:P, C:P, tannin activity, work to shear
and specific work to shear (Table 3). No correlations
were recorded with SLA, N, C:N, Ni and total
phenolics. PCA reduced the data to three main
components explaining 69% of the variation among
species (Fig. 7a). Lignin:P, lignin:N, ADL, NDF and
C:P contributed most strongly to Component 1, C:N,
N and SLA to Component 2, and work to shear,
specific work to shear and cellulose to Component 3.
The configuration plot of the first two components
showed no clustering of Nothofagus species or of sites
(Fig. 7a). Leaf decomposition rate correlated negatively with Component 1 (Fig. 7b), with more
variation explained (71%) than by any individual
variable (Table 3). Multiple regression showed that
82% of the variation in decomposition rate (log-
60
A.G. Van der Valk (ed.)
transformed) was explained by the nine variables
indicated in Table 3, but hierarchical partitioning
indicated a significant independent contribution of
only NDF (Table 3).
Discussion
Foliar element concentrations of green leaves
Fig. 7 PCA of the traits of senesced leaves. (a) The
configuration plot of the first two component axes, and (b)
the relationship between leaf decomposition rate and Component 1. The PCA was undertaken on the full set of variables
given in Table 3. The Nothofagus species are shown by their
initials in (a)
Foliar macronutrient concentrations were commonly
on the low side of the range found in forests on
ultramafic and non-ultramafic soils (Table 4), suggesting that the study species on average have high
INUE. P appeared to be the limiting nutrient, with
N:P ratios [16 (cf. Güsewell 2004) for all species,
consistent with global trends suggesting that Plimitation is particularly common in tropical environments (Reich and Oleksyn 2004). However, in
contrast to predictions, Nothofagus species did not
have lower green-leaf nutrient concentrations (higher
INUE) on average than co-occurring species for P or
other nutrients, although K concentrations were lower
in Nothofagus at three sites.
High levels of soil Ni have been recorded at these
sites, but to a lesser extent at Col de Mouirange (Read
et al. 2006). The degree of toxicity of Ni is sitedependant, with factors such as low soil pH and Ca
concentrations influencing uptake (Proctor and Woodell 1975). Since Ca concentrations are low in these
soils (Read et al. 2006), there is potential for Ni
toxicity, particularly at Yaté and Dzumac. In addi-
Table 4 Foliar element concentrations (mg g-1) of green leaves in rain forest species on ultramafic and non-ultramafic soils
New Caledonian Nothofagusa
New Caledonian ultramaficb
Malaysian ultramaficc
6.1–26.4
Tropical rain forest non-ultramaficd
N
7.4–12.7
6.50–15.4
P
0.25–0.48
0.18–0.60
K
2.5–7.5
2.36–12.7
1.3–17.7
2.4–38.5
Ca
4.1–14.5
2.4–24.4
1.9–31.5
1.1–34.9
0.24–1.8
6.1–35.3
0.2–2.5
Mg
1.3–3.5
0.8–5.2
0.9–12.8
0.1–10.4
Ni
0.002–0.017
0.01–0.12
0.00–0.65
0.001–0.007
a
From this study, Jaffré (1980) and Read et al. (2002)
b
From this study, Jaffré (1980) (mean of 118–140 species), Enright et al. (2001) and Read et al. (2002)
c
Proctor et al. (1989)
d
Data from lowland to upper montane rain forests (Ovington and Olson 1970; Stark 1970; Tanner 1977; Grubb 1977; Thompson
et al. 1992; Read et al. 2002; Plummer 2007), with few data available for Ni (Read et al. 2002; Plummer 2007)
Forest Ecology
tion, Nothofagus litter appears to be more acidic (pH
4) than that of mixed rain forest (ca. pH 5.5) (CDMh:
McCoy 1991) and may increase Ni availability.
Resistance to Ni toxicity is influenced by the ability
to limit its uptake and transport, as well as tissue
tolerance, each mechanism potentially incurring an
energy cost. Foliar Ni content was quite variable
among species in this study, but was generally at
lower levels (7% of species with [0.05 mg g-1 Ni)
than found in New Caledonian rain forests by Jaffré
(1980) (58% of forest species on ultramafic soils
[0.05 mg g-1 Ni), probably due largely to differences in soil and topographic situation between the
studies. The low foliar concentration of Ni in
Nothofagus species could be partly due to their
ectomycorrhizae (VAM also weakly present in at
least one species: Perrier et al. 2006). Mycorrhizal
associations may play a crucial role in heavy metal
tolerance in some plants (Baker and Walker 1989;
Perrier et al. 2006), either due to general benefits
conferred by the relationship (Meharg and Cairney
2000) or due to reduced metal uptake (Wilkinson and
Dickinson 1995). If the ectomycorrhizal association
reduces Ni uptake, it may provide an energetically
efficient resistance mechanism. However, the low
foliar Ni might be partly a consequence of the high
cell wall fraction in Nothofagus species (Fig. 5), and
so may not reflect trends among species in protoplasmic concentrations.
Post-leaf NUE—resorption efficiency
and proficiency
Recycling of nutrients by decomposition and mineralisation is less efficient than resorption (Aerts and
Chapin 2000), and we predicted that Nothofagus
species would show higher PNUE than co-occurring
species. Instead, resorption efficiency for N, P and K
was highly variable across species, and the only
significant difference was the higher resorption of P
by N. discoidea at CDMb than co-occurring species.
Low resorption rates of N (\50%) have been
recorded in species with low green leaf concentrations (\1%), probably due to proportionately less leaf
N being accessible for resorption (Diehl et al. 2003).
Similarly, all species in this study had N concentrations below ca. 1.5%, with correspondingly low N
resorption efficiency (\50%) in most species. P, like
N, is fairly mobile in plants, but resorption was more
61
variable among sites and species, suggesting lower
reliance of some species on resorption even in a Plimited system. Most tropical rain forest species have
mycorrhizae that are expected to increase substrate-P
uptake, arbuscular mycorrhizae in most species
(Connell and Lowman 1989; Smith and Read 1997)
and ectomycorrhizae in Nothofagus (Read and Hope
1996) and some other monodominant trees (e.g.
Connell and Lowman 1989; Newbery et al. 1997). It
may be less energetically expensive for species to
rely on substrate-available P than the resorption
pathway if the mycorrhizal relationship is highly
efficient (Newbery et al. 1997). The variation
recorded in K resorption is not unusual as K is
highly leachable, although leaching is likely to be
lower on these infertile soils (Chapin 1980). Ca
concentrations increased from green to senesced
leaves (data not presented), as expected, given that
Ca is not phloem-mobile (Chapin 1980).
Resorption proficiency may provide a better
estimate of PNUE if plants control the minimum
N and P concentrations of senesced leaves rather
than the proportion of nutrients withdrawn (Killingbeck 1996; Aerts and Chapin 2000). N and P
concentrations were low in senesced leaves of all
species; most species had high resorption proficiency of N by Killingbeck’s (1996) criterion, and
all fell well within the range for highly proficient P
resorption, further suggesting P-limitation. Levels of
N and P were relatively low even compared with
leaf litter from an ultramafic mountain in Sabah
(means of 0.8 to 1.3 mg g-1 N and 0.17 to
0.26 mg g-1 P) (Proctor et al. 1989). However,
there was no evidence of higher resorption proficiency in Nothofagus, except for high P-proficiency
in N. discoidea at CDMb.
Differences in community-level traits can occur
among sites due to both differences in species
composition and effects of growth conditions on
plant traits (phenotypic plasticity) (Richardson et al.
2005). In this study, differences among Nothofagus
species, and in the trends between Nothofagus and
co-occurring species among sites, may be due to
either varying species composition or due to site
effects on phenotypes. Killingbeck (1996) suggested
that resorption is driven less by nutrient availability
than by controls at the species level. Aerts and
Chapin (2000) found no consistent phenotypic differences in resorption efficiency between low and
62
high fertility conditions, and similarly, nutrient
availability did not strongly affect resorption in a
fertilisation experiment (Vitousek 1998). However,
phenotypic variation has been recorded in green-leaf
nutrient concentrations in N. aequilateralis among
sites (Read et al. 2002). Hence, phenotypic variation
may occur in resorption efficiency and proficiency.
Mg and Ni concentration of senesced leaves
The deciduous tree Peltogyne gracilipes forms
monodominant forests on Mg-rich soils in Brazilian
Amazonia and has high concentrations of Mg in
both green and senesced leaves (Villela and Proctor
1999). Villela and Proctor (2002) suggested that
pulses of Mg from decomposing leaves of P.
gracilipes may be mildly toxic on these soils and
linked to its monodominance if these levels of Mg
suppress growth of less tolerant species. However,
Nothofagus species on the New Caledonian ultramafic soils showed no evidence of consistent high
concentrations of Mg and Ni in either green or
senesced leaves; instead, levels of Ni in green and
senesced leaves were lower on average than in cooccurring species. Hence, there is no evidence for
the allelopathic mechanism of monodominance
suggested for Peltogyne.
Litter decomposition rates and causal factors
Falling leaves typically account for more than 70% of
all aboveground litter (Killingbeck 1996). Therefore,
leaf decomposition has a large influence on nutrient
recycling rates. In nutrient-poor habitats, such as
studied here, leaf litter must represent a major pool of
accessible nutrients that are deficient in the mineral
soil. Litter quality appears to be the best determinant
of decay rates in the tropics (Meentemeyer 1978;
Aerts 1997), with both chemical and physical traits
influencing its value as a nutrient resource to
decomposers (Swift et al. 1979). N and P are often
the most limiting nutrients (Heal et al. 1997; Aerts
and Chapin 2000), but unlike some other studies in
tropical rain forests (Santiago 2007), in this study, N
did not correlate with decay rates, and P only weakly.
On these severe soils, carbon:nutrient ratios may be
of more relevance. C:N ratios exceeding 25:1 are
considered to be high (Heal et al. 1997); for all of our
A.G. Van der Valk (ed.)
study species, the mean C:N was above 50:1, but
litter decay did not correlate with C:N, although it did
with C:P. The largest quantitative demand by
decomposers is the energy released from organic
compounds, including cell wall polysaccharides (La
Caro and Rudd 1985), but cell wall components can
also provide a physical barrier to nutrients. We found
that lignin:nutrient ratios were better predictors of
decay rates than C:nutrient ratios, reflecting the
influence of both nutrient limitation and the quality
of carbon as an energy source (and barrier) to
decomposers, lignin being highly recalcitrant to
decay (Hammel 1997). Leaf decay was also negatively correlated with lignin concentration alone,
consistent with some other studies (La Caro and Rudd
1985; Cornelissen 1996; Hobbie and Vitousek 2000).
Lignin concentrations of our study species were not
particularly high; the mean across all species was
16%, with none exceeding 35%, compared with the
level of 50% considered likely to significantly retard
litter decay rates (Mesquita et al. 1998), although
Aerts (1997) found that low lignin concentrations
could inhibit decay if there was a lack of specialised
lignin-degrading organisms. However, in our study,
the measure of total cell wall (NDF) correlated more
strongly with decomposition rates than lignin.
Some secondary compounds can interact with the
nutritional quality of litter to slow decay (Heal et al.
1997). Phenolics, in particular tannins, can suppress
litter decomposition (Northup et al. 1998; Kraus et al.
2003), especially in the later stages of decay (Mesquita et al. 1998; Loranger et al. 2002). Consistent with
this, tannin activity (but not total phenolics) was
negatively correlated with litter decay rates in this
study, but with no significant independent effect; a
stronger effect may have been observed if the leaves
of more species had decayed substantially. The
physical characteristics of leaves can also affect litter
decay rates (Swift et al. 1979). In particular, leaf
mechanical properties have been found to correlate
negatively with litter decomposition (Cornelissen
et al. 1999). We measured leaf toughness as both
work to shear and specific work to shear, but only the
latter correlated with decomposition rates. However,
it did not make a significant independent contribution
to decomposition rates; its negative correlation with
decomposition rate may be largely due to its positive
correlation with NDF (R = 0.64, P \ 0.001), but
different results may have occurred if we had not
Forest Ecology
excluded larger detritivores by the small mesh size.
Even though only NDF made a significant independent contribution to variation in decomposition rates,
the higher correlation of decomposition rate with the
main component axis of the PCA than with any
individual variables, and the high amount of variation
explained when multiple variables were included in
the regression model, is consistent with the influence
of multiple factors in leaf decomposition, despite
these apparently being minor individual contributions
relative to NDF. The relatively low rate of Nothofagus leaf decay appears to be largely influenced by
high cell wall content and ratios of lignin:N and
lignin:P.
The decomposition experiment may not reflect
rates of decay in nature. We attempted to mimic
climatic conditions and to introduce a variety of soil
biota, but the experiment is nevertheless artificial,
and does not allow the complexity of interactions that
exist in nature (e.g. Milton and Kaspari 2007). In
addition, more detailed interpretation is precluded by
using only a single harvest. More importantly, decay
rates of individual leaves often exceeded those of leaf
packs, not surprisingly as individual leaves had
greater contact with the substrate and could be
accessed more easily by decomposers. Interactions
among species can also be important. Mixed litters
can interact to affect decomposition rates of individual leaves, either positively or negatively (Gartner
and Cardon 2004), with decomposition strongly
controlled by traits of the dominant species (Hoorens
et al. 2003). A study at Col de Mouirange found that
in areas dominated by N. aequilateralis, the leaf litter
was dominated by a large amount of undecomposed
leaves and branch material of this species with a low
pH (McCoy 1991). Our experiment found that leaf
litter of Nothofagus was slow to decompose compared with co-occurring species, and together with
the low pH of Nothofagus litter, must have a
significant effect on nutrient cycling and availability
in the monodominant forests.
Does leaf-level NUE influence monodominance
of Nothofagus?
We found no evidence that leaves of Nothofagus
species generally function at lower nutrient concentrations or are better on average at conserving scarce
nutrients than co-occurring species. We note that our
63
main analyses were limited in power by the low
number of sites, and so, conclusions should be
cautious. Within-site comparisons, however, showed
high variability among species for many leaf traits,
with no suggestion that Nothofagus species show
consistent superior leaf-level NUE compared with
other common canopy species. Some species that
dominate tropical forest canopies produce leaf litter
that is slow to decompose, e.g. Gilbertiodendron
dewevrei in monodominant Congolese forest (Torti
et al. 2001), and Cecropia sciadophylla in Amazonian secondary forest (Mesquita et al. 1998), and so is
slow to release nutrients. Similarly, the litter of
Nothofagus is relatively slow to decompose, but it is
not clear whether this provides an advantage to
Nothofagus species since there is no evidence to date
of higher NUE than co-occurring species. However,
the low decomposability of Nothofagus leaf litter
may confer an advantage by slowing the establishment and growth of its competitors by other means:
litter accumulation may change topsoil conditions by
intercepting light and rain and the transfer of heat and
water may be affected (Facelli and Pickett 1991), and
the low pH may increase availability of Ni. In
addition, succession may be delayed by effects of
litter accumulation on fuel loads, increasing the
probability of the forests burning during a severe dry
season and potentially creating an environment
suitable for Nothofagus regeneration. However, while
these factors might explain the maintenance of
dominance by delayed growth of competitors, or
interruption of succession, they do not easily explain
how Nothofagus achieves early dominance of these
forests after disturbance. Other components of plantlevel NUE, such as nutrient uptake rates and carbon
gain per unit nutrient acquired, including effects of
leaf lifespan (Aerts and Chapin 2000) and ectomycorrhizal relationships (Newbery et al. 1997; Smith
and Read 1997; Aerts 2002; Erland and Taylor 2002),
may be contributing to dominance by Nothofagus
species. These warrant investigation in the context of
understanding the mechanisms that promote monodominance by some species on severe soils.
Acknowledgments We thank G. Sanson and R. Carpenter for
assistance in collecting leaves, G. Sanson for assistance in
measurement of leaf toughness, M. Logan for advice regarding
sampling design and data analysis, S. Kerr for measurement of
tannin activity and for general laboratory assistance, B. Lees
for measurement of N and C, and D. Griepsma (ASIRC Pty
64
Ltd) for ICP-OES analysis. We also thank an anonymous
reviewer for helpful suggestions.
References
Aerts R (1996) Nutrient resorption from senescing leaves of
perennials: are there general patterns? J Ecol 84:597–608.
doi:10.2307/2261481
Aerts R (1997) Climate, leaf litter chemistry and leaf litter
decomposition in terrestrial ecosystems: a triangular
relationship. Oikos 79:439–449. doi:10.2307/3546886
Aerts R (2002) The roles of various types of mycorrhizal fungi
in nutrient cycling and plant competition. In: van der
Heijden MGA, Sanders I (eds) Mycorrhizal ecology.
Springer-Verlag, Berlin
Aerts R, Chapin FS (2000) The mineral nutrition of wild plants
revisited: a re-evaluation of processes and patterns. Adv
Ecol Res 30:1–67. doi:10.1016/S0065-2504(08)60016-1
Asquith TN, Butler LG (1985) Use of dye-labeled protein as
spectrophotometric assay for protein precipitants such as
tannin. J Chem Ecol 11:1535–1544. doi:10.1007/BF01012
199
Baker AJM (1981) Accumulators and excluders—strategies in the
response of plants to heavy metals. J Plant Nutr 3:643–654
Baker AJM, Walker PL (1989) Ecophysiology of metal uptake
by tolerant plants. In: Shaw AJ (ed) Heavy metal tolerance
in plants: evolutionary aspects. CRC Press, Boca Raton, FL
Brooks RR, Lee J, Reeves RD, Jaffré T (1977) Detection of
nickeliferous rocks by analysis of herbarium specimens of
indicator plants. J Geochem Explor 7:49–57. doi:10.1016/
0375-6742(77)90074-7
Chapin FS (1980) The mineral nutrition of wild plants. Annu
Rev Ecol Syst 11:233–260. doi:10.1146/annurev.es.11.
110180.001313
Clarkson DT, Hanson JB (1980) The mineral nutrition of
higher plants. Annu Rev Plant Physiol 31:239–298. doi:
10.1146/annurev.pp.31.060180.001323
Connell JH, Lowman MD (1989) Low-diversity tropical rain
forests: some possible mechanisms for their existence.
Am Nat 134:88–119. doi:10.1086/284967
Cork SJ, Krockenberger AK (1991) Methods and pitfalls of
extracting condensed tannins and other phenolics from
plants: insights from investigations on Eucalyptus leaves.
J Chem Ecol 17:123–134. doi:10.1007/BF00994426
Cornelissen JHC (1996) An experimental comparison of leaf
decomposition rates in a wide range of temperate plant
species and types. J Ecol 84:573–582. doi:10.2307/
2261479
Cornelissen JHC, Pérez-Harguindeguy N, Diaz S, Grime JP,
Marzano B, Cabido M et al (1999) Leaf structure and
defence control litter decomposition rate across species and
life forms in regional floras on two continents. New Phytol
143:191–200. doi:10.1046/j.1469-8137.1999.00430.x
Diehl P, Mazzarino MJ, Funes F, Fontenla S, Gobbi M, Ferrari J
(2003) Nutrient conservation strategies in native AndeanPatagonian forests. J Veg Sci 14:63–70. doi:10.1658/
1100-9233(2003)014[0063:NCSINA]2.0.CO;2
Enright NJ, Rigg L, Jaffré T (2001) Environmental controls on
species composition along a (maquis) shrubland to forest
A.G. Van der Valk (ed.)
gradient on ultramafics at Mont Do, New Caledonia. S Afr
J Sci 97:573–580
Erland S, Taylor AFS (2002) Diversity of ecto-mycorrhizal
fungal communities in relation to the abiotic environment.
In: van der Heijden MGA, Sanders I (eds) Mycorrhizal
ecology. Springer-Verlag, Berlin
Facelli JM, Pickett STA (1991) Plant litter: its dynamics and
effects on plant community structure. Bot Rev 57:1–32.
doi:10.1007/BF02858763
Gartner TB, Cardon ZG (2004) Decomposition dynamics in
mixed-species leaf litter. Oikos 104:230–246. doi:
10.1111/j.0030-1299.2004.12738.x
Graham HD (1992) Stabilization of Prussian Blue color in the
determination of polyphenols. J Agric Food Chem
40:801–805. doi:10.1021/jf00017a018
Grubb PJ (1977) Control of forest growth and distribution on
wet tropical mountains: with special reference to mineral
nutrition. Annu Rev Ecol Syst 8:83–107. doi:10.1146/
annurev.es.08.110177.000503
Güsewell S (2004) N:P ratios in terrestrial plants: variation and
functional significance. New Phytol 164:243–266. doi:
10.1111/j.1469-8137.2004.01192.x
Hammel KE (1997) Fungal degradation of lignin. In: Cadisch G, Giller KE (eds) Driven by nature: plant litter
quality and decomposition. CAB International, Wallingford, UK
Hart TB, Hart JA, Murphy PG (1989) Monodominant and species-rich forests of the humid tropics: causes for their cooccurrence. Am Nat 133:613–633. doi:10.1086/284941
Heal OW, Anderson JM, Swift MJ (1997) Plant litter quality
and decomposition: an historical overview. In: Cadisch G,
Giller KE (eds) Driven by nature: plant litter quality and
decomposition. CAB International, Wallingford, UK
Hobbie SE, Vitousek PM (2000) Nutrient limitation of decomposition in Hawaiian forests. Ecology 81:1867–1877
Hoorens B, Aerts R, Stroetenga M (2003) Does initial litter
chemistry explain litter mixture effects on decomposition?
Oecologia 137:578–586. doi:10.1007/s00442-003-1365-6
Jaffré T (1980) Étude écologique du peuplement végétal des
sols dérivés de roches ultrabasiques en Nouvelle Calédonie. Collection Travaux et Documents de l’ORSTOM
no. 124, ORSTOM, Paris
Jaffré T (1992) Floristic and ecological diversity of the vegetation on ultramafic rocks in New Caledonia. In: Baker
AJM, Proctor J, Reeves RD (eds) The vegetation of
ultramafic (serpentine) soils. Intercept Ltd, Andover, UK
Jaffré T, Morat P, Veillon J-M, Rigault F, Dagostini G (2004)
Composition et caractéristiques de la flore indigène de
Nouvelle-Calédonie. Nouméa. 2nd ed. IRD Documents
Scientifiques et Techniques II 4, pp 121
Killingbeck KT (1996) Nutrients in senesced leaves: keys to
the search for potential resorption and resorption proficiency. Ecology 77:1716–1727. doi:10.2307/2265777
Kraus TEC, Dahlgren RA, Zasoski RJ (2003) Tannins in
nutrient dynamics of forest ecosystems—a review. Plant
Soil 256:41–66. doi:10.1023/A:1026206511084
La Caro F, Rudd RL (1985) Leaf litter disappearance rates in
Puerto Rican montane rain forest. Biotropica 17:269–276.
doi:10.2307/2388589
Loranger G, Ponge J-F, Imbert D, Lavelle P (2002) Leaf
decomposition in two semi-evergreen tropical forests:
Forest Ecology
influence of litter quality. Biol Fertil Soils 35:247–252.
doi:10.1007/s00374-002-0467-3
Mac Nally R, Walsh CJ (2004) Hierarchical partitioning public-domain software. Biodivers Conserv 13:659–660. doi:
10.1023/B:BIOC.0000009515.11717.0b
McCoy S (1991) Edaphic controls influencing the distribution
of Nothofagus aequilateralis on ultrabasic soils at the Col
de Mouirange, New Caledonia. Honours Dissertation,
Australian National University, Australia
Meentemeyer V (1978) Macroclimate and lignin control of litter
decomposition rates. Ecology 59:465–472. doi:10.2307/
1936576
Meharg AA, Cairney JWG (2000) Co-evolution of mycorrhizal
symbionts and their hosts to metal-contaminated environments. Adv Ecol Res 30:69–112. doi:10.1016/S00652504(08)60017-3
Mesquita RCG, Workman SW, Neely CL (1998) Slow litter
decomposition in a Cecropia-dominated secondary forest
of central Amazonia. Soil Biol Biochem 30:167–175. doi:
10.1016/S0038-0717(97)00105-3
Milton Y, Kaspari M (2007) Bottom-up and top-down regulation of decomposition in a tropical forest. Oecologia
153:163–172. doi:10.1007/s00442-007-0710-6
Nascimento MT, Barbosa RI, Villela DM, Proctor J (2007)
Above-ground biomass changes over an 11-year period in
an Amazonian monodominant forest and two other lowland forests. Plant Ecol 192:181–191. doi:10.1007/
s11258-007-9303-z
Newbery DM, Alexander IJ, Rother JA (1997) Phosphorus
dynamics in a lowland African rain forest: the influence of
ectomycorrhizal trees. Ecol Monogr 67:367–409
Northup RR, Dahlgren RA, McColl JG (1998) Polyphenols as
regulators of plant-litter-soil interactions in northern
California’s pygmy forest: a positive feedback? Biogeochemistry 42:189–220. doi:10.1023/A:1005991908504
Ovington JD, Olson JS (1970) Biomass and chemical content
of El Verde lower montane rain forest plants. In: Odum
HT, Pigeon RF (eds) A tropical rain forest. U.S. Atomic
Energy Commission, Oak Ridge
Perrier N, Amir H, Colin F (2006) Occurrence of mycorrhizal
symbioses in the metal-rich lateritic soils of the Koniambo
massif, New Caledonia. Mycorrhiza 16:449–458. doi:
10.1007/s00572-006-0057-6
Plummer A (2007) Decomposition rates of leaves on two
contrasting soil types in a tropical rainforest—are the
factors that control herbivory persistent and effective in
reducing rates of decomposition? Honours Dissertation,
Monash University
Price ML, Butler LG (1977) Rapid visual estimation and
spectrophotometric determination of tannin content of
sorghum grain. J Agric Food Chem 25:1268–1273. doi:
10.1021/jf60214a034
Proctor J, Woodell SRJ (1975) The ecology of serpentine soils.
Adv Ecol Res 9:255–366. doi:10.1016/S0065-2504
(08)60291-3
Proctor J, Phillipps C, Duff GK, Heaney A, Robertson FM
(1989) Ecological studies on Gunung Silam, a small
ultrabasic mountain in Sabah, Malaysia. II. Some forest
processes. J Ecol 77:317–331. doi:10.2307/2260752
R Development Core Team (2004) R: a language and environment for statistical computing. R Foundation for
65
Statistical Computing, Vienna, Austria. http://www.rproject.org. Accessed 3–6 August 2007
Read J, Hope GS (1996) Ecology of Nothofagus forests of New
Guinea and New Caledonia. In: Veblen TT, Hill RS, Read
J (eds) The ecology and biogeography of Nothofagus
forests. University of Yale Press, New Haven
Read J, Sanson GD (2003) Characterizing sclerophylly: the
mechanical properties of a diverse range of leaf types.
New Phytol 160:81–99. doi:10.1046/j.1469-8137.2003.
00855.x
Read J, Jaffré T, Godrie E, Hope GS, Veillon J-M (2000)
Structural and floristic characteristics of some monodominant and adjacent mixed rain forest in New Caledonia. J
Biogeogr 27:233–250. doi:10.1046/j.1365-2699.2000.
00407.x
Read J, Ferris JM, Jaffré T (2002) Foliar mineral content of
Nothofagus species on ultramafic soils in New Caledonia
and non-ultramafic soils in Papua New Guinea. Aust J Bot
50:607–617. doi:10.1071/BT01091
Read J, Jaffré T, Ferris JM, McCoy S, Hope GS (2006) Does
soil determine the boundaries of monodominant rain forest with adjacent mixed rain forest and maquis on
ultramafic soils in New Caledonia? J Biogeogr 33:1055–
1065. doi:10.1111/j.1365-2699.2006.01470.x
Reich PB, Oleksyn J (2004) Global patterns of plant leaf N and P
in relation to temperature and latitude. Proc Natl Acad Sci
USA 101:11001–11006. doi:10.1073/pnas.0403588101
Richardson SJ, Peltzer DA, Allen RB, McGlone MS (2005)
Resorption proficiency along a chronosequence: responses
among communities and within species. Ecology 86:20–
25. doi:10.1890/04-0524
Santiago LS (2007) Extending the leaf economics spectrum to
decomposition: evidence from a tropical forest. Ecology
88:1126–1131. doi:10.1890/06-1841
Smith SE, Read DJ (1997) Mycorrhizal symbiosis, 2nd edn.
Academic Press, London
Stark N (1970) The nutrient content of plants and soils from
Brazil and Surinam. Biotropica 2:51–60. doi:10.2307/
2989789
Swift MJ, Heal OW, Anderson JM (1979) Decomposition in
terrestrial ecosystems. Blackwell, Oxford
Tanner EVJ (1977) Four montane rain forests of Jamaica: a
quantitative characterization of the floristics, the soils and
the foliar mineral levels, and a discussion of the interrelations. J Ecol 65:883–918. doi:10.2307/2259385
Thompson J, Proctor J, Viana V, Milliken W, Ratter JA, Scott
DA (1992) Ecological studies on a lowland evergreen rain
forest on Maraca Island, Roraima, Brazil. I. Physical
environment, forest structure and leaf chemistry. J Ecol
80:689–703. doi:10.2307/2260860
Torti SD, Coley PD, Kursar TA (2001) Causes and consequences of monodominance in tropical lowland forests.
Am Nat 157:141–153. doi:10.1086/318629
Van Soest PJ (1994) Nutritional ecology of the ruminant, 2nd
edn. Comstock Publications, Ithaca, New York
Villela DM, Proctor J (1999) Litterfall mass, chemistry, and
nutrient retranslocation in a monodominant forest on
Maracá Island, Roraima, Brazil. Biotropica 31:198–211
Villela DM, Proctor J (2002) Leaf litter decomposition and
monodominance in the Peltogyne forest of Maracá Island,
Brazil. Biotropica 34:334–347
66
Vitousek PM (1984) Litterfall, nutrient cycling, and nutrient
limitation in tropical forests. Ecology 65:285–298. doi:
10.2307/1939481
Vitousek PM (1998) Foliar and litter nutrients, nutrient
resorption, and decomposition in Hawaiian Metrosideros
polymorpha. Ecosystems 1:401–407. doi:10.1007/s1002
19900033
A.G. Van der Valk (ed.)
Walsh C, Mac Nally R (2007) The hier.part package.
http://cran.r-project.org/doc/packages/hier.part.pdf
Wilkinson DM, Dickinson NM (1995) Metal resistance in
trees: the role of mycorrhizae. Oikos 72:298–300. doi:
10.2307/3546233
Dendroecological study of a subalpine fir (Abies fargesii)
forest in the Qinling Mountains, China
Haishan Dang Æ Mingxi Jiang Æ Yanjun Zhang Æ
Gaodi Dang Æ Quanfa Zhang
Originally published in the journal Plant Ecology, Volume 201, No. 1, 67–75.
DOI: 10.1007/s11258-008-9491-1 Springer Science+Business Media B.V. 2008
Abstract Dendroecological techniques were used to
investigate the stand dynamics and the disturbance
history of the subalpine fir forest in the Qinling
Mountains of Shaanxi Province, China. The results
indicated that 68% of the fir trees experienced 1–2
release events for a total of 10–29 (an average of 15.8)
years, and 1–2 suppression events for a total of 10–27
(an average of 13.4) years before they reached canopy.
Large number of Abies fargesii and Betula albosinensis recruitment coincided temporally with larger
increases in the ring-width index from the 1830s to
1880s, suggesting occurrence of a major stand-wide
disturbance during this time period. Few seedlings and
saplings were found in the forest, and there was a
dramatic decline in recruitment after 1890, probably
because of the intensive cover of understory umbrella
bamboo (Fargesia spathacea). Radial growth analyses
indicated frequent canopy opening resulting from
small-scale disturbances in the forest. Thus, the
subalpine fir forest experienced frequent small-scale
disturbances and infrequent large-scale disturbances
H. Dang M. Jiang Y. Zhang Q. Zhang (&)
Wuhan Botanical Garden, The Chinese Academy of
Sciences, Wuhan 430074, People’s Republic of China
e-mail: qzhang@wbgcas.cn
H. Dang
e-mail: dangkey@hotmail.com
G. Dang
Foping National Nature Reserve, Foping 723400,
Shaanxi, People’s Republic of China
in its developmental history, and these disturbances
coupled with the understory umbrella bamboo might
have influenced tree growth and species recruitment.
Keywords Growth release Age distribution
Disturbance Subalpine fir forest
The Qinling Mountains
Introduction
Forest’s current structure, composition, and pattern
are influenced by many factors over its developmental
history, including disturbance, competitive interactions between trees, and microsite differences in
resources (North et al. 2004). In fact, stand dynamics
is an important subject of ecological research, and the
studies on stand dynamics and developmental history
have provided a substantial database of information
on the regeneration ecology and population structure
of forests, particularly in relation to canopy gaps
(Namikawa 1996; Abrams et al. 1999). Reconstruction of forest history through identifying growth
releases and tree recruitment is the primary information used to understand the ecosystem processes such
as population dynamics, community structure and
stand development (Taylor and Qin 1988; Taylor et al.
1996; Bergeron et al. 2002).
Over the past decades, dendroecological techniques have become an important tool in the studies
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_6
67
68
of stand dynamics and ecological history (Fritts
1976; Fritts and Swetnam 1989; Nowacki and
Abrams 1994), such as species recruitment patterns,
periodicity and intensity of disturbances, and influences of climatic variation and extreme weather
events (Henry and Swan 1974; Oliver and Stephens
1977; Foster 1988; Abrams and Orwig 1995;
Abrams et al. 1995; Druckenbrod 2005). Tree rings
present valuable and long-term records of tree
growth in forest environments. Variations in treering width can be used to reconstruct the occurrence
of past forest disturbance and to speculate on the
origin of the forest stand (Foster 1988; Lorimer
and Frelich 1989; McClaran and Bartolome 1989;
Veblen et al. 1994; Frelich 2002). As a result,
coupling tree-ring chronologies and age structure
have proven to be a particular robust approach for
understanding long-term variation in forest dynamics and history (Foster 1988; Abrams and Orwig
1995; Abrams et al. 1999).
Disturbances in forested landscapes influence
community structure, composition, and dynamics.
Community response to disturbance varies widely
and depends on the types, size, severity, and
frequency of disturbance and species’ life history
attributes (Taylor et al. 1996). Large-scale disturbances (e.g. severe insect outbreaks, extreme climatic
events, etc.) are usually visible on the mean ringwidth chronology, while detection of medium- and
small-scale disturbances needs more sophisticated
methods of signal analysis (Rathgeber and Roche
2003). The moving average techniques allow us to
derive the disturbance events that are associated with
the medium-frequency signal and remove the highand low-frequency signal (age and climatic effects) in
a chronology (Lorimer and Frelich 1989; Nowacki
and Abrams 1997). Small-scale disturbances regularly cause tree-fall and branch breakage, thus
creating canopy openings that trigger regeneration
and increase of growth of suppressed trees in forests,
i.e. growth release (Lorimer and Frelich 1989; Veblen
et al. 1989; Schweingruber et al. 1990; Cao and
Ohkubo 1999). Growth releases occur synchronously
in neighbouring trees and show a slow decrease in the
following years due to ageing or to closure of canopy
(Veblen et al. 1991). So a disturbance that occurred in
a stand can be identified by the growth releases of
suppressed trees with an accuracy of a few years
(Motta et al. 1999).
A.G. Van der Valk (ed.)
Abies fargesii is a subalpine tree species widely
distributing in the Qinling Mountains of China. It
occurs over a wide range in elevation and dominates
the forests above 2,300 m a.s.l. in the Qinling
Mountains. The subalpine fir forests remain undisturbed by human activities for more than a century in
the Qinling Mountains (Zhang 1989). Thus, these
forests represent a rare opportunity to study forest
dynamics and developmental history for the subalpine conifer forests. However, a very little is known
about the stand dynamics and developmental history
of the subalpine fir forests in the Qinling Mountains,
a biodiversity hotspot in China, due to insufficient
research (Taylor and Qin 1988; Taylor et al. 1996).
Understanding the forest dynamics and developmental history of the subalpine conifer forests is of great
importance for the sustainable forestry.
In this study, we report the forest history and stand
development of the subalpine fir forests in the Qinling
Mountains of Shaanxi province, China. The primary
objectives of this study are: (1) to describe the
composition and structure of the subalpine fir forest;
(2) to investigate the long-term patterns of species
recruitment and radial growth variation; (3) to
reconstruct the disturbance history of the forest using
dendroecological evidence.
Methods
Study area
The study area is located in the Foping National
Nature Reserve in the south aspect of the Qinling
Mountains of Shaanxi Province, China (E107490 ,
N33420 ) (Fig. 1). The Qinling Mountains run East–
West and form the basin divider between China’s two
longest rivers, the Yellow River and the Yangtze
River. The Qinling Mountains are situated in the
transitional zone between two macroclimatic regimes
(subtropical and warm-temperate zones), making it
biologically rich area and sensitive to climatic change
in China (Chen 1983). Elevation in the study area
ranges from 980 to 2,838 m. The climate is characterized as subtropical with mild, wet summers and
cold, dry winters. Annual precipitation ranges from
950 to 1,200 mm, most of which falls between July
and September. Snow cover usually lasts five or more
months (from November to March), and annual mean
Forest Ecology
69
interior at the elevations of 2,360 and 2,450 m,
respectively, which were orientated parallel to the
isoline for a total length of 100 m. These transects
were selected based on the criteria that there were
similar habitats and species composition within and
among transects and that the stands should represent
the forest structure and composition in the study area.
Each transect was divided into five 20 9 20 m
plots. Within each plot, all living trees (DBH C 5 cm,
measured at 1.37 m above ground) were labeled,
counted and mapped with respect to a reference tree.
Species, height, crown projection area and DBH were
recorded. Increment cores were taken in the direction
parallel to the slope contour using increment borers at
the breast height. Trees of abnormal growth form
(bent, twisted or hollow) were excluded from sampling. Usually one core per tree was taken, but for a
few trees with unusable cores, additional cores were
collected until a usable core was obtained. In total,
197 trees were cored (118 for A. fargesii, 46 for
B. albo-sinensis, and 33 for the other species).
Saplings and seedlings were counted by species in
five 2 9 2 m quadrats within each of the 20 9 20 m
plots. The cover of umbrella bamboo (F. spathacea)
was estimated in the 2 9 2 m quadrats.
Fig. 1 Location of the study site in the Qinling Mountains of
Shaanxi Province, China
temperature ranges from 6 to 11C below 2,000 m
and from 1 to 6C above 2,000 m a.s.l. (Yue et al.
1999).
Conifers dominated by subalpine fir (Abies
fargesii) occupy the area above 2,300 m in elevation,
usually developing into mixed forests with birch
(Betula albo-sinensis) or forming pure conifer forests
in the study area. Umbrella bamboo (Fargesia
spathacea) is a common understory species above
2,300 m. The mixed conifer and deciduous forests
occur between elevations of 1,800 and 2,300 m.
Patchy subalpine meadow also occurs above 2,600 m
a.s.l. in this study area (Yue et al. 1999). Because of
the limited accessibility, there have been no visible
human activities in the region.
Field survey
In the summer of 2005, two transects of 20 m wide
were established through the subalpine fir forest
Radial growth analysis
All increment cores were air-dried, mounted, and
sanded. The Abies cores were cross-dated using the
signature year technique to identify the missing,
partial, or false rings (Yamaguchi 1991). For cores,
where the pith was missed, a graphical procedure was
used to estimate their age (Villalba and Veblen 1997).
Annual growth increments were measured to the
nearest 0.01 mm with a tree ring-measuring device
(Regent Instruments Inc., Quebec, Canada). The
cores of B. albo-sinensis and other species were only
used for age estimations. Recruitment was defined as
the date when trees reached 1.37 m in height.
The percentage growth change filter (Lorimer
and Frelich 1989; Nowacki and Abrams 1997)
was applied to the fir tree ring series. The formulae %GCr = [(M2 - M1)/M1] 9 100 and %GCs =
[(M1 - M2)/M2] 9 100 were used to identify growth
release and suppression, respectively, where M1 and
M2 are the preceding and subsequent 15-year ringwidth mean, respectively. The minimum threshold of
100% for %GC was considered for event recognition,
70
A.G. Van der Valk (ed.)
and the identified periods of release and suppression
had to be at least 10 years to eliminate annual
variations of radial growth. The criterion, coupled
with tree canopy recruitment dates, was used to
distinguish disturbance events from responses attributed to climatic factors (Lorimer and Frelich 1989;
Abrams and Orwig 1996).
After cross-dating using signature years and the
quality control program COFECHA (Holmes 1983),
the raw ring widths from the oldest 30 fir cores were
indexed using a detrending filter with a 10-year
window (Guiot and Goeury 1996), and a mean
indexed series was constructed. A site chronology
was then developed using the oldest 30 fir cores. In
addition, the tree-ring series at decadal intervals were
used to reconstruct the disturbance history for the
subalpine fir forest (Motta et al. 1999; Abrams et al.
2001; Bergeron et al. 2002; Druckenbrod 2005).
represent a combined relative importance value of
23%. The total density and basal area are 520 stems/
ha and 68.31 m2/ha, respectively. Tree regeneration
was very sparse within the stand. Only 11 seedlings
and 6 saplings, mainly Acer maximowiczii and
Prunus tomentosa, were counted in the fifty
2 9 2 m quadrats (data not shown). Umbrella bamboo (Fargesia spathacea) with a height of 1.5–3.0 m
averaged 95% cover in the understory.
The diameter-class structure showed a bell-shape
distribution (Fig. 2). Abies fargesii occurred in all
Results
Composition and size structure
The forest is comprised of nine tree species and is
dominated by A. fargesii (47.86% importance value),
B. albo-sinensis (19.66%), and Acer maximowiczii
(9.13%) (Table 1). The six remaining species
Fig. 2 Diameter (at 1.37 m above ground) distribution of tree
species of the subalpine fir (Abies fargesii) forest in the Qinling
Mountains of Shaanxi province, China. Other species include
Abelia macroptera, Sorbus koehneana, and Corylus tibetica
Table 1 Species composition for the ten plots of the subalpine fir (Abies fargesii) forest in the Qinling Mountains of Shaanxi
province, China
Species
Frequency
Density
(stems/ha)
Dominance
(m2/ha)
Relative
frequency
Relative
density
Relative
dominance
Importance
value
Abies fargesii
10
317.5
40.85
22.73
61.06
59.80
47.86
Betula albosinensis
8
115.0
12.75
18.18
22.12
18.67
19.66
Acer maximowiczii
6
22.5
6.44
13.64
4.33
9.43
9.13
Maddenia
hypoxantha
5
17.5
5.76
11.36
3.37
8.44
7.72
Prunus tomentosa
5
20.0
2.14
11.36
3.85
3.14
6.12
Crataegus
kansuensis
4
12.5
0.27
9.09
2.40
0.40
3.96
Abelia macroptera
3
7.5
0.04
6.82
1.44
0.06
2.77
Sorbus koehneana
2
5.0
0.03
4.55
0.96
0.04
1.85
Corylus tibetica
Totals
1
2.5
0.01
2.27
0.48
0.02
0.92
44
520.0
68.31
100
100
100
100
Frequency is based on presence in the 10 sampled plots; density is based on the number of individuals; and dominance is based on
basal area. Importance value is the average of relative frequency, relative density, and relative dominance (Cottam and Curtis 1956)
Forest Ecology
71
Table 2 Summary of growth release and suppression of canopy fir trees of the subalpine fir forest in the Qinling Mountains
of Shaanxi province, China (n = 118)
Mean SD
Range
Release
Numbers of release periods per core
Years of each release period
1.1
Years of each suppression period
11.9
2.7 10–24
Years of suppression per core
13.4
4.8 10–27
Ages when the first suppression was 60.5
identified
0.3
1–2
32.7 15–119
(ranging from 15 to 124) and 60.5 (ranging from 15
to 119) years, respectively.
Growth releases occurred frequently over the past
240 years in the fir trees except the time periods of the
1770s to 1790s and the 1810s (Fig. 4). The number of
trees showing releases was fairly low prior to 1840
because of the scarcity of fir trees greater than
180 years old. The number of trees showing releases
each decade had generally remained constant after
1840, with the largest number in the 1940s and 1960s.
Dendroecology
Fir trees had an average growth rate of 1.14 mm/year,
ranging from 0.11 mm/year to 3.67 mm/year. The
160
Total number of
cores
140
Number of cores
showing releases
120
100
80
60
40
1990s
1970s
1950s
1930s
1910s
1890s
0
1870s
20
1850s
The radial growth analysis indicated that 68% of the
canopy fir trees (n = 118) showed periods of release
and suppression. These trees experienced 1–2 (an
average of 1.2) releases for 10–29 (an average of
15.8) years, and 1–2 (an average of 1.1) suppression
for 10–27 (an average of 13.4) years (Table 2). There
were average of 12.8 and 11.9 years for release and
suppression period, respectively. The average ages
for the first release and suppression were 45.2
Suppression
Numbers of suppression periods per
core
1830s
Release and suppression
5.8 10–29
32.7 15–124
1810s
Based on the sample of 197 cored trees, the forest is
uneven-aged (Fig. 3). Six fir trees are older than
200 years (239 years maximum at 1.37 m above
ground). Long tree-ring series also exist for B. albosinensis (205 years old). A large number of Abies
fargesii trees were recruited from the 1830s to 1880s
with peak number in the 1840s. Few fir trees were
recruited since the 1890s, and none after the 1950s. A
large number of Betula albo-sinensis were recruited
from the 1790s to 1880s, while very few after the
1880s. All recruitment of the rest species occurred in
the 20th century with a few exceptions.
15.8
45.2
1790s
Age structure
1–2
Ages when the first release was
identified
1770s
diameter classes and dominated the middle and larger
diameter classes from 31 to 70 cm. In contrast,
Betula albo-sinensis was in the relatively smaller
diameter classes from 11 to 40 cm, and most of the
remaining species were represented in the smaller
diameter classes less than 20 cm.
0.4
3.3 10–29
Years of release per core
Number of cores
Fig. 3 Age class (at 1.37 m above ground) distribution of tree
species of the subalpine fir forest (Abies fargesii) in the Qinling
Mountains of Shaanxi province, China. Other species include
Abelia macroptera, Sorbus koehneana, and Corylus tibetica
1.2
12.8
Period (10 years)
Fig. 4 Number of cores showing releases and total number of
cores analyzed at 10-year intervals of the subalpine fir forest in
the Qinling Mountains of Shaanxi province, China
72
A.G. Van der Valk (ed.)
1.6
Ring-Width Index
1.4
1.2
1
0.8
0.6
2000
1980
1960
1940
1920
1900
1880
1860
1840
1820
1800
1780
0.4
Year
Fig. 5 Site chronology of Abies fargesii of the subalpine fir
forest in the Qinling Mountains of Shaanxi province, China
ring-width index chronology for the oldest 30 fir trees
indicated a period of generally high radial growth
between 1836 and 1910 with a few exceptions
(Fig. 5). Increases in radial growth also occurred in
the time period of 1936–1954. The radial growth was
Fig. 6 Individual tree-ring
chronologies and release
dates for six selected cores
of the subalpine fir forest in
the Qinling Mountains of
Shaanxi province, China.
Arrows indicate release
dates
substantially below average for most of the time
period before 1835, and from 1911 to 1935 and 1955
to 1994.
Figure 6 illustrated variation in growth patterns
and frequency of release dates of the representative
fir trees. The timing of the releases reflected great
variation among the individual trees, i.e. the releases
were present only in individual cores (Fig. 6). The
trees in Fig. 6a, c showed high early growth followed
by a steep growth decline, characteristics of trees
with gap origin followed by gap closure. The fir tree
in Fig. 6b showed slow early growth followed by
progressively higher growth before a dramatic
increase of its growth rate in 1811, and then a peak
growth was recognized after the release in 1860. The
fir tree in Fig. 6d showed fairly consistent growth for
more than 150 years before its growth began to
decrease in 1938. The tree in Fig. 6f had major early
growth, followed by a great decrease in growth for
Forest Ecology
the next 60 years, and its growth rate drastically
increased after a release in 1874, and then it
experienced another release in 1914.
Discussion
The results of this study suggest that the existing fir
and birch trees were primarily recruited in the
subalpine fir forest between 1830 and 1890 in these
plots on the Qinling Mountains. Prior to 1890, the
forest was comprised of a large number of A. fargesii,
B. albo-sinensis, and Acer maximowiczii, based on
the current stand age distribution (Fig. 3). However,
care must be taken, when explaining static age
structure data because of difference in species
mortality with various age and canopy classes and
stand-history events (Johnson et al. 1994; Abrams
et al. 2001). The oldest and the largest trees in our
study sites are Abies species and the domination of
the uneven-aged forest by Abies apparently dated
back 200 years ago. The absence of A. fargesii trees
older than 250 years in the subalpine fir forest may be
attributable to their maximum life expectancy.
Prunus tomentosa, Maddenia hypoxanth, Crataegus
kansuensis, and the other species are companion
species and have only invaded the forest recently.
A unique dendroecological feature of the
A. fargesii stand is the association of the large
recruitment pulse with high radial growth observed in
the ring-width index chronology between 1830 and
1890 (Figs. 3 and 5). The forests were logged several
times during the time period from the 1790s to 1870s,
and most of the old trees were selectively cut in those
logging activities (Zhang 1989). Logging may have
greatly altered the composition and number of tree
recruits in an old-grown forest (Abrams and Nowacki
1992; Nowacki and Abrams 1994; Orwig and
Abrams 1999; Abrams et al. 2001). The several
logging events may cause the sustained period of tree
recruitment and the several increases in the ringwidth index chronology from 1830 to 1890, culminating with a very large pulse around 1850 (Figs. 3
and 5). Prior to 1830, although A. fargesii recruitment
in the subalpine fir forest was also associated with
releases in the ring-width index chronology, e.g.
1789–1792, 1800–1802, and 1824–1827, the association between recruitment and high radial growth of
fir trees might not be substantial due to the
73
insufficient number of fir trees (Figs. 3 and 5).
Catastrophic fire has been described as a pervasive
disturbance in the coniferous forests of China (Wang
1961), while there are no recently burned stands in
our study sites. Whether, natural disturbance events
such as windstorms (Ren 1998) stimulated as much
tree recruitment as the nineteenth logging is
unknown, because of mortality in older fir trees.
Releases in radial growth indicate occurrence of
disturbances almost over the developmental history
of the fir forest (Fig. 4). While the 1770s to 1790s
period without growth releases would perhaps be the
stem exclusion stage for the fir forest (Oliver and
Larson 1996). A series of release episodes are
identified, but a very few of these releases are
recorded in the same year (Figs. 4 and 6), indicating
that the forest experienced frequent small-scale
disturbances (such as tree-fall and branch breakage
caused by windstorms, etc.) each of which impacted a
relatively small number of trees in the stand development history (Abrams et al. 2001). Only a few
major stand-wide disturbance events (such as logging
activities, etc.) occurred in the subalpine fir forest
according to the largest pulse in tree recruitment and
the largest pulse in ring-width index chronology
around the 1850s (Figs. 3 and 5). Thus, the subalpine
forest experienced frequent small-scale disturbances
and infrequent large-scale disturbances in its developmental history.
Tree recruitment after 1890 became infrequent,
excluding a small pulse of recruitments between 1920
and 1960 (Fig. 3). Perhaps the intensive cover of
understory umbrella bamboo (Fargesia spathacea) is
responsible for the low frequency of seedling establishment. Bamboos, which are common understory
plants in temperate and tropical forests, appear
particularly effective in reducing tree regeneration
where they achieve a high degree of dominance
(Taylor and Qin 1988; Holz and Veblen 2006). In
other subalpine forests with dense bamboo stands in
China, Japan and South America (Franklin et al.
1979; Koyama 1984; Taylor and Qin 1988;
Nakashizuka 1991; Taylor and Qin 1992; Taylor
et al. 1996; Holz and Veblen 2006), bamboos with a
50% cover seem sufficiently to impede tree establishment and forest gaps fill slowly, and most
regeneration of tree species occurs in canopy gaps
caused by disturbances and such regeneration is
sparse. In our study sites, the umbrella bamboo with a
74
height of 1.5–3.0 m and a density of 70–140 stems/
m2 usually covers more than 85%, sometimes 100%
of the understory of the subalpine fir forest (Ren
1998). As a result, it seems that regeneration of the A.
fargesii population depends on the occurrence of
large-scale disturbances in the Qinling Mountains
(Ren 1998) or in areas of umbrella bamboo dieback
(Taylor and Qin 1992; Namikawa 1996).
Patterns of tree recruitment in the subalpine fir
forests of the Qinling Mountains appear to be similar
to other subalpine forests in China, Japan and South
America (Franklin et al. 1979; Koyama 1984; Taylor
and Qin 1988; Nakashizuka 1991; Taylor and Qin
1992; Taylor et al. 1996; Holz and Veblen 2006). In
these subalpine forests, bamboos reduce seedlings
establishment and stands typically have few seedlings
and saplings. Rapid vegetative growth of bamboos
reduces space for other species within the community, and small-scale disturbances induce most trees’
recruitment. Mass death of the bamboo understory
following synchronized flowering is a large-scale
disturbance, which increases space-related resources,
especially light, that may allow for pulses of tree
recruitment, otherwise tree recruitment is inhibited by
the bamboo cover (Taylor and Qin 1988, 1992; Hiura
et al. 1996; Abrams et al. 1999). However, there are
no reports on bamboo die-off in the study area, which
might contribute to the patterns of tree recruitment
observed in this study. If the intensity of umbrella
bamboo cover on tree regeneration does not lessen in
the future, we anticipate very little opportunity for
canopy recruitment in the subalpine fir forest in the
Qinling Mountains.
Conclusions
The majority of the subalpine fir trees (68%) in the
Qinling Mountains of China experienced 1–2 times
of growth release and suppression before they
reached canopy. A large number of A. fargesii and
B. albo-sinensis were recruited from the 1830s to
1880s. In combination with frequent releases from
radial growth analyses, it seems that tree growth and
species recruitment have been influenced by the
coupling of frequent small-scale disturbances and
infrequent large-scale disturbances. There are very
few seedlings and saplings in the forest, and canopy
recruitment declined after 1890, probably as a result
A.G. Van der Valk (ed.)
of the dense cover of understory umbrella bamboo
(95%) which prevents tree recruitment in the
subalpine fir forest.
Acknowledgments The authors sincerely thank comments
and suggestions from Dr. Jianqing Ding and two anonymous
reviewers. We thank Mr. Xinping Ye for help on fieldwork. We
also acknowledge Mr. Xu Pang and Ms. Xiuxia Chen for
assistance with tree core processing. This research was
supported by the ‘‘Hundred-Talent Project’’ of the Chinese
Academy of Sciences (0629221C01) and the Kuancheng Wang
Education Foundation of Hong Kong.
References
Abrams MD, Nowacki GJ (1992) Historical variation in fire,
oak recruitment, and post-logging accelerated succession in central Pennsylvania. Bull Torrey Bot Club 119:
19–28
Abrams MD, Orwig DA (1995) Structure, radial growth
dynamics and recent climatic variations of a 320-year-old
Pinus rigida rock outcrop community. Oecologia
101:353–360
Abrams MD, Orwig DA (1996) A 300-year history of disturbance and canopy recruitment for co-occurring white pine
and hemlock on the Allegheny Plateau USA. J Ecol
84:353–363
Abrams MD, Orwig DA, DeMeo TE (1995) Dendroecological
analysis of successional dynamics for a presettlementorigin white-pine-mixed-oak forest in the southern
Appalachians, USA. J Ecol 83:123–133
Abrams MD, Copenheaver CA, Terazawa K, Umeki K,
Takiya M, Akashi N (1999) A 370-year dendroecological history of an old-growth Abies-Acer-Quercus forest in
Hokkaido, northern Japan. Can J For Res 29:1891–1899
Abrams MD, Copenheaver CA, Black BA, van de Gevel S
(2001) Dendroecology and climatic impacts for a relict,
old-growth, bog forest in the Ridge and Valley Province
of central Pennsylvania USA. Can J Bot 79:58–69
Bergeron Y, Denneler B, Charron D, Girardin MP (2002)
Using dendrochronology to reconstruction disturbance
and forest dynamics around Lake Duparquet, northwestern Quebec. Dendrochronologia 20:175–189
Cao KF, Ohkubo T (1999) Suppression and release during
canopy recruitment in Fagus crenata and Acer mono in
two old-growth beech forests in Japan. Plant Ecol
145:281–290
Chen MR (1983) Climate and agriculture in the Qinling
Mountains. Shaanxi People’s Press, Xi’an (in Chinese)
Cottam G, Curtis JT (1956) The use of distance measures in
phytosociological sampling. Ecology 37:451–460
Druckenbrod DL (2005) Dendroecological reconstructions of
forest disturbance history using time-series analysis with
intervention detection. Can J For Res 35:868–876
Foster DR (1988) Disturbance history, community organization and vegetation dynamics of the old-growth Pisgah
Forest, south-western New Hampshire USA. J Ecol 76:
105–134
Forest Ecology
Franklin JF, Maeda T, Oshumi T, Matsui M, Yagia H, Hawk
GM (1979) Subalpine coniferous forests of central Honshu, Japan. Ecol Monogr 49:311–334
Frelich LE (2002) Forest dynamics and disturbance regimes:
studies from temperate evergreen-deciduous forests.
Cambridge University Press, New York
Fritts HC (1976) Tree Rings and Climate. Academic Press,
New York
Fritts HC, Swetnam TW (1989) Dendroecology: a tool for
evaluating variations in past and present forest environments. Adv Ecol Res 19:111–188
Guiot J, Goeury C (1996) PPPBASE, a software for statistical
analysis of paleoecological and paleoclimatological data.
Dendrochronologia 14:295–300
Henry JD, Swan JMA (1974) Reconstructing forest history
from live and dead plant material—an approach of the
study of forest succession in southwest New Hampshire.
Ecology 55:772–783
Hiura T, Sano J, Konno Y (1996) Age structure and response to
fine scale disturbance of Abies sachalinensis, Picea jezoensis, Picea glehnii, and Betula ermanii growing under
the influence of a dwarf bamboo understory in northern
Japan. Can J For Res 26:289–297
Holmes RL (1983) Computer-assisted quality control in treering dating and measurement. Tree-ring Bulletin 43:69–87
Holz CA, Veblen TT (2006) Tree regeneration responses to
Chusquea montana bamboo die-off in a subalpine Nothofagus forest in the southern Andes. J Veg Sci 17:19–28
Johnson EA, Miyanishi K, Kleb H (1994) The hazards of
interpretation of static age structures as shown by stand
reconstructions in a Pinus contorta-Picea engelmannii
forest. J Ecol 82:923–931
Koyama T (1984) Regeneration and coexistence of two Aibes
species dominating subalpine forests in central Japan.
Oecologia 62:156–161
Lorimer CG, Frelich LE (1989) A methodology for estimating
canopy disturbance frequency and intensity in dense
temperate forests. Can J For Res 19:651–663
McClaran MP, Bartolome JW (1989) Fire-related recruitment
in stagnant Quercus douglasii populations. Can J For Res
19:580–585
Motta R, Nola P, Piussi P (1999) Structure and stand development in three subalpine Norway spruce (Picea abies
(L.) Karst) stands in Paneveggio (Trento, Italy). Glob Ecol
Biogeogr 8:455–471
Nakashizuka T (1991) Population dynamics of coniferous and
broad-leaved trees in a Japanese temperate mixed forest.
J Veg Sci 2:413–418
Namikawa K (1996) Stand dynamics during a 12 year period in
an old-growth, cool temperate forest in northern Japan.
Ecol Res 11:23–33
North M, Chen J, Oakley B, Song B, Rudnicki M, Gray A,
Innes J (2004) Forest stand structure and pattern of oldgrowth western Hemlock/Douglas-fir and mixed-conifer
forests. For Sci 50:299–311
Nowacki GJ, Abrams MD (1994) Forest composition, structure, and disturbance of the Alan Seeger Natural Area,
75
Huntingdon County, Pennsylvania. Bull Torrey Bot Club
121:277–291
Nowacki GJ, Abrams MD (1997) Radial-growth averaging
criteria for reconstructing disturbance histories from presettlement-origin oaks. Ecol Monogr 67:225–249
Oliver CD, Stephens EP (1977) Reconstruction of a mixed
forest in central New England. Ecology 58:562–572
Oliver CD, Larson BC (1996) Forest Stand Dynamics (update
edition). John Wiley and Sons Inc, New York
Orwig DA, Abrams MD (1999) Impacts of early selective
logging on the dendroecology of an old-growth, bottomland hemlock-white pine-northern hardwood forest on the
Allegheny Plateau. J Torrey Bot Soc 126:234–244
Rathgeber C, Roche P (2003) Spatio-temporal growth
dynamics of a subAlpine Pinus uncinata stand in the
French Alps. C R Biol 326:305–315
Ren Y (1998) Vegetation of Giant Panda’s Habitat of the
Qinling Mountains. Shaanxi Science and Technology
Press, Xi’an (in Chinese)
Schweingruber FH, Eckstein D, Serre-Bachet F, Braker OU
(1990) Identification, presentation and interpretation of
event years and pointer years in dendrochronology. Dendrochronologia 8:9–39
Taylor AH, Qin ZS (1988) Regeneration patterns in old-growth
Abies-Betula forests in the Wolong Natural Reserve,
Sichuan, China. J Ecol 76:1204–1218
Taylor AH, Qin ZS (1992) Tree regeneration after bamboo dieback in Chinese Abies-Betula forests. J Veg Sci 3:253–
260
Taylor AH, Qin ZS, Liu J (1996) Structure and dynamics of
subalpine forests in the Wang Lang Natural Reserve, Sichaun, China. Vegetatio 124:25–38
Veblen TT, Hadley KS, Reid MS, Rebertus AJ (1989) Blowdown and stand development in a Colorado subalpine
forest. Can J For Res 19:1218–1225
Veblen TT, Hadley KS, Reid MS (1991) Disturbance and stand
development of a Colorado subalpine forest. J Biogeogr
18:707–716
Veblen TT, Hadley KS, Nel EM, Kitzberger T, Reid MS,
Villalba R (1994) Disturbance regime and disturbance
interactions in a Rock Mountains subalpine forest. J Ecol
82:125–135
Villalba R, Veblen TT (1997) Improving estimates of total tree
ages based on increment cores samples. EcoSceince
4:535–542
Wang CW (1961) The Forests of China with a Survey of
Grassland and Desert Vegetation. Maria Moors Cabot
Foundation, Harvard University, Publication Series 5
Yamaguchi DK (1991) A simple method for cross-dating
increment cores from living trees. Can J For Res 21:414–
416
Yue M, Ren Y, Dang GD, Gu TQ (1999) Species diversity of
higher plant communities in Foping Nature Reserve.
Chinese Biodiversity 7:263–369 (in Chinese with English
abstract)
Zhang YQ (1989) Shaanxi Forest. Xi’an Science and Technology Press, Xi’an (in Chinese)
A conceptual model of sprouting responses in relation to fire
damage: an example with cork oak (Quercus suber L.) trees
in Southern Portugal
Francisco Moreira Æ Filipe Catry Æ Inês Duarte Æ
Vanda Acácio Æ Joaquim Sande Silva
Originally published in the journal Plant Ecology, Volume 201, No. 1, 77–85.
DOI: 10.1007/s11258-008-9476-0 Springer Science+Business Media B.V. 2008
Abstract The sprouting response types of 1,151
cork oak (Quercus suber) trees one and half years
after a wildfire in southern Portugal were characterised. It was hypothesised that different response types
should occur according to the following conceptual
model: an increased level of damage (fire severity) on
a sprouting tree that suffered a crown fire was
expected to be reflected in a sequence of four
alternative events, namely (a) resprouting exclusively
from crown, (b) simultaneous resprouting from crown
and base, (c) resprouting exclusively from base and
(d) plant death. To assess whether the level of
expected damage was influenced by the level of
protection from disturbance, we explored the relationships between response types and tree size, bark
thickness and cork stripping, using an informationtheoretic approach. The more common response type
was crown resprouting (68.8% of the trees), followed
by plant death (15.8%), simultaneous resprouting
from crown and base (10.1%) and basal resprouting
(5.3%). In agreement with the conceptual model,
trees which probably suffered a higher level of
damage by fire (larger trees with thinner bark;
exploited for cork) died or resprouted exclusively
F. Moreira (&) F. Catry I. Duarte V. Acácio
J. S. Silva
Centro de Ecologia Aplicada ‘‘Prof. Baeta Neves’’,
Instituto Superior de Agronomia, Universidade Técnica
de Lisboa, Tapada da Ajuda, 1349-017 Lisbon, Portugal
e-mail: fmoreira@isa.utl.pt
from base. On the other hand, trees that were well
protected (smaller trees with thicker bark not
exploited for cork) were able to rebuild their canopy
through crown resprouting. Simultaneous resprouting
from the crown and base was determined mainly by
tree size, and it was more common in smaller trees.
Keywords Apical dominance Mediterranean
Model Mortality Resource allocation
Resprouting Severity
Introduction
Resprouting is an efficient mechanism through which
many plants from the Mediterranean region recover
above-ground biomass after they have suffered total
crown consumption from a wildfire (Whelan 1995;
Bond and van Wilgen 1996; Keeley 2006). Sprouting
shoots can originate from dormant buds located
above ground (axillary, branch epicormic or stem
epicormic) or from the base of the plant (i.e. from the
collar, roots or underground stems) (Bond and van
Wilgen 1996; Miller 2000; Del Tredici 2001).
Hereafter, these two sprouting modes will be referred
to as ‘crown’ and ‘basal’ sprouting (Bond and van
Wilgen 1996).
Bellingham and Sparrow (2000) presented a
general model of resprouting responses as a function
of increasing disturbance severity (severity defined as
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_7
77
78
a measure of a plant0 s perception of a disturbance
event). This gradient of increasing severity was
expected to create a sequence of hierarchical regenerative responses ranging from crown (e.g. axillary
and branch epicormic) to basal sprouting, where the
loss of one type of tissue (e.g. in twig) induces a
regenerative response from the next level of hierarchy (e.g. twig axil on the branch) (Bellingham and
Sparrow 2000). In their model, disturbance severity is
expressed as proportion of above-ground biomass lost
(Bellingham and Sparrow 2000). For one particular
type of disturbance, wildfires, and in particular crown
fires, often all the canopy foliage, buds and twigs are
consumed (crown consumption). When this happens,
severity will depend mainly on the fire intensity and
the level of fire-protection mechanisms at the individual level (e.g. Bond and van Wilgen 1996).
Although the hierarchical nature of sprouting
responses presented in Bellingham and Sparrow0 s
model could also be expected in this situation, the
fact that in a few studies sprouting responses at
different hierarchical levels were simultaneously
registered in the same individual plant (Trollope
1984, current study) suggests that the factors underlying response types will be more complex than just
above-ground biomass lost.
In situations where wildfires caused total crown
consumption in sprouting trees we allege that disturbances of differing levels of damage (severity), and
corresponding sprouting responses, not necessarily
organised as an hierarchical model, can still be
recognised. These responses will be determined by
the amount of bud damage in the twigs and branches,
the level of damage to stem and root cambial tissue
and the amount of below-ground reserves which
determines how much carbohydrate reserves can be
mobilised to rebuild the lost biomass (Chapin et al.
1990; Bond and van Wilgen 1996; Iwasa and Kubo
1997; Bellingham and Sparrow 2000) (Fig. 1). When
the level of fire damage is low (e.g. caused by low fire
intensity on trees with thicker bark, and where the
stem cambium is not affected), the plant is expected
to resprout from crown buds that survived the fire
(Fig. 1a). If the level of damage is extreme (e.g.
caused by high fire intensity on trees with thinner
bark or where the stem cambium is damaged), the
most likely outcome is plant death (Fig. 1d). At
intermediate levels of severity two response types can
be identified. If the level of damage is higher, all
A.G. Van der Valk (ed.)
a
b
c
d
Level of damage / fire severity
Bud
damage
Cambial tissue
damage
Below-ground
reserves
Fig. 1 A conceptual model of post-fire responses of a
sprouting tree that suffered total crown consumption (combustion of leaves and twigs during a wildfire) in relation to a
gradient of increasing level of damage/fire severity. (a) Crown
sprouting, (b) simultaneous sprouting from crown and base, (c)
basal sprouting, (d) plant death (for further explanations see
text)
crown buds will be killed, either directly through heat
or indirectly through the destruction of the vascular
cambium in the stem, as the carbohydrate reserves
that support sprouting are primarily stored in belowground structures (Del Tredici 2001). Furthermore,
apical dominance will be suppressed directly through
bud destruction by heat or indirectly via damage to
the cambium (Kozlowski 1971; Kozlowski et al.
1991; Miller 2000), and the tree is therefore expected
to respond through basal resprouting (Fig. 1c).
Alternatively, if the level of damage is not so severe,
partial damage to the crown buds and cambium will
cause weakened apical dominance (Kozlowski 1971)
and at least some accessibility to below-ground
reserves, thereby resulting in the simultaneous resprouting of the crown and base (Fig. 1b). Since the
amount of carbohydrate, nitrogen and phosphorus
resources that can be used for growth also determines
the extent to which plants can resprout (Chapin et al.
1990), the observed resprouting patterns will therefore also be influenced, and plants with depleted
below-ground resources may suffer higher levels of
damage since they are unable to allocate enough
energy to restore the lost biomass. An example of
these above-mentioned four types of responses can be
found in a study of Acacia karroo savanna by
Trollope (1984), where different responses were
related to tree size and fire intensity. However, no
other examples were found in the literature where the
Forest Ecology
occurrence of different responses was registered and
characterised for other tree species.
The cork oak Quercus suber L. is a very important
tree species within the Mediterranean basin, both
from an economic and ecological perspective (Silva
and Catry 2006). The existence of a thick cork bark
plays an important role in the capacity of this species
to withstand the frequent occurrence of fire typical of
Mediterranean climates (e.g. Pausas 1997; Moreira
et al. 2007). Another feature of cork oak trees is the
capacity of post-fire resprouting from the base and
crown after complete defoliation, hence the species is
a good model for studying the different response
patterns previously described.
In general, there is little information available on
the relative frequency of the different response types
as well as the factors influencing these responses in
cork oak. Previous studies (e.g. Cabezudo et al. 1995;
Pausas 1997; Barberis et al. 2003; Catry et al. 2007;
Moreira et al. 2007) focused mainly on the factors
influencing post-fire survival, and showed the key
role of cork stripping, cork thickness and tree size on
determining oak survival. In this article, we hypothesise that these three variables also influence other
post-fire response patterns (as described in Fig. 1)
besides death, since they are expected to influence the
level of resistance to fire and, consequently, the level
of damage.
Cork stripping is a common operation that is
normally performed after the tree attains a certain
circumference at breast height (70 cm in Portugal).
Cork is a valuable raw material for industry and is
periodically removed with an axe by manually
cutting along vertical and horizontal lines on the
stem and thicker branches and stripping off cork
planks (Pereira and Tomé 2004). After each cork
stripping, the tree has the capacity to produce new
cork bark by adding new layers of cork every year
(Pereira and Tomé 2004), Moreira et al. (2007)
showed that unstripped trees (with unharvested virgin
cork) had higher survival rates than trees that had
been exploited for cork (i.e. trees debarked at least
once). These authors suggested that the higher
survival rates of unstripped trees may be explained
by the higher insulating properties of virgin cork (for
a given bark thickness) and the absence of stress
caused by cork extraction. In fact, cork extraction is a
disturbance that has negative effects on tree health
and growth (Costa et al. 2004). Thus, stripping
79
probably requires a greater allocation of belowground energy reserves that will subsequently not be
available for investment in resprouting. Consequently, unstripped trees are expected to show
lower levels of damage when compared to exploited
trees since their buds are more protected and their
below-ground reserves may be better preserved.
Cork thickness depends on the harvesting cycle and
the time elapsed between harvesting events. Cork can
only be harvested every 9–15 years (minimum 9 years
according to Portuguese legislation), and several
studies have shown that cork age (and thus thickness)
is inversely related to post-fire mortality (e.g. Lamey
1893; Pampiro et al. 1992; Cabezudo et al. 1995;
Pausas 1997; Barberis et al. 2003; Catry et al. 2007;
Moreira et al. 2007). The thicker the bark, the lower the
expected level of post-fire damage (again, buds and
cambium are more protected from fire).
Barberis et al. (2003) and Moreira et al. (2007)
provided evidence that trees with larger diameter at
breast height (DBH) had a lower probability of
survival. Possible explanation for this pattern include
a likely higher amount of stripping damages, higher
susceptibility to stress or diseases and higher frequency
of poor management practices (e.g. deep ploughing,
excessive pruning) in older trees (Costa et al. 2004;
Moreira et al. 2007). A bigger tree that has suffered
several damage events across its lifespan is therefore
prone to higher levels of post-fire damage, mainly
because of the lack of carbohydrate reserves to invest in
resprouting (Iwasa and Kubo 1997).
The aim of this article is to explore the importance
of tree size, bark thickness and cork stripping in
determining the whole range of post-fire response
types in cork oak. In particular we aimed to: (a)
quantify the relative frequency of four different postfire responses in burned cork oak trees 1.5 years after
an intense wildfire and (b) explore whether stripping,
bark thickness and tree size influenced each of the
observed types of post-fire responses as hypothesised.
Methods
Study area and plot definition
The study area is located in ‘‘Serra do Caldeirão’’, a
mountain range in the Algarve province, southern
Portugal. The climate is Mediterranean with an
80
average annual temperature and rainfall of 16.6C
and 900 mm, respectively. The altitude ranges from
150 to 580 m above sea level. Soils consist mainly
of shallow schist lithosols that have a low fertility
and are prone to erosion. The landscape is characterised by vast expanses of cork oak forests ranging
from areas with high tree cover, to ‘‘montados’’ that
have scattered trees and an understory of crops or
pastures. In the 2004 summer, an intense wildfire
burned ca. 25,000 ha in this region. A 1 9 1 km2
grid of points covering part of the burned area was
used to define a 50 m-radius circle (sampling plot)
around each point. Plots were checked in the field
for accessibility and to confirm whether they had
burned and were dominated by cork oak trees. A
total of 40 plots were ultimately selected. Large
within-plot variability in tree size and cork age (and
consequently bark thickness) was common since
cork debarking was not carried out simultaneously
on all individual trees (for further details see
Moreira et al. 2007).
Tree variables
Individual tree evaluation in the plots took place
between December 2005 and April 2006, approximately 1.5 years after the fire. Trees were assessed
along four 50-m strip transects departing from the
plot centres at right angles. Given the very high
density of young trees in many plots, only trees larger
than ca. 9 cm DBH were measured. Approximately
30 trees per plot were assessed (mean ± s.e. of
28.8 ± 0.51, range = 14–30, n = 40) yielding a total
of 1,151 individuals. For each tree, several variables
were measured (see Moreira et al. 2007 for details);
however, for the purposes of this article only the
following variables are presented: (a) tree size (DBH,
cm), taken as the average of two measurements at
1.3 m above ground level, (b) bark thickness (average
thickness, cm) at breast height, calculated from four
measurements using a bark gauge and (c) presence/
absence of cork stripping in order to distinguish
unstripped trees with virgin cork from exploited trees
where cork debarking (stripping) had occurred at
least once. The types of post-fire responses were also
assessed and classified into four mutually exclusive
categories: (a) dead trees (no resprouting from the
base or crown), (b) trees that resprouted exclusively
from the crown, (c) trees that resprouted exclusively
A.G. Van der Valk (ed.)
from the base (thus with a dead stem) and (d) trees
that resprouted from both the crown and base.
Data analysis
To examine the influence of tree variables on post-fire
response types, an information theoretic approach was
used based on the Akaike information criterion
corrected for small sample sizes (AICc) (Burnham
and Anderson 2002). This approach starts with the
formulation of a series of models that rely on an
understanding of the system being studied, followed by
an assessment of how different putative models
compare to the reality (Rushton et al. 2004). The suite
of candidate models is compared using AICc, and the
smaller the AICc value the better the model fits the data.
Each of the four response types was modelled
separately using a binary variable taking the value 1
for the specific response type and 0 for the remaining
types. A generalised linear model with binomial error
structure and a logit link function (McCullagh and
Nelder 1989) was used to test a group of biologically
plausible models, including separate models for each
of the three variables (stripping, bark thickness,
DBH) assumed to be biologically significant, and all
possible combinations of these variables. Two interaction terms were also added to this list of variables:
stripping 9 bark thickness, as previous analyses
showed that we could expect different responses,
for a given bark thickness, of unstripped or exploited
trees (Moreira et al. 2007); and stripping 9 DBH, as
the effects of tree size could also vary according to
stripping status. This yielded four groups (one group
per response type) of 27 models each, resulting from
all combinations of these five variables. The smaller
AICc among the models in each group was used to
identify the more parsimonious model (Burnham and
Anderson 2002) for each response type.
The fit and predictive performance of the models
with smaller AICc was evaluated through the likelihood ratio statistic (full model v2) and by calculating
the area under the receiver operating characteristics
(ROC) curve (Saveland and Neueschwander 1990;
Pearce and Ferrier 2000). This has the advantage of
assessing model performance in a threshold-independent fashion, being independent of the prevalence of
the several response types. The AUC varies between
0.5 (no discrimination ability) to 1 (perfect discrimination ability) (Pearce and Ferrier 2000). Usually,
Forest Ecology
81
AUC values of 0.5–0.7 are taken to indicate low
accuracy, values of 0.7–0.9 indicate useful applications and values above 0.9 indicate high accuracy
(Swets 1988). The calculation of the AUC and standard
error was based on a non-parametric assumption. For a
better visualization of the expected probabilities of the
fitted models, data from bark thickness and tree size
were grouped into classes. The former was divided into
three classes: B2 cm (33.8% of the trees), 2–4 cm
(54.4%) and [ 4 cm (11.8%). Tree size was also
divided into three DBH categories: B20 cm (28.8% of
the trees), 20–40 cm (58.5%) and[ 40 cm (12.7%).
There was no correlation between bark thickness and
DBH (r = 0.021, n = 1151, P = 0.487). However,
exploited trees (n = 859) had significantly larger DBH
than unstripped ones (n = 292) (mean ± s.e. of
30.7 ± 0.406 cm and 16.5 ± 0.253 cm, respectively,
t-test, t = 29.6, P \ 0.001), and had a slightly thinner
bark (mean ± s.e. of 2.39 ± 1.289 cm and
2.93 ± 0.835, respectively, t-test, t = 8.1, P \ 0.001).
Results
Influence of predictor variables on response types
The more parsimonious model for tree death, among
the set of models compared, is shown in Table 1 and
Fig. 2. The probability of a tree dying increased if it
had been exploited and had a larger DBH. Bark
thickness was also a key variable but only if trees
were exploited, in this case the probability of death
increased as bark thickness decreased. Similarly to
death, the model with the lowest AIC for resprouting
only from base showed that this response type was
also more likely in stripped trees (Table 1; Fig. 2).
Bark thickness was an important variable in the case
of stripped trees, and was negatively correlated to
basal resprouting probability. The more parsimonious
model for simultaneous resprouting from the base and
crown (Table 1; Fig. 2) included only DBH, with
larger trees being less likely to show this response
type. Finally, resprouting exclusively from the crown
was more likely in unstripped trees (Table 1; Fig. 2).
For stripped trees, this resprouting type increased
with bark thickness and decreased with DBH. Overall, model performance was low to moderate with
AUC values ranging from 0.64 to 0.82.
Response types
For the 1,151 sampled trees, the most common
response type was resprouting exclusively from
crown (68.8%, n = 792 trees), followed by death
(15.8%, n = 182), simultaneous resprouting from the
crown and base (10.1%, n = 116) and lastly, resprouting exclusively from the base (5.3%, n = 61).
Discussion
Differences in sprouting behaviour are important for
understanding vegetation dynamics, extinction risks
for threatened species and for defining management
regimes for woody plants (Bond and Midgley 2003).
Table 1 Generalized linear models with the lowest AICc among the set of models compared, for each of the four post-fire response
types in cork oak (death, resprouting exclusively from crown, resprouting exclusively from base, resprouting from both crown and
base).
Variable
Stripping
Death
1.645 ± 0.280
Base only
Crown and base
2.955 ± 0.440
Crown only
-1.464 ± 0.278
Bark thickness
DBH
Stripping 9 bark thickness
0.031 ± 0.007
-0.688 ± 0.086
-0.055 ± 0.012
-1.272 ± 0.182
0.809 ± 0.075
Stripping 9 DBH
Constant
Model v2
AUC
-0.016 ± 0.007
-2.722 ± 0.229
101.95
0.71 ± 0.022
-3.570 ± 0.358
-0.842 ± 0.283
81.03
27.79
0.82 ± 0.026
0.64 ± 0.026
2
0.940 ± 0.130
153.65
0.70 ± 0.017
The variables entering each model (linear predictor), their coefficients (±s.e.), the model v and the area under the ROC curve
(AUC ± s.e.) are shown for each response type. See Fig. 2 for model visualization. All model v2, variable coefficients and AUC
values are significant (P \ 0.05)
82
Death
nao
0.60
sim
nao
0.20
exploited
unstripped
Base only
sim
exploited
unstripped
0.50
0.40
Probability
Probability
0.15
0.30
0.10
0.20
0.05
0.10
0.00
0.00
2
2-4
>4
2
2-4
>4
2
2-4
bark thickness (cm)
nao
1.00
Crown only
sim
nao
0.20
unstripped
>4
2
2-4
>4
bark thickness (cm)
exploited
Crown and base
unstripped
sim
exploited
0.90
0.15
Probability
0.80
Probability
Fig. 2 Mean (±95%
confidence intervals) of the
predicted probability of
each response type (dead,
base only, crown and base,
crown only), according to
the GLM models shown in
Table 1, for each
combination of bark
thickness and DBH classes.
Unstripped and exploited
trees correspond,
respectively, to the left and
right panels. The three DBH
classes are shown by
different line styles (——:
[40 cm, - - - -: 20–40 cm,
: \20 cm)
A.G. Van der Valk (ed.)
0.70
0.60
0.50
0.10
0.05
0.40
0.30
0.00
2
2-4
>4
2
2-4
bark thickness (cm)
Bellingham and Sparrow’s (2000) model of resprouting response as a function of increasing disturbance
severity assumes a hierarchical sequence of regenerative responses that depends on the proportion of the
above-ground biomass lost. Here, a conceptual model
is presented (Fig. 1) which can be applied to situations where a specific disturbance (fire) caused crown
consumption on a tree with resprouting abilities. In
this situation, different levels of damage are expected
to create a sequence of response types where the
hierarchical nature of sprouting type is not necessarily followed. These damage levels are assumed to be
determined by the amount of damage to buds and
cambial tissue, and by the available below-ground
reserves that can be used to rebuild the lost biomass
(e.g. Bond and van Wilgen 1996). Different levels of
damage will cause four different types of post-fire
responses that were identified in cork oaks in
southern Portugal, 1.5 years after being burned in
an intense wildfire. The majority of trees (ca. 70%)
resprouted exclusively from the crown, which is
expected to correspond to the lower level of damage.
The second most common response category was
death (16% of the trees), which corresponds to the
highest level of damage. Response types expected to
correspond to intermediate damage levels were less
>4
2
2-4
>4
2
2-4
>4
bark thickness (cm)
common: simultaneous resprouting of the crown and
base was the third more common response type (10%
of the trees), whereas the least common response type
was resprouting only from the base (i.e. stem death)
(5% of the trees).
The relationship between the expected level of
damage and the degree to which a tree is protected
from disturbance was also addressed. The focus was
on three variables previously known to have an
important impact on cork oak post-fire survival,
namely cork stripping, bark thickness and tree size.
It was hypothesised that stripping would be a
strong determinant of the expected level of post-fire
damage to trees, since the process of extracting
highly insulating virgin cork for the first time
initialises major periodic stresses across the life span
of a tree (e.g. Natividade 1950; Costa et al. 2004;
Moreira et al. 2007). This is consistent with the result
that stripping is positively correlated with the likelihood of response types with higher levels of damage
(i.e. dead trees and resprouting from base), and was
negatively related to the probability of crown
resprouting (corresponding to lower levels of
damage).
Bark thickness is a well-known determinant of
post-fire survival in cork oak (e.g. Cabezudo et al.
Forest Ecology
1995; Pausas 1997; Barberis et al. 2003; Catry et al.
2007; Moreira et al. 2007). Insulating capacity
increases with bark thickness (Dikinson and Johnson
2001), thereby providing a higher level of protection
to both the buds and the living tissues in the vascular
cambium from which resprouting closely depends. As
expected, the thicker the bark the lower the probability of a greater level of damage (expressed in the
negative correlation of this variable with dead and
basal resprouting probability), and the higher the
probability of a low level of damage (expressed in the
positive correlation with crown resprouting probability). However, the effect of bark thickness on postfire responses is expressed only in exploited trees,
suggesting that trees with virgin cork have an
additional degree of protection that appears independent of bark thickness, as previously discussed in
Moreira et al. (2007).
The relationship between tree size (DBH) and
level of protection from fire was hypothesised to be
related to the amount of damage and to the
availability of below-ground carbohydrate reserves
that may be allocated to resprouting. To be able to
sprout and support regrowth, a plant needs surviving
meristems and stored carbohydrate reserves (Iwasa
and Kubo 1997; Bond and Midgley 2001). Older
plants should have larger below-ground reserves
(Gurvich et al. 2005) and consequently a higher
capacity to mobilise reserves in response to disturbance (Bellingham and Sparrow 2000). For example,
Malanson and Trabaud (1988) found that a 9-year-old
Q. coccifera resprouted more vigorously than a 3year-old, presumably because the latter had less
developed below-ground reserves. Other empirical
studies, however, suggest that resprouting ability
declines with age and that below-ground carbohydrate storage in larger trees may be invested in
survival rather than growth (Bond and van Wilgen
1996; Bond and Midgley 2001). In the case of cork
oak, once cork exploitation has begun, the older (and
therefore larger) trees have probably experienced a
higher number of stripping events and poor management practices (e.g. deep ploughing or excessive
canopy pruning) (Natividade 1950; Costa et al. 2004;
Silva and Catry 2006). Therefore, if below ground
reserves are allocated to recover from damages
accumulated across the plant’s lifespan, they will be
diverted from resprouting (Bellingham and Sparrow
2000; Chapin et al. 1990). Reduced vigour and
83
survivorship of resprouting shrubs have been related
to increasing disturbance frequency due to the
impossibility of rebuilding or maintaining energy
reserves in storage organs between consecutive
disturbances (Bellingham and Sparrow 2000). Thus,
it was hypothesised that bigger trees would be prone
to higher levels of damage, and this was consistent
with the fact that DBH was positively correlated with
the likelihood of death, particularly in exploited trees.
Bigger trees were also less likely to resprout from the
crown, which also indicated a higher level of damage,
and of simultaneous resprouting from the crown and
base.
The fact that the models obtained had moderate
predictive performance suggests that other variables,
which may not be directly related to tree features,
also contribute to the level of damage suffered by
each individual. If these other factors had been taken
into account, they would probably explain a larger
proportion of variability in the observed patterns. For
example, Moreira et al. (2007) showed that variables
related to stand structure (e.g. tree density or
understory vegetation height) and topographic location (slope and aspect) as well as indicators of fire
severity (e.g. charring height) were significant predictors of cork oak mortality in the same study area.
Other factors known to influence post-fire sprouting
responses include site quality (López Soria and
Castell 1992), disturbance frequency (Bond and
Midgley 2001), fire season (Konstantidinis et al.
2006) and the existence of herbivory (Moreno and
Oechel 1991).
The expected probabilities of the different
response type models reflected the relative frequency
of these types in the field. Thus, the more likely
response type (the one with higher expected probability) for all possible combinations of stripping
status, cork thickness and bark thickness was resprouting from crown. The only exception was when trees
were exploited, had a very thin bark (\2 cm) and
were very big ([40 cm DBH). This corresponds to
the higher level of damage in our model, which is
consistent with the fact that death was the most likely
outcome (Fig. 2).
In summary, the influence of the studied variables
on the post-fire response patterns of cork oak after fire
were in agreement with the hypothesis that different
levels of damage (and corresponding response types)
may be found in sprouting trees where all the crown
84
was destroyed by wildfire. We provided evidence that
four different response types may occur, and that
these are influenced by stripping status, bark thickness and tree size, which are probably related to the
level of protection of buds and cambial tissue, and to
the amount of below ground reserves available for the
plant to invest in resprouting.
Acknowledgements Thanks are due to Raimundo Duarte,
Rebeca Alvarez, Ana Oliveira and Rui Morgado, for lab and field
work. Thanks also due to the contribution of two anonymous
referees, whose commentaries significantly improved the
manuscript. This research was carried out within the scope of
projects INTERREG III-B RECOFORME, POCI/AGR/58896/
2004, POCI/AGR/61407/2004 and FFP - Recuperação de áreas
ardidas.
References
Barberis A, Dettori S, Filigheddu MR (2003) Management
problems in Mediterranean cork oak forests: post-fire
recovery. J Arid Environ 54:565–569. doi:10.1006/
jare.2002.1079
Bellingham PJ, Sparrow AD (2000) Resprouting as a life history strategy in woody plant communities. Oikos 89:409–
416. doi:10.1034/j.1600-0706.2000.890224.x
Bond WJ, van Wilgen BW (1996) Fire and plants. Chapman &
Hall, London
Bond WJ, Midgley JJ (2001) Ecology of sprouting in woody
plants: the persistence niche. Trends Ecol Evol 16:45–51.
doi:10.1016/S0169-5347(00)02033-4
Bond WJ, Midgley JJ (2003) The evolutionary ecology of
sprouting in woody plants. Int J Plant Sci 164:103–114.
doi:10.1086/374191
Burnham KP, Anderson DR (2002) Model selection and multimodel inference. A practical information-theoretic
approach, 2nd edn. Springer, New York
Cabezudo B, Latorre A, Nieto J (1995) After fire regeneration
in a Quercus suber forest in the South of Spain (Istan,
Malaga). Acta Bot Malacitana 20:143–151
Catry FX, Rego FC, Bugalho MN, Lopes T, Silva JS, Moreira
F (2007) Post-fire tree mortality and regeneration in a
mixed forest: evaluating the cumulative impacts of herbivory. In: Rokich D, Wardell-Johnson, Yates C, Stevens
J, Dixon K, McLellan R, Moss G (eds) Proceedings of the
International Mediterranean Ecosystems ConferenceMedecos XI 2007. Kings Park and Botanic Garden, Perth,
Australia, pp 45–46
Chapin FS, Ernst-Detlef S, Mooney HA (1990) The ecology
and economics of storage in plants. Annu Rev Ecol Syst
21:423–447. doi:10.1146/annurev.es.21.110190.002231
Costa A, Pereira H, Oliveira A (2004) The effect of corkstripping damage on diameter growth of Quercus suber L.
Forestry 77:1–8. doi:10.1093/forestry/77.1.1
Del Tredici P (2001) Sprouting in temperate trees: a morphological and ecological review. Bot Rev 67:121–140.
doi:10.1007/BF02858075
A.G. Van der Valk (ed.)
Dikinson MB, Johnson EA (2001) Fire effects on trees. In:
Johnson EA, Miyanishi K (eds) Forest fires: behaviour
and ecological effects. Academic Press, New York, pp
477–525
Iwasa Y, Kubo T (1997) Optimal size of storage for recovery
after unpredictable disturbances. Evol Ecol 11:41–65.
doi:10.1023/A:1018483429029
Gurvich DE, Enrico L, Cingolani AM (2005) Linking plant
functional traits with post-fire sprouting vigour in woody
species in central Argentina. Austral Ecol 30:789–796.
doi:10.1111/j.1442-9993.2005.01522.x
Keeley J (2006) Fire severiy and plant age in postfire resprouting
of woody plants in sage scrub and chaparral. Madrono
53:373–379.
doi:10.3120/00249637(2006)53[373:FSAPAI]2.0.CO;2
Konstantidinis P, Tsiourlis G, Xofis P (2006) Effect of fire
season, aspect and pre-fire plant size on the growth of
Arbutus unedo L. (strawberry tree) resprouts. For Ecol
Manag 225:359–367
Kozlowski TT (1971) Growth and development of trees, vol 1.
Academic Press, New York
Kozlowski TT, Kramer PJ, Pallardy JG (1991) The physiological ecology of woody plants. Academic Press, San
Diego
Lamey A (1893) Le chêne-liège, sa culture et son exploitation.
Paris Nancy, Berger-Levrault et Cie éditeurs
López Soria L, Castell C (1992) Comparative genet survival
after fire in woody Mediterranean species. Oecologia
91:493–499. doi:10.1007/BF00650321
Malanson GP, Trabaud L (1988) Vigour of post-fire resprouting by Quercus coccifera L. J Ecol 76:351–356.
doi:10.2307/2260598
McCullagh P, Nelder JA (1989) Generalized Linear Models.
Chapman and Hall, London
Miller M (2000) Fire autecology. In: Brown JK, Smith JK (eds)
Wildland fire in ecosystems: effects of fire on flora. Gen
Tech Rep RMRS-GTR-42, vol 2. Department of Agriculture, Forest Service, Rocky Mountain Research
Station, Ogden, UT, US, pp 9–34
Moreira F, Duarte I, Catry F, Acácio V (2007) Cork extraction
as a key factor determining post-fire cork oak survival in a
mountain region of southern Portugal. For Ecol Manag
253:30–37
Moreno JM, Oechel WC (1991) Fire intensity and herbivory
effects on post-fire resprouting of Adenostoma fasciculatum in southern California chaparral. Oecologia 85:429–
433. doi:10.1007/BF00320621
Natividade JV (1950) Subericultura. DGSFA, Lisboa
Pampiro F, Pintus A, Ruiu PA (1992) Interventi di recupero di
una giovane sughereta percorsa da incêndio. In: Instituto de
Promoción del Corcho (ed) Simpósio Mediterrâneo sobre
Regeneración del Monte Alcornocal, Mérida, pp 174–177
Pausas J (1997) Resprouting of Quercus suber in NE Spain
after fire. J Veg Sci 8:703–706. doi:10.2307/3237375
Pearce J, Ferrier S (2000) Evaluating the predictive performance of habitat models developed using logistic
regression. Ecol Modell 133:225–245. doi:10.1016/
S0304-3800(00)00322-7
Pereira H, Tomé M (2004) Cork oak. In: Burley J, Evans J,
Youngquist JA (eds) Encyclopedia of Forest Sciences.
Elsevier, Oxford, pp 613–620
Forest Ecology
Rushton SP, Ormerod SJ, Kerby G (2004) New paradigms for
modelling species distributions? J Appl Ecol 41:193–200.
doi:10.1111/j.0021-8901.2004.00903.x
Saveland JM, Neueschwander LF (1990) A signal detection
framework to evaluate models of tree mortality following
fire damage. For Sci 36:66–76
Silva JS, Catry F (2006) Forest fires in cork oak (Quercus
suber) stands in Portugal. Int J Environ Stud 63:235–257.
doi:10.1080/00207230600720829
85
Swets JA (1988) Measuring the accuracy of diagnostic systems.
Science 240:1285–1293. doi:10.1126/science.3287615
Trollope WSW (1984) Fire in savanna. In: de Booysen PV,
Tainton NM (eds) Ecological effects of fire in south
African ecosystems. Springer Verlag, Berlin, pp 199–218
Whelan RJ (1995) The ecology of fire. Cambridge University
Press, Cambridge
Non-woody life-form contribution to vascular plant species
richness in a tropical American forest
Reynaldo Linares-Palomino Æ Victor Cardona Æ Ernest I. Hennig Æ
Isabell Hensen Æ Doreen Hoffmann Æ Jasmin Lendzion Æ Daniel Soto Æ
Sebastian K. Herzog Æ Michael Kessler
Originally published in the journal Plant Ecology, Volume 201, No. 1, 87–99.
DOI: 10.1007/s11258-008-9505-z The Author(s) 2008. This article is published with open access at Springerlink.com
Abstract We provide total vascular plant species
counts for three 1-ha plots in deciduous, semideciduous and evergreen forests in central Bolivia.
Species richness ranged from 297 species and
22,360 individuals/ha in the dry deciduous forest to
382 species and 31,670 individuals/ha in the
Electronic supplementary material The online version of
this article (doi:10.1007/978-90-481-2795-5_8) contains
supplementary material, which is available to authorized users.
R. Linares-Palomino (&) M. Kessler
Department of Systematic Botany, Albrecht-von-HallerInstitute for Plant Sciences, University of Göttingen,
Untere Karspüle 2, 37073 Göttingen, Germany
e-mail: r.linaresp@yahoo.co.uk
V. Cardona
Herbario Nacional de Bolivia, Instituto de Ecologı́a,
Universidad Mayor de San Andrés, Casilla 10077, La Paz,
Bolivia
E. I. Hennig
Department of Environmental Sciences, Institute of
Terrestrial Ecosystems, ETH Zürich, Universitätstrasse
16, 8092 Zurich, Switzerland
I. Hensen D. Hoffmann
Plant Ecology, University of Halle – Wittenberg, Am
Kirchtor 1, 06108 Halle (Saale), Germany
evergreen forest. Orchidaceae, Pteridophyta and
Leguminosae were among the most species-rich
major plant groups in each plot, and Peperomia
(Piperaceae), Pleurothallis (Orchidaceae) and Tillandsia (Bromeliaceae), all epiphytes, were the most
species-rich genera. This dominance of a few but
very diverse and/or widespread taxa contrasted with
the low compositional similarity between plots. In a
neotropical context, these Central Bolivian forest
plots are similar in total species richness to other dry
Present Address:
J. Lendzion
Institute of Botany and Landscape Ecology,
Ernst-Moritz-Arndt-University Greifswald,
Grimmer Str. 88,
17487 Greifswald, Germany
D. Soto
Herbario del Oriente Boliviano, Museo Noel Kempff
Mercado, Santa Cruz de la Sierra, Bolivia
S. K. Herzog
Asociación Armonı́a - BirdLife International, Casilla
3566, Santa Cruz de la Sierra, Bolivia
Present Address:
M. Kessler
Institute of Systematic Botany, University of Zürich,
Zollikerstrasse 107, 8008 Zurich, Switzerland
J. Lendzion
Department of Plant Ecology and Ecosystem Research,
Albrecht-von-Haller-Institute for Plant Sciences,
University of Göttingen, Untere Karspüle 2, 37073
Göttingen, Germany
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_8
87
88
deciduous and humid montane forests, but less rich
than most Amazonian forests. Nevertheless, lianas,
terrestrial herbs and especially epiphytes proved to be
of equal or higher species richness than most other
neotropical forest inventories from which data are
available. We therefore highlight the importance of
non-woody life-forms (especially epiphytes and terrestrial herbs) in Andean foothill forest ecosystems in
terms of species richness and numbers of individuals,
representing in some cases nearly 50% of the species
and more than 75% of the individuals. These figures
stress the need for an increased inventory effort on
non-woody plant groups in order to accurately direct
conservation actions.
Keywords Alpha diversity Andean foothills
forest ecosystem Life-form diversity
Non-woody plants Total species inventory
Introduction
Statements about the diversity of plant species in
forest ecosystems are usually based on results from
vegetation inventories that are mostly restricted to a
certain plant subgroup. Woody species, usually trees
and shrubs with diameter at breast height of C1 cm
(e.g. the STRI 50-ha plots, Condit 1995), C2.5 cm
(e.g. 0.1-ha transects, Gentry 1982) and C10 cm (e.g.
Gentry 1988; Valencia et al. 1994; Smith and Killleen 1998), are the most commonly studied plant
groups. In contrast, herbs (e.g. Poulsen and Balslev
1991; Poulsen and Nielsen 1995), lianas (PérezSalicrup et al. 2001; Mascaro et al. 2004) and
epiphytes (Ingram et al. 1996; Arévalo and Betancur
2004; Benavides et al. 2005; Krömer et al. 2005) are
less commonly used to characterize the diversity of
vegetation types. These non-woody life-forms, however, have been shown to be of importance in the few
assessments of tropical plant alpha diversity in which
all vascular plants were counted (Whitmore et al.
1985; Gentry and Dodson 1987; Duivenvoorden
1994; Balslev et al. 1998; Galeano et al. 1998;
Langenberger et al. 2006). The scarcity of such
studies can be attributed to the difficulties associated
with identification of more (and usually less well
known) plant groups (restricting inventories to some
A.G. Van der Valk (ed.)
life-form groups in the tropics is already a huge
identification task) and the difficulty of collecting
epiphyte specimens from the forest canopy. Whitmore and colleagues have undertaken the most
comprehensive study of vascular plants to date in a
Costa Rican rain forest. To accomplish their task of
inventorying all species (including non-vascular
plants), destructive sampling of a 10 m 9 10 m plot
was required (Whitmore et al. 1985).
The few full tropical plant inventories performed
to date have focused on a single and homogeneous
vegetation type, usually tropical lowland rain forests.
Although some of these studies (e.g. Duivenvoorden
1994; Langenberger et al. 2006) inventoried plots
and transects along edaphic and physiographic gradients, only two have inventoried and compared
different vegetation types using a uniform sampling
methodology throughout. Álvarez et al (2003)
reported total vascular plant counts in three 0.1-ha
plots in Amazonian, Chocoan and Andean forests in
Colombia. This study, however, was not published
formally, and epiphytes in the Amazonian plot were
not sampled, restricting the total vascular plant count
to the Chocó and Andean forest only. The other study
by Gentry and Dodson (1987) compared three 0.1-ha
plots in wet, moist and dry forests in Ecuador. The
lack of standardized inventory methods hampers the
quantitative comparison between both of these studies. The use of florulas could be an option to compare
different forests (e.g. Gentry 1990), but the size of the
areas studied and collection intensities are not
uniform (Tobler et al. 2007).
We chose Central Bolivia, a region where four
major biomes occur in close proximity to each other
(humid and moist vegetation from Amazonia, seasonal subtropical lowland vegetation from the Chaco,
subtropical highland vegetation from the Andes and
seasonal vegetation of the Chiquitanı́a (Ibisch et al.
2003)), as our study region. We established within
this complex biogeographic setting three permanent
1-ha plots. We used a uniform methodology along a
humidity gradient from deciduous to evergreen
forest, inventorying all vascular plants present. Our
main objective was to quantitatively assess the
relative importance of different life-form groups
and taxa within the different vegetation types we
surveyed and to compare our results with similar
studies in the neotropical region.
Forest Ecology
Methods
Study area
The study was carried out at the Refugio Los
Volcanes in Santa Cruz, Bolivia. Los Volcanes is a
private reserve of approximately 300 ha. It is located
about 18060 S and 63360 E and is adjacent to the
southern border of Amboró National Park, directly on
the transition from the humid inner tropics to the
seasonally dry subtropics (Fig. 1a). The substrate of
the study area consists primarily of red sandstone and
locally of loamy sedimentary rocks (lutite). These red
sandstones form cliffs several hundred metres high
and are intersected by narrow valleys providing the
area with dramatic scenery. Annual precipitation is
about 1200–1500 mm, with most of the rainfall from
October/November to March/April, but with high
temporal variability.
The general vegetation of the area has been
classified as ‘subhumid to humid deciduous forest
of southeastern Amboró’ (Navarro et al. 1996)
and is usually found at 900–1100 masl. Among
the dominant tree species are Aspidosperma cylindrocarpon (Apocynaceae), Cariniana estrellensis
(Lecythidaceae),
Cedrela
lilloi
(Meliaceae),
Gallesia integrifolia (Phytolaccaceae), Pachystroma
Fig. 1 Main vegetation types and plot shape and location in
the study area. a Aerial photograph of the Refugio Los
Volcanes area in central Bolivia, inset showing map of Bolivia
and location of study area. b Schematic representation of the
89
longifolium (Euphorbiaceae), Pogonopus tubulosus
(Rubiaceae) and Tabebuia lapacho (Bignoniaceae)
(Navarro et al. 1996). Locally, however, vegetation
types are determined by differences in topography,
aspect and precipitation regimes that lead to ecologically relevant differences in water availability within
the study area. Consequently, the dominant zonal
vegetation is semi-deciduous forest (about 30–50%
deciduous trees) mainly found on shaded southfacing slopes. Steep, sunny and north-facing slopes
are occupied by deciduous forest (70–90% deciduous
trees), whereas flat, shaded valleys with groundwater
supply support evergreen forest (10–20% deciduous
trees) (Fig. 1b).
Vegetation sampling
A permanent plot of 1 ha was established in each
forest type (deciduous, semi-deciduous and evergreen) between 2002 and 2003. Each plot was
subdivided into 25 adjacent 20 m 9 20 m subplots.
Plots where laid out in such a way as to include only
the forest type under study, avoiding other forest
types, young secondary vegetation and non-forest
vegetation (e.g. rock outcrops). Thus, our plots are
not the traditional square 100 m 9 100 m inventory
plots but have rather irregular shapes (Fig. 1c).
major vegetation types and geographical characteristics of the
area, showing plot locations and orientation (ss: sandstone, ca:
cleared area, LV Station: Los Volcanes Research Station). c
Shape of inventory plots, where numbers denote subplots
90
All vascular plants in each plot were inventoried
between 2002 and 2004, mainly in the season
following the summer rains (i.e. May–August). J.
Lendzion inventoried herbs, shrub and tree seedlings,
E. I. Hennig epiphytes, D. Hoffmann lianas and V.
Cardona, D. Soto and S. K. Herzog woody plants. For
the herb inventories, we recorded all species with
stem diameter below 1 cm. Additionally, we recorded
all Cactaceae, Bromeliaceae and Costaceae below
1 m height and epiphytes on fallen branches. For
lianas, we recorded all individuals, including Araceae, with a diameter of [1 cm at 1.3 m above soil
level. All epiphytes were observed and counted.
Collections of epiphytes were made either with a
clipper pole or with the help of rope-climbing
techniques. Binoculars were used to aid identification
when they were too inaccessible to collect. Finally,
we recorded all woody plants, excluding lianas, with
diameter at breast height (dbh) of [1 cm.
Voucher specimens of all species were collected
for later determination and are deposited at USZ
(Santa Cruz) and LPB (La Paz), with a small subset
of samples at the Göttingen Herbarium (GOET)
(herbarium acronyms follow Holmgren and Holmgren 1998). Several sterile specimens could not be
fully identified and were sorted into morphospecies.
The final stage of data production was completed at
USZ (by R. Linares-Palomino) by cross-checking all
collected vouchers in order to unify morphospecies
delimitations.
A.G. Van der Valk (ed.)
understanding of the phylogeny of extant ferns,
familial composition and relationships are still
unsatisfactorily solved (Smith et al. 2006; Schuettpelz and Pryer 2007). We therefore refrained from
assigning our collections to families and treated all
ferns and fern allies as a single taxon Pteridophyta.
We computed species accumulation curves based
on the 20 m 9 20 m subplots using EstimateS (Colwell 2005). Similarity between forest plots was
evaluated by subtracting the Bray–Curtis distance
between two forest plots from unity. Pair-wise Bray–
Curtis distances (DBC) were calculated in the vegan
package for R (Oksanen et al. 2006; R Development
Core Team 2006) using presence/absence data by
DBC = 2a/(2a ? b ? c), where a is the total number
of species present in both forest plots, b is the number
of species present only in the first forest plot, and c is
the number of species present only in the second
forest plot (Magurran 2004). In order to compare the
species richness of the Los Volcanes plots with that
of other forests in the neotropics, we searched for
other published full plant, epiphyte, liana, terrestrial
herb and tree/woody plant inventories (Appendix 1)
and plotted species accumulation curves for each
forest type at Los Volcanes against the species
richness data of the other studies.
Results
Taxonomic diversity
Data analysis
We used a conservative approach in calculating
species numbers by lumping highly similar morphospecies into one group instead of considering them as
several distinct species. The herb inventory, which
included life-forms other than herbaceous plants, was
split into terrestrial herbs, tree seedlings, shrub
seedlings and epiphytes. Thus, terrestrial herbs
formed a life-form group by itself in subsequent
analyses. The other three subgroups were crossreferenced with the tree, shrub and epiphyte inventories and merged accordingly. We follow the
TROPICOS and Flora of Bolivia online databases
for nomenclatural purposes (both available at
http://mobot.mobot.org/W3T/Search/vast.html and
http://www.efloras.org/flora_page.aspx?flora_id=40,
respectively). Despite much progress in the
We recorded 80,352 individual plants belonging to
670 species (including morphospecies) on the three
plots (Appendix 2). We were able to completely
identify 52% of our collections to species level (341
species), an additional 25% could be assigned to
genus (172 morphospecies) and 14% to family (95
morphospecies). Nine percent (62 morphospecies)
could not be assigned to a family or lower taxon.
The most species-rich plots were in the evergreen
and semi-deciduous forest, both of which had an
almost identical number of species (381 and 382,
respectively). The deciduous forest had 297 species.
Of the 273 genera, most were recorded in the
evergreen and semi-deciduous forest (190 and 185,
respectively) compared to 162 genera in the deciduous forest. Of the 92 families, 75 were found in the
evergreen, 72 in the semi-deciduous and 60 in the
Forest Ecology
91
Table 1 Number of families, genera and species of three 1-ha plots in Santa Cruz, Central Bolivia (A: angiosperms, P:
pteridophytes)
Total from the three forest plots
Deciduous forest plot
Semi-deciduous forest plot
Evergreen forest plot
A
A
A
A
P
Total
P
Total
P
Total
P
Total
Families
80
12
92
55
5
60
64
8
72
65
10
75
Genera
245
28
273
149
13
162
168
17
185
166
24
190
Species
617
53
670
279
18
297
353
29
382
337
44
381
Life-form
Epiphyte
Hemiepiphyte
142
67
80
109
9
1
4
8
Liana
153
64
86
44
Shrub
57
97
45
49
Tree, liana
1
1
–
1
Tree, shrub
39
17
16
20
Terrestrial herb 79
Tree
148
30
71
42
105
57
84
Other
1
–
–
1
Parasite
1
1
–
–
Life-form composition values show the number of species assigned to each plant group
deciduous forest. The contribution of ferns and
lycophytes (‘‘pteridophytes’’) to species richness
was higher in the evergreen forest than in the two
other forest types (Table 1).
Of the 10 most species-rich families, seven were
shared between all three plots, although with different
ranking within each plot (Table 2). Taking all three
plots together, the most species-rich families were
Orchidaceae, pteridophytes, Leguminosae and Bignoniaceae. Orchidaceae, a family containing mostly
epiphytic species, was by far the most species-rich in
all plots. Pteridophytes, composed mostly of ground
herbs, ranked second in the evergreen and semideciduous and fourth in the deciduous forest. Leguminosae, which was mainly composed of woody
species in our plots, decreased in importance from the
deciduous (second) to the evergreen forest (fourth).
Absolute species numbers were similar in the evergreen forest and higher in the semi-deciduous forest
as compared to the deciduous forest. Bignoniaceae (a
family including liana, shrub and tree species), the
third most important family in the deciduous forest,
was the fourth most important family in the semideciduous forest (again with a higher species number)
but was ranked only eighth in the evergreen forest.
Only two other families were important in terms of
species numbers, and these were shared by two forest
types: Apocynaceae (mostly trees) present in the
deciduous and semi-deciduous forest and Rubiaceae
(shrubs and trees) present in the semi-deciduous and
evergreen forest.
In contrast to families, only five species-rich
genera were common to all three forest plots
(Peperomia, Pleurothallis, Tillandsia, Acalypha and
Eugenia). Of these, the three most species-rich genera
were Peperomia, Tillandsia and Pleurothallis,
although ranking varied between forest plots
(Table 2).
Two species of Tillandsia had the highest numbers
of individuals on all three plots (Table 2): T. bryoides
had highest numbers in the deciduous forest, whereas
T. tenuifolia had most individuals in the semi-deciduous and evergreen forests. The 10 species with
highest number of individuals in the deciduous forest
included epiphytes and terrestrial herbs (three species
each) and shrubs and trees (two species each). The
contribution of non-woody plants increased in the
semi-deciduous forest, including epiphytes (four
species), terrestrial herbs (three species) and one
species each of shrubs, trees and lianas. Non-woody
plant contribution was highest in the evergreen forest
with six species of epiphyte, two species of terrestrial
herb and one hemiepiphyte species dominating. Only
one shrub species was included among the top 10.
Deciduous
Semi-deciduous
Evergreen
Total
Orchidaceae (85, e)
92
Table 2 Most species-rich families and genera (number of species in parentheses) and most abundant species (number of individuals in parentheses)
Families (Pteridophyta is considered as one family)
Orchidaceae (34, e)
Orchidaceae (45, e)
Orchidaceae (60, e)
Leguminosae (20, t)
Pteridophyta (29, h)
Pteridophyta (44, h)
Pteridophyta (53, h)
Bignoniaceae (19)
Pteridophyta (18, h)
Leguminosae (26, t)
Bignoniaceae (23)
Bromeliaceae (20, e)
Leguminosae (20, t)
Leguminosae (39, t)
Bignoniaceae (35)
Bromeliaceae (16, e)
Rubiaceae (16)
Piperaceae (16)
Bromeliaceae (26, e)
Euphorbiaceae (13, s,t)
Bromeliaceae (15, e)
Rubiaceae (14)
Rubiaceae (23)
Piperaceae (12)
Sapindaceae (13)
Euphorbiaceae (12, s,t)
Euphorbiaceae (21, s,t)
Acanthaceae (12, h)
Euphorbiaceae (12, s,t)
Araceae (9, e)
Acanthaceae (20, h)
Apocynaceae (10, t)
Apocynaceae (11, t)
Moraceae (9, t)
Apocynaceae (18, t)
Myrtaceae (9, s,t)
Piperaceae (11)
Asteraceae (8)
Piperaceae (18)
Cactaceae (9)
Bignoniaceae (8)
Lauraceae (8, t)
Melastomataceae (8)
Genera
Peperomia (9)
Tillandsia (11)
Tillandsia (13)
Tillandsia (15)
Pleurothallis (8)
Peperomia (8)
Pleurothallis (10)
Pleurothallis (14)
Tillandsia (7)
Pleurothallis (8)
Peperomia (9)
Peperomia (11)
Acalypha (6)
Forsteronia (6)
Forsteronia (6)
Acalypha (5)
Epidendrum (8)
Acalypha (7)
Acalypha (10)
Epidendrum (9)
Forsteronia (9)
Acacia (5)
Eugenia (5)
Begonia (7)
Eugenia (5)
Machaerium (5)
Piper (7)
Begonia (8)
Machaerium (4)
Philodendron (5)
Eugenia (6)
Eugenia (7)
Epidendrum (4)
Psychotria (5)
Ficus (6)
Ficus (7)
Philodendron (4)
Randia (5)
Thelypteris (6)
Piper (7)
Ruellia (4)
Thelypteris (5)
Tillandsia bryoides (2410, e)
Tillandsia tenuifolia (2008, e)
Tillandsia tenuifolia (5576, e)
Tillandsia tenuifolia (8059, e)
Peperomia comarapana (1654, h)
Piper callosum (1132, s)
Racinaea parviflora (1879, e)
Peperomia tetragona (2549, e)
Rinorea ovalifolia (1534, s)
Paspalum humboldtianum (972, h)
Vriesea maxoniana (1745, e)
Racinaea parviflora (2426, e)
Petiveria alliacea (1056, s)
Olyra fasciculata (932, h)
Blechnum occidentale (1058, h)
Tillandsia bryoides (2410, e)
Peperomia tetragona (1012, e)
Peperomia tetraphylla (925, e)
Philodendron camposportoanum (837, e)
Piper callosum (2156, s)
A.G. Van der Valk (ed.)
Species (number of individuals)
The dominant life-form is indicated for families and individual species (e = epiphytic, h = terrestrial herb, s = shrub, t = tree, l = liana). Nomenclature follows TROPICOS
and the Flora of Bolivia online databases (http://mobot.mobot.org/W3T/Search/vast.html and http://www.efloras.org/flora_page.aspx?flora_id=40, respectively)
Peperomia comarapana (1678, h)
Commelina erecta (1574, h)
Piper amalago (681, t)
Herreria montevidensis (521, l)
Tillandsia streptocarpa (755, e)
Coursetia brachyrhachis (522, t)
Peperomia aceroana (735, e)
Piper callosum (692, s)
Lomagramma guianensis (741)
Philodendron camposportoanum (745, e)
Commelina erecta (777, h)
Olyra fasciculata (2012, h)
Peperomia tetragona (748, e)
Peperomia tetragona (789, e)
Philodendron camposportoanum (2007, e)
Thelypteris dentata (803, h)
Adiantum tetraphyllum (800, h)
Acacia polyphylla (802, t)
Olyra fasciculata (986, h)
Evergreen
Vriesea maxoniana (2120, e)
93
Semi-deciduous
Deciduous
Table 2 continued
Total
Forest Ecology
Life-form composition
In the deciduous forest, trees, epiphytes and lianas
contributed similar species numbers (22–24% of the
total, Table 1). In the semi-deciduous forest, trees
were slightly more species-rich than other life-forms
(28%), followed by lianas (23%) and epiphytes
(21%). The most species-rich life-form in the evergreen forest was epiphytes (29%), followed by trees
(22%), shrubs (15%) and ground herbs (15%)
(Table 1). Woody plants (including lianas and
shrubs) contributed to approximately 66%, 57% and
54% of total species richness in the deciduous, semideciduous and evergreen forest plots, respectively.
There was a highly significantly statistical difference
between the proportions of life-forms in the three
studied forest types (G-test, G = 28.34, P = 0.0004,
df = 8).
Pearson correlation analyses among life-form
richness patterns within each forest plot showed that
tree species richness patterns were positively and
significantly correlated with total plant species richness: deciduous forest r = 0.67 (P = 0.0002), semideciduous forest r = 0.54 (P = 0.0049), evergreen
forest r = 0.59 (P = 0.0021). No other significant
correlation could be detected between trees and other
life-forms, except with shrubs in the semi-deciduous
forest (r = 0.51, P = 0.01) and with lianas in the
evergreen forest (r = 0.53, P = 0.0064). Apart from
trees, lianas were the only other life-form that showed
similar levels of positive and significant correlations
with total plant species richness across the three
plots: deciduous forest r = 0.63 (P = 0.0007), semideciduous forest r = 0.66 (P = 0.0003), evergreen
forest r = 0.54 (P = 0.0051). Other positive significant correlations were detected in the evergreen
forest plot between total plant species richness and
epiphytes (r = 0.60, P = 0.0017) and terrestrial
herbs (r = 0.69, P = 0.0001).
Similarity among plots
Similarity among plots was 0.46 between deciduous
and semi-deciduous (155 species shared), 0.37
between deciduous and evergreen (125 species
shared) and 0.57 between semi-deciduous and evergreen (216 species shared). We recorded 106 species
occurring in all three plots. One hundred and twentythree species were recorded only in the deciduous
94
plot, 117 only in the semi deciduous plot and 146
only in the evergreen forest plot.
Discussion
Plot shape and its influence on species richness
estimations
Spatial distribution patterns of plant species in
tropical forests are influenced by niche assembly
and/or random dispersal assembly processes acting at
both local and landscape scales (Chave 2004; Gaston
and Chown 2005; John et al. 2007), resulting in
mostly clumped distributions (e.g. Condit et al.
2000). Within this scenario, plot shape has been
documented to influence estimates of species richness, i.e. longer and narrower rectangular plots are
prone to capture more species than square plots of
similar area (Condit et al. 1996; Laurance et al.
1998). The differences, however, were found to be
small and statistically non-significant in a study
comparing tree species richness in 100 m 9 100 m
square plots versus 40 m 9 250 m rectangular plots
in Central Amazonia (Laurance et al. 1998). Given
the patchy and fragmented nature of the deciduous
and evergreen forests in our study area, it was
impossible to survey the vegetation types in traditional square or rectangular plots. Rather, we tried to
survey as environmentally homogeneous an area as
possible, leading to our irregular plot shape design.
There is little doubt that plot shape has influenced our
results, especially those of the evergreen forest plot.
The extent of this influence, however, seems to be
small, since the species accumulation curves for all
vascular plants together and for individual life-forms
significantly decrease or level off when the hectare is
completely surveyed, suggesting that sampling was
representative in all three forest types.
Alpha diversity and plant density at Los Volcanes
Additional plant surveys across the entire Los
Volcanes reserve (approx. 300 ha) have documented
65 species of pteridophytes (M. Kessler, unpublished
data), of which 53 (82%) were recorded in the plots.
The corresponding figures are 83% for Acanthaceae,
90% for Araceae, 84% for Bromeliaceae and 92% for
Cactaceae. If these groups are considered to be
A.G. Van der Valk (ed.)
representative of the total flora, then our plots contain
roughly 80–90% of the vascular plant flora of Los
Volcanes reserve, which would then be estimated to
be around 740–840 species, corresponding to about
6.4–7.2% of the Bolivian vascular plant flora (estimated at 11,600 species, Jørgensen et al. 2006).
Species that were not encountered in the plots are
either forest species that are patchily distributed or
non-forest species occurring on sandstone walls of
the area, secondary vegetation on landslides, rock
falls or along streams.
The impressive number of individuals of T. tenuifolia and T. bryoides in the Los Volcanes plots is not
unique to these forests. Both have been documented
as the characteristic and dominant species of some
seasonal forests of Central and Southern Bolivia
(Navarro 2001). Likewise, Bonnet (2006) and Bonnet
et al. (2007) reported high densities and wide
regional distribution for T. tenuifolia in Paraná,
Brazil. They attributed the success of this species in
humid and seasonal semi-deciduous forests to its
small size (ca. 25 cm long, Smith and Downs 1977),
the presence of plumose wind-dispersed diaspores,
CAM metabolism, its atmospheric nutrient acquisition strategy (i.e. species that have no form of
absorptive root system, in which the tank habit is
lacking and where epidermal trichomes cover the
whole shoot system and are entirely responsible for
nutrient and water uptake, cf. Griffiths and Smith
1983) and the fact that this species usually forms
dense monospecific associations with no explicit
preference for some position on the phorophyte. Most
of these factors are also true for T. bryoides, in
particular the small size of the plants (usually no
longer than 5 cm), their plumose diaspores (Smith
and Downs 1977) and the formation of dense
monospecific populations (Navarro 2001).
Los Volcanes plots in a neotropical context
To our knowledge, there is only one other study
(Balslev et al. 1998) that includes an inventory of all
vascular plants in 1 ha of tropical forest and is
therefore directly comparable to our study (although
the mentioned study inventoried trees and shrubs of
1–5 cm dbh only in a 0.49-ha subplot). In the
Amazonian terra firme rain forest of Cuyabeno,
Ecuador, a perhumid area with 3555 mm of annual
rainfall and no dry season, Balslev et al. (1998) found
Forest Ecology
95
0.2
0.4
0.6
0.8
0.5
1.5
0.0
0.2
0.4
0.6
0.8
1.0
Area (ha)
0
50
100
No. Species
150
(e)
0
No. Species
1.0
Area (ha)
50 100 150 200 250 300 350
Area (ha)
(d)
300
0
0.0
1.0
200
No. Species
50
0
0
0.0
100
200 250
150
No. Species
600
400
(c)
100
800 1000
(b)
Lowland humid
Lowland deciduous
Montane humid
sacD
sacSD
sacE
200
No. Species
(a)
400
higher than terrestrial herbs (ca. 10%). The most
striking difference is in the evergreen forest in which
epiphytes (31%) were more species-rich than trees
(27%).
For a more representative comparison of the Los
Volcanes data with other neotropical sites, we
constructed species accumulation curves for Los
Volcanes that allowed comparisons with other surveys with plot sizes of up to 1 ha (Fig. 2). As Los
Volcanes is found in a biogeographical transition
zone and has high moisture variability between plots,
our study plots cannot be easily assigned to any of the
usual broad categories used for neotropical forests
(lowland humid, montane humid, lowland dry, etc.).
We therefore compared our richness counts with data
from a wide range of other neotropical forest habitats.
Most inventories in lowland humid forests (mostly
in Amazonia) have higher vascular plant, epiphyte,
liana and terrestrial herb counts than any of the plots at
Los Volcanes (Fig. 2). In contrast, plots inventoried in
dry deciduous or humid montane forests have similar
levels of species richness (Fig. 2). For example,
942 species of vascular plants in 88 families. Our
plots contained between 297 and 382 species in 60–
75 families/ha for the three forest types studied. This
is clearly much lower than the Cuyabeno plot, but
lower rainfall and strong seasonality set the forests at
Los Volcanes apart from the Amazonian site. There
were further differences in the relative contribution of
different life-forms to overall species richness at both
sites. At Cuyabeno, trees (which made up 50% of
species) were clearly the most species-rich group,
distantly followed by epiphytes (making up 18% of
species) and shrubs and lianas (both making up 11%
of species). At Los Volcanes, trees were the most
species-rich life-form in the deciduous and semideciduous forest but contributed only 32% of all
species. Epiphytes (including hemiepiphytes) and
lianas (23% and 22% in the deciduous forest, 22%
and 23% in the semi-deciduous forest, respectively)
followed them closely. Shrubs were less important in
these forests, contributing 13% and 15% of all
species in the deciduous and semi-deciduous forest,
respectively, and having percentages only slightly
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
Area (ha)
Fig. 2 Species accumulation curves (sac) for each forest type
at Los Volcanes (D = deciduous, SD = semi-deciduous,
E = evergreen) against species richness data of: a vascular
plant counts, b epiphyte inventories, c liana inventories, d
0.4
0.6
0.8
1.0
Area (ha)
terrestrial herb inventories and e woody plant inventories in the
neotropics. Symbols described in the legend apply to the entire
figure. Note the different scales of the x and y axes. (Values for
individual sites are provided in Appendix 1)
200
150
100
50
No. Species
Los Volcanes (dbh ≥ 10)
Lowland humid, Bolivia
Lowland deciduous, Bolivia
Montane humid, Bolivia
Los Volcanes (all trees)
Lowland humid, Neotropics (n=22)
Lowland deciduous, Neotropics (n=15)
Montane humid, Neotropics (n=15)
0
vascular plant counts at Los Volcanes are similar to
those in the humid mountains of the Carrasco National
Park (Ibisch 1996) and those in the deciduous forest of
Capeira and moist forest of Jauneche in Ecuador
(Gentry and Dodson 1987), at least at small plot sizes
(Fig. 2a). Epiphytes are perhaps the best represented
group at Los Volcanes. They are more diverse here
than a humid forest plot in Ecuador (Jauneche, Gentry
and Dodson 1987) and at least as diverse as the humid
forest plots in French Guyana (Sinamary, Bordenave
et al. 1998), Venezuela (Surumoni, Nieder et al.
2000) and interestingly also as the Chocoan forests
in Caquetá (Colombia, Duivenvoorden 1994). They
also have higher species richness than other deciduous
forests and similar species richness as montane forests
(Fig. 2b). Lianas have similar levels of species
richness to several humid forests in Ecuador (Yasunı́,
Nabe-Nielsen 2001; Burnham 2004), Colombia (Nuqui and Coqui, Galeano et al. 1998) and Bolivia
(Oquiriquia, Pérez-Salicrup et al. 2001). In contrast,
the deciduous forest of Capeira in coastal Ecuador
(Gentry and Dodson 1987) has a much higher species
richness of lianas in smaller plots (Fig. 2c). Terrestrial
herb counts are similar to other humid forests from
Costa Rica (Whitmore et al. 1985), Panama (Royo
and Carson 2005), French Guyana (Bordenave et al.
1998), Colombia (Galeano et al. 1998), Ecuador
(Gentry and Dodson 1987; Poulsen and Balslev
1991) and Brazil (Costa 2004). The deciduous forest
in Capeira (Gentry and Dodson 1987) again has a
higher count of terrestrial herbs than any plot at Los
Volcanes (Fig. 2d).
Tree species richness at Los Volcanes is similar to
several other forest types in the neotropics. If only
trees with dbh C10 cm are considered, the plots at Los
Volcanes are similar to other deciduous forests in
Bolivia and reach the lower end of species richness of
1-ha plots in humid and montane forests (Figs. 2e, 3).
If we compare the overall richness of trees on the Los
Volcanes plots (including those with dbh \10 cm)
with several forest types inventoried by A. Gentry
(0.1-ha transects using the exploded quadrat method,
available at http://www.mobot.org/MOBOT/research/
gentry/transect.shtml), the studied plots have higher
species richness values than most dry deciduous forests in the neotropics. They have also values similar to
the average species richness of humid montane forests
but only reach the lower end of the species richness
values of humid lowland forest (Fig. 3).
A.G. Van der Valk (ed.)
250
96
Fig. 3 Species richness values of woody plants and trees at
Los Volcanes (LV) with dbh C10 cm compared against species
richness data of other woody plant inventories (1 ha,
dbh C10 cm) in Bolivia (filled symbols), and all trees
inventoried at Los Volcanes compared to species richness data
of woody plant inventories (0.1 ha, dbh C 2.5 cm) in the
neotropics (empty symbols)
Contribution of non-woody plant groups
to overall plant species richness
Non-woody plants, and specifically epiphytes, have
been highlighted by Gentry and Dodson (1987) as the
most important plant group in terms of species
richness and individual numbers in wet tropical rain
forests in Ecuador, whereas tree species with dbh
C10 cm were more or less equally well represented
in dry, moist and wet forest. In the wet forest sampled
by them, 35% of the species and 49% of the
individuals were epiphytes. At Los Volcanes, epiphytes (including hemiepiphytes) included ca. 30%
of the species and nearly 60% of the individuals in
the evergreen forest plot. If terrestrial herbs are
included, non-woody life-forms represent 45% of the
species and more than 76% of the individuals. In the
semi-deciduous and deciduous forest plots, figures
are somewhat lower but still impressive. Epiphytes
(including hemiepiphytes) comprised 22% and 23%
of the species and 36% and 31% of the individuals on
the semi-deciduous and deciduous forest plots,
respectively. If terrestrial herbs are included, the
figures are ca. 32% of the species in both forests and
58% and 52% of the individuals, respectively. Nonwoody life-forms showed a consistent pattern across
the different forest types and represented an important component of neotropical forests. This is
Forest Ecology
certainly overlooked when plots are only sampled for
trees. In contrast, trees with dbh C10 cm represented
only 14%, 18% and 16% of the species and 2.4%,
2.5% and 1.7% of the individuals on our deciduous,
semi-deciduous and evergreen forest plot,
respectively.
Despite these facts, species richness and floristic
data arising from tree inventories are often used (for
want of better and more complete, but also usually
more work-intensive, information) to characterize all
the surrounding vegetation because they are the
major structural element (e.g. ter Steege et al. 2000a;
La Torre-Cuadros et al. 2007). While this may be
enough to identify the major forest types, and indeed
some studies in tropical forests do confirmed a
positive correlation between the richness of the
woody component of a forest and its accompanying
non-woody component (e.g. Webb et al. 1967), there
is also evidence that this is not a consistent pattern
(Duivenvoorden and Lips 1995 in Colombia; ter
Steege et al. 2000b in Guyana; Williams-Linera et al.
2005 in Mexico; Tchouto et al. 2006 in Cameroon).
Our own data indicate that there is no consistent
correlation between tree species richness and the
other life-forms studied at Los Volcanes. The comparisons of inventories across the neotropics
discussed above also show that forests with higher
tree species richness do not necessarily contain higher
species richness in non-woody life-forms. In a
neotropical context, the forests at Los Volcanes
may be poor in terms of tree species richness, but
they do show remarkable species richness of lianas,
terrestrial herbs and, especially, epiphytes, challenging even those of the most diverse forests of the
continent, the Colombian Chocó.
We have shown that tree species richness alone
does not always correlate with species richness
patterns in other life-forms (although it consistently
did so with total species richness across the three
forest types studied, as did the species richness
patterns of lianas). We thus advocate that more effort
be put into non-woody plant inventories in order to
better assess the biodiversity of an area and to allow
more informed conservation decisions to be made.
Acknowledgements We thank three anonymous reviewers
who provided helpful critiques of and insightful suggestions to
the manuscript. We thank P. Wilkie for his suggestions and for
correcting the English of the manuscript. We also thank the
owner of Los Volcanes, A. Schwiening, for allowing us to
97
work on his land. SKH and MK are indebted to the Colección
Boliviana de Fauna and the Dirección General de
Biodiversidad for research permits. We thank the curators of
the Herbario Nacional de Bolivia (LPB) and Herbario del
Oriente Boliviano (USZ) for providing us with working
facilities and allowing access to their collections. Several
botanists aided in species identification (S. Beck, T. Krömer, J.
F. Morales, M. Nee, C. Taylor, R. Vasquez). We are grateful to
C. Hamel and M. Valverde for help in the field. Financial
support
is
acknowledged
from
the
Deutsche
Forschungsgemeinschaft (DFG) and the German Academic
Exchange Service (DAAD).
Open Access This article is distributed under the terms of the
Creative Commons Attribution Noncommercial License which
permits any noncommercial use, distribution, and reproduction
in any medium, provided the original author(s) and source are
credited.
References
Álvarez E et al (2003) Conteo total de plantas vasculares en
parcelas de 0.1 ha en Bosques de Amazonı́a, Chocó y los
Andes de Colombia. In: Romero-Fernández L, LuceroMosquera H, Aguirre-Mendoza Z et al (eds) Libro de
Resúmenes II Congreso de Conservación de la Biodiversidad en los Andes y la Amazonia y IV Congreso
Ecuatoriano de Botánica. Editoral de la Universidad
Técnica Particular de Loja, Loja, Ecuador, p 171
Arévalo R, Betancur J (2004) Diversidad de epı́fitas vasculares
en cuatro bosques del sector Suroriental de la Serranı́a de
Chiribiquete, Guayana Colombiana. Caldasia 26:359–380
Balslev H, Valencia R, Paz y Miño G et al (1998) Species
count of vascular plants in 1-hectare of humid lowland
forest in Amazonian Ecuador. In: Dallmeier F, Comiskey
JA (eds) Forest biodiversity in North, Central and South
America and the Caribbean: research and monitoring.
Man and the biosphere series, vol 21. UNESCO and
Parthenon, Lancashire
Benavides A, Duque A, Duivenvoorden J et al (2005) A first
quantitative census of vascular epiphytes in rain forests of
Colombian Amazonia. Biodivers Conserv 14:739–758.
doi:10.1007/s10531-004-3920-9
Bonnet A (2006) Caracterização fitossociológica das bromeliáceas epifı́ticas e suas relações com os fatores
geomorfológicos e pedológicos da planı́cie do Rio Iguaçu,
Paraná, Brasil. Dissertation, Universidade Federal do Paraná, Brazil
Bonnet A, Curcio GR, Barddal ML et al (2007) Estratificação
vertical de bromélias epifı́ticas na planı́cie do rio Iguaçu,
Paraná, Brasil. Rev Bras Biociencias 5:492–494
Bordenave BG, De Granville JJ, Hoff M (1998) Measurement
of species richness of vascular plants in a neotropical rain
forest in French Guiana. In: Dallmeier F, Comiskey J
(eds) Forest biodiversity, research, monitoring and modelling: conceptual background and Old World case
studies. Man and the biosphere series, vol 20. UNESCO
and Parthenon, Paris
98
Burnham RJ (2004) Alpha and beta diversity of lianas in
Yasunı́ National Park, Ecuador. For Ecol Manag 190:43–
55. doi:10.1016/j.foreco.2003.10.005
Chave J (2004) Neutral theory and community ecology. Ecol
Lett 7:241–253. doi:10.1111/j.1461-0248.2003.00566.x
Colwell RK (2005) EstimateS: statistical estimation of species
richness and shared species from samples. Version 7.5.
User’s guide and application published at: http://purl.oclc.
org/estimates. Accessed 31 May 2007
Condit R (1995) Research in large, long-term tropical forest
plots. Trends Ecol Evol 10:18–22. doi:10.1016/S01695347(00)88955-7
Condit RG, Hubbell SP, Lafrankie JV et al (1996) Species-area
and species-individual relationships for tropical trees: a
comparison of three 50-ha plots. J Ecol 84:549–562. doi:
10.2307/2261477
Condit R, Ashton PS, Baker P et al (2000) Spatial patterns in
the distribution of tropical tree species. Science
288:1414–1418. doi:10.1126/science.288.5470.1414
Costa FRC (2004) Structure and composition of the groundherb community in a terra-firme Central Amazonian forest. Acta Amazon 34:53–59
Duivenvoorden JF (1994) Vascular plant species counts in the
rain forests of the middle Caquetá area. Colombian
Amazonia. Biodivers Conserv 3:685–715. doi:10.1007/BF
00126860
Duivenvoorden JF, Lips JM (1995) A land ecological study of
soils, vegetation and plant diversity in Colombian
Amazonia. Tropenbos series 12. The Tropenbos Foundation, Wageningen
Galeano G, Suárez S, Balslev H (1998) Vascular plant species
count in a wet forest in the Chocó area on the Pacific coast
of Colombia. Biodivers Conserv 7:1563–1575. doi:
10.1023/A:1008802624275
Gaston KJ, Chown SL (2005) Neutrality and niche. Funct Ecol
19:1–6. doi:10.1111/j.0269-8463.2005.00948.x
Gentry AH (1982) Patterns of neotropical plant species
diversity. Evol Biol 15:1–84
Gentry AH (1988) Tree species richness of upper Amazonian
forests. Proc Natl Acad Sci USA 85:156–159. doi:10.1073/
pnas.85.1.156
Gentry AH (ed) (1990) Four neotropical rainforests. Yale
University Press, New Haven
Gentry AH, Dodson C (1987) Contribution of nontrees to
species richness of a tropical rain forest. Biotropica
19:149–156. doi:10.2307/2388737
Griffiths H, Smith JAC (1983) Photosynthetic pathways in the
Bromeliaceae of Trinidad: relations between life-forms,
habitat preference and the occurrence of CAM. Oecologia
60:176–184. doi:10.1007/BF00379519
Holmgren PK, Holmgren NH (1998) Index herbariorum: a
global directory of public herbaria and associated staff.
New York Botanical Garden’s Virtual Herbarium.
http://sweetgum.nybg.org/ih/. Cited 20 May 2007 (continuously updated)
Ibisch PL (1996) Neotropische Epiphytendiversität, das Beispiel Bolivien. Martina-Galunder-Verlag, Wiehl
Ibisch PL, Beck SG, Gerkmann B et al (2003) Ecoregiones y
ecosistemas. In: Ibisch PL, Mérida G (eds) Biodiversidad:
la riqueza de Bolivia. Estado de conocimiento y
A.G. Van der Valk (ed.)
conservación. Ministerio de Desarrollo Sostenible y
Planificación, Editorial FAN, Santa Cruz
Ingram SW, Ferrell-Ingram K, Nadkarni NM (1996) Floristic
composition of vascular epiphytes in a neotropical cloud
forest, Monteverde, Costa Rica. Selbyana 17:88–103
John R, Dalling JW, Harms KE et al (2007) Soil nutrients
influence spatial distributions of tropical tree species. Proc
Natl Acad Sci USA 104:864–869. doi:10.1073/pnas.
0604666104
Jørgensen PM, Ulloa C, Maldonado C (2006) Riqueza de
plantas vasculares. In: Moraes M, Øllgaard B, Kvist LP
et al (eds) Botánica Económica de los Andes Centrales.
Universidad Mayor de San Andrés, Plural Editores, La
Paz
Krömer T, Kessler M, Gradstein SR et al (2005) Diversity
patterns of vascular epiphytes along an elevational gradient in the Andes. J Biogeogr 32:1799–1810. doi:
10.1111/j.1365-2699.2005.01318.x
Langenberger G, Martin K, Sauerborn J (2006) Vascular plant
species inventory of a Philippine lowland rain forest and
its conservation value. Biodivers Conserv 15:1271–1301.
doi:10.1007/s10531-005-2576-4
Laurance WF, Ferreira LV, Rankin-De Merona JM et al (1998)
Influence of plot shape on estimates of tree diversity and
community composition in Central Amazonia. Biotropica
30:662–665. doi:10.1111/j.1744-7429.1998.tb00106.x
La Torre-Cuadros MA, Herrando-Pérez S, Young KR (2007)
Diversity and structural patterns for tropical montane and
premontane forests of central Peru, with an assessment of
the use of higher-taxon surrogacy. Biodivers Conserv
16:2965–2988. doi:10.1007/s10531-007-9155-9
Magurran AE (2004) Measuring biological diversity. Blackwell Science, Oxford
Mascaro J, Schnitzer SA, Carson WP (2004) Liana diversity,
abundance, and mortality in a tropical wet forest in Costa
Rica. For Ecol Manag 190:3–14. doi:10.1016/j.foreco.
2003.10.002
Nabe-Nielsen J (2001) Diversity and distribution of lianas in a
neotropical rain forest, Yasunı́ National Park, Ecuador. J
Trop Ecol 17:1–19. doi:10.1017/S0266467401001018
Navarro G (2001) Contribución al conocimiento fitosociológico de la vegetación de epı́fitos vasculares del centro y sur
de Bolivia. Rev Bol Ecol 10:59–79
Navarro G, Vargas I, Jardim A et al (1996) Clasificación y
diagnóstico para la conservación de la vegetación de la
región del Parque Nacional Amboró. Santa Cruz, Bolivia.
Universidad Complutense, Fundación Amigos de la Naturaleza, Museo de Historia Natural Noel Kempff
Mercado, Universidad Mayor de San Andrés and New
York Botanical Garden, Santa Cruz
Nieder J, Engwald S, Klawun M et al (2000) Spatial distribution of vascular epiphytes (including hemiepiphytes) in a
lowland Amazonian rain forest (Suromoni Crane Plot) of
Southern Venezuela. Biotropica 32:385–396
Oksanen J, Kindt R, Legendre P et al (2006) vegan: Community Ecology Package version 1.8-3. http://cc.oulu.fi/
*jarioksa/. Accessed 10 Oct 2006
Pérez-Salicrup DR, Sork VL, Putz FE (2001) Lianas and trees
in a liana forest of Amazonian Bolivia. Biotropica 33:34–
47
Forest Ecology
Poulsen AD, Balslev H (1991) Abundance and cover of ground
herbs in an Amazonian rain forest. J Veg Sci 2:315–322.
doi:10.2307/3235922
Poulsen AD, Nielsen IH (1995) How many ferns are there in
one hectare of tropical rain forest? Am Fern J 85:29–35.
doi:10.2307/1547678
R Development Core Team (2006) R: a language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna. ISBN 3-900051-07-0,
URL http://www.R-project.org
Royo AA, Carson WP (2005) The herb community of a tropical
forest in central Panama: dynamics and impact of mammalian herbivores. Oecologia 145:66–75. doi:10.1007/
s00442-005-0079-3
Schuettpelz E, Pryer KM (2007) Fern phylogeny inferred from
400 leptosporangiate species and three plastid genes.
Taxon 56:1037–1050
Smith AR, Pryer KM, Schuettpelz E et al (2006) A classification for extant ferns. Taxon 55:705–731
Smith DN, Killleen TJ (1998) A comparison of the structure
and composition of montane and lowland in the Serranı́a
Pilon Lajas, Beni, Bolivia. In: Dallmeier F, Comiskey JA
(eds) Forest biodiversity in North, Central and South
America, and the Caribbean: research and monitoring.
Man and the biosphere series, vol 21. UNESCO and
Parthenon, Lancashire
Smith LB, Downs RJ (1977) Tillansioideae (Bromeliaceae).
Flora Neotrop Monogr 4:663–1492
Tchouto MGP, de Boer WF, de Wilde JJFE et al (2006)
Diversity patterns in the flora of the Campo-Ma’an rain
forest, Cameroon: do tree species tell it all? Biodivers
Conserv 15:1353–1374. doi:10.1007/s10531-005-5394-9
99
ter Steege H, Sabatier D, Castellanos H et al (2000a) An
analysis of the floristic composition and diversity of
Amazonian forests including those of the Guiana Shield. J
Trop Ecol 16:801–828. doi:10.1017/S0266467400001735
ter Steege H, Ek R, van Andel T (2000b) A comparison of
diversity patterns of tree and non-tree groups. In: ter
Steege H (ed) Plant diversity in Guyana. With recommendation for a protected areas strategy. Tropenbos series
18. Tropenbos Foundation, Wageningen
Tobler MW, Janovec J, Honorio E et al (2007) Implications of
collection patterns of botanical specimens on their usefulness for conservation planning: an example of two
neotropical plant families (Moraceae and Myristicaceae)
in Peru. Biodivers Conserv 16:659–677. doi:10.1007/
s10531-005-3373-9
Valencia R, Balslev H, Paz y Miño G (1994) High tree alphadiversity in Amazonian Ecuador. Biodivers Conserv 3:21–
28. doi:10.1007/BF00115330
Webb LJ, Tracey JG, Williams WT et al (1967) Studies in the
numerical analysis of complex rain-forest communities:
II. The problem of species-sampling. J Ecol 55:525–538.
doi:10.2307/2257891
Whitmore TC, Peralta R, Brown K (1985) Total species count
in a Costa Rican tropical rain forest. J Trop Ecol 1:375–
378
Williams-Linera G, Palacios-Rios M, Hernández-Gomez R
(2005) Fern richness, tree species surrogacy, and fragment
complementarity in a Mexican tropical montane cloud
forest. Biodivers Conserv 14:119–133. doi:10.1007/
s10531-005-4053-5
Relationships between spatial configuration of tropical
forest patches and woody plant diversity in northeastern
Puerto Rico
Ileana T. Galanes Æ John R. Thomlinson
Originally published in the journal Plant Ecology, Volume 201, No. 1, 101–113.
DOI: 10.1007/s11258-008-9475-1 Springer Science+Business Media B.V. 2008
I. T. Galanes (&)
Department of Biological Science, Faculty of General
Studies, University of Puerto Rico, Rı́o Piedras Campus,
P.O. Box 23323, San Juan, PR 00931-3323, USA
e-mail: ileana.galanes@vmail.uprrp.edu;
i.galanes@gmail.com
pasturelands. The spatial data were obtained from
aerial color photographs from year 2000. Each photo
interpretation was digitized into a GIS package, and
12 forest patches (24–34 years old) were selected
within a study area of 28 km2. The woody plant
species composition of the patches was determined
by a systematic floristic survey. The species diversity (Shannon index) and species richness of woody
plants correlated positively with the area and the
shape of the forest patch. Larger patches, and
patches with more habitat edge or convolution,
provided conditions for a higher diversity of woody
plants. Moreover, the distance of the forest patches
to the LEF, which is a source of propagules,
correlated negatively with species richness. Plant
species composition was also related to patch size
and shape and distance to the LEF. These results
indicate that there is a link between landscape
structure and species diversity and composition and
that patches that have similar area, shape, and
distance to the LEF provide similar conditions for
the existence of a particular plant community. In
addition, forest patches that were closer together had
more similarity in woody plant species composition
than patches that were farther apart, suggesting that
seed dispersal for some species is limited at the
scale of 10 km.
J. R. Thomlinson
Department of Biology, California State University
Dominguez Hills, 1000 E. Victoria St., Carson,
CA 90747, USA
Keywords Biodiversity Landscape structure
Plant species composition Tropical moist forests
Patch area Patch shape
Abstract The destruction and fragmentation of
tropical forests are major sources of global
biodiversity loss. A better understanding of anthropogenically altered landscapes and their relationships with species diversity and composition is
needed in order to protect biodiversity in these
environments. The spatial patterns of a landscape
may control the ecological processes that shape
species diversity and composition. However, there is
little information about how plant diversity varies
with the spatial configuration of forest patches
especially in fragmented tropical habitats. The
northeastern part of Puerto Rico provides the
opportunity to study the relationships between
species richness and composition of woody plants
(shrubs and trees) and spatial variables [i.e., patch
area and shape, patch isolation, connectivity, and
distance to the Luquillo Experimental Forest (LEF)]
in tropical forest patches that have regenerated from
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_9
101
102
Introduction
Human land-use practices and the accelerating rate of
human population growth are negatively impacting
ecosystems and landscapes on a global scale. Landuse change is projected to have the largest worldwide
impact on biodiversity by the year 2100, especially in
the tropics (Sala et al. 2000). Tropical forests are the
world’s richest in terms of species number; thus, their
destruction is a major source of global loss of species
(Lugo 1988; Brown and Lugo 1990; du Toit et al.
2004). The rate of humid tropical deforestation
between 1990 and 1997 was approximately 5.8 million ha annually, and 2.3 million ha of forest was
visibly degraded as observed in satellite imagery
(Achard et al. 2002). According to McCloskey (1993),
two-thirds of tropical forests have been fragmented
and are especially vulnerable. Scattered and isolated
forest patches of less than 100 ha are found in many
tropical regions, restricting the dispersal capacity of
organisms and reducing their habitat (Turner and
Corlett 1996). Patch size is related positively to the
presence of interior species (Grashof-Bokdam 1997;
Bender et al. 1998). Thus, small forest patches may
lack the interior conditions necessary for the survival
of some species that require the remoteness from the
surroundings or particular microclimate conditions
(Kremen et al. 1994; Forman 1999). In addition,
larger areas support larger populations, which are
associated to lower extinction rates (Rosenzweig
1995). These forest patches need to be managed
appropriately in order to prevent future species
extinctions, given that the main goal of conservation management is to maintain species diversity
(Coleman et al. 1996).
Spatial patterns at landscape scales may control
the ecological processes that affect species richness
and composition (Turner 1989; Haines-Young and
Chopping 1996; Gustafson 1998; de Blois et al. 2002;
Opdam et al. 2003). Studies have shown that the size,
shape, and degree of connectivity of habitat patches
influence patterns of species diversity and abundance
due to the effects of spatial patterns on the dispersal,
distribution, and persistence of species (Burkey 1988;
Turner 1989; Bierregaard et al. 1992; Pearson 1993;
Beier and Noss 1998; Gibbs 1998; Mazerolle and
Villard 1999; Jeanneret et al. 2003; Waldhardt 2003).
In addition, landscape-scale studies have shown that
matrix attributes are important for the dispersal of
A.G. Van der Valk (ed.)
plants. For example, flying seed dispersers and
pollinators may use small fragments or solitary trees
as ‘‘stepping stones’’ to move between forest patches
(Tewksbury et al. 2002; Murphy and Lovett-Doust
2004; Turner 2005).
Conservation planning in anthropogenically fragmented landscapes must include a better understanding of biodiversity patterns and spatial relationships
at a landscape scale (Wu and Hobbs 2002; Waldhardt
2003). However, there is little information on how
plant species diversity and composition vary with
changes in the parameters of landscape structure,
especially in fragmented tropical habitats (Laurance
et al. 1998; Metzger 1997, 2000; de Blois et al. 2002;
McGarigal and Cushman 2002; Hernandez-Stefanoni
2006).
The purpose of this research was to determine how
woody plant species diversity and composition of
forest patches in northeastern Puerto Rico relate to
spatial variables such as patch area and shape and the
degree of connectivity or isolation between forested
areas. Several questions that need to be addressed for
the appropriate management and conservation of
tropical forest patches were considered in this study:
(1) Do forest patches near a reserve, which is a rich
source of propagules, have a higher diversity of
woody plants? (2) Do larger forest patches and/or
patches with a convoluted shape have a higher diversity of woody plants? (3) Do forest patches with
vegetation corridors and/or a vegetated buffer zone
have greater woody plant diversity? (4) Which spatial
variables are related to forest patches with similar
woody plant species composition? (5) Do closer
forest patches are more similar in species composition than forest patches farther apart?
The northeastern part of Puerto Rico was selected
for this study, because the landscape in this location
has become a mosaic of forest patches and corridors
in a matrix of mixed urban and pasture, as a result of
human intervention. These forest patches regenerated
from pasturelands that were abandoned during the
1950s through the 1970s (Aide et al. 1995; Thomlinson et al. 1996; Chinea and Helmer 2003). During
this period the economy of Puerto Rico shifted from
an agrarian economy to an industrial one (Dietz 1986;
Grau et al. 2003). Thus, the chosen site provided an
excellent opportunity to study the spatial configuration of forest patches with more or less equal age of
abandonment, similar land-use history and edaphic
Forest Ecology
conditions. In addition, evolutionary factors that may
affect species composition were minimized in this
study because of the young age of the patches.
There is a need to consider these forest patches in
conservation and management efforts because of
their role in sustaining species diversity and as a
source of propagules at a time of accelerating
increase of urban development in this part of Puerto
Rico (Helmer 2004; Lugo and Helmer 2004) and in
many tropical areas of the world (Achard et al.
2002). In addition, subtropical moist forests in
Puerto Rico are the least protected among the six
ecological life zones found in this Caribbean island
(only 3.2% are protected), and land development
occurs mostly in this ecological zone (Helmer 2004;
Lugo 2006).
Fig. 1 Study site in
northeastern Puerto Rico.
The polygons, digitized
from aerial photographs,
represent land-use classes
as shown in the legend. The
12 forest patches examined
are marked by a circle. The
Luquillo Experimental
Forest (LEF) reserve is
indicated
103
Materials and methods
Study site
The northeastern part of Puerto Rico (between
18220 N, 65470 W and 18170 N, 65570 W) was
selected for this study (Fig. 1). The study site includes
part of the municipalities of Carolina, Canóvanas, and
Rı́o Grande and has an area of 28 km2. It is adjacent to
the northwestern part of the Luquillo Experimental
Forest (LEF), a rainforest reserve of 11,491 ha. The
studied forest patches are located in the subtropical
moist forest life zone (Ewel and Whitmore 1973) and
have a mean elevation ranging from 25 to 105 m above
sea level. The soil of these forest patches is predominantly silty clay loam inceptisols (USDA Soil
104
A.G. Van der Valk (ed.)
Conservation Service 1977). The mean annual rainfall
ranges from 1,893 mm at Canóvanas to 2,506 mm at
Rı́o Grande, and the average annual temperature
minimum and maximum at the Canóvanas Station
are 23 and 30C, respectively (Puerto Rico Station
Data Inventory, NOAA/NWS cooperative observer
network).
3.
4.
5.
6.
Aerial photo interpretation
The spatial configuration of the study site was obtained
from true-color aerial photographs at a scale of
1:20,000 taken in year 2000. Land use was delineated
on 10 of these 9 9 9 inch photographs. A mirror
stereoscope was used to delineate polygons with a
minimum mapping unit of 0.36 ha, equivalent to
3 9 3 mm on the photographs.
Each polygon was assigned one of eight land-use
classes as follows: dense forest (80% or more of forest
cover), less dense forest (between 80% and 20% of
forest cover), agriculture, pasture, dense urban (80% or
more of construction), less dense urban (less than 80%
of construction), cleared land, and others (business,
factories, stable buildings, poultry farms). The polygons were rectified and digitized into a geographical
information system (ArcView version 3.2, ESRI 1999)
(Fig. 1).
Selection of forest patches
Soil maps from the United States Department of
Agriculture, Soil Conservation Service (1977) were
used to determine the type of soil and topography for
the studied forest patches. Aerial photographs taken
in years 1936, 1964, 1971, 1981, and 1990 were used
to determine the land-use history and age of these
patches. Twelve dense forest patches (80% of forest
cover or more) were selected according to the criteria
of same soil type and similar topography, state of
secondary succession (between 24 and 34 years since
abandonment), and land-use history (sugar cane until
the 1950s and then pasture). This was to minimize the
effects of these factors on the species composition. In
addition, each forest patch was classified according to
the following seven spatial variables:
1.
2.
Patch size.
Patch shape, calculated with the Shoreline
Development Index (SLD) (Patton 1975).
7.
Mean elevation.
Distance to the LEF.
Isolation index (Whitcomb et al. 1981) within a
600-m radius from the center of the study patch.
This radius was chosen because it assured that
there was at least one other forest patch within
the radius.
Percent of forest area (defined as polygons with
[33% forest cover) within the matrix, in a buffer
of 50 m radius around the forest patch. Forest
area in the buffer was determined as the proportion of dots on a 1-mm grid transparent overlay
that intercepted tree crowns on the aerial photo.
Connectivity: the number of vegetation corridors
from each patch that connected to other forest
patches. A corridor was defined as a continuous
strip between patches that showed a distinct
narrowing compared to a patch.
Forest patch sampling
Woody plant species composition (shrubs and trees
C1 cm diameter and [1.3 m height) of the 12 study
patches was determined by a systematic floristic
survey. The survey consisted of several belt transects
of 2 m 9 50 m spaced regularly in the interior and at
the edge of each patch, perpendicular to the topographic gradient. Cumulative species–area curves
were constructed. Rarefaction curves based on
individuals and samples were also created using the
software package EstimateS, version 6 (Colwell
2000). These rarefaction and species–area curves
aided in determining the number of transects needed
to sample each forest patch; typically four to six
transects were needed. Collected specimens of each
species were identified in the University of Puerto
Rico, Rı́o Piedras herbarium, and voucher specimens
were deposited.
Data analysis
Woody plant species diversity (i.e., species number
and relative abundance or evenness) and species
composition of the 12 study patches were related to
patch size, patch shape, distance between each forest
patch and the LEF, degree of isolation from other
forest patches (within a 600-m radius), percent of
forest in the matrix around the patch (within a buffer
Forest Ecology
105
of 50 m radius), and number of vegetation corridors
connected to other forest patches. Differences in
woody plant species composition were also related to
the geographic distance among the 12 forest patches.
Plant species diversity for each forest patch was
measured using species richness (S) and the Shannon
diversity index (H0 ). The relative importance of all
the spatial variables in relation to species diversity
was analyzed using a stepwise regression analysis of
H0 (dependent variable) against the spatial variables
of each patch. This analysis has the advantage that it
reduces the multicollinearity among the explanatory
variables (Draper and Smith 1981). The analysis was
repeated with species richness as the dependent
variable.
Nonmetric multidimensional scaling (NMS) analysis (Mather 1976; Kruskal 1964) was used to compare
dissimilarity among the study patches in relation to
plant species composition and to correlate these
dissimilarities with the spatial variables. The computer
program PC-ORD version 3.18 (McCune and Mefford
1997) was used for multivariate analysis of the data. A
two-way samples-by-species data matrix was constructed using the abundance values of each woody
plant species within each forest patch. The data were
relativized by the species maximum to reduce the
coefficient of variation between species columns, in
this case from 297% to 53%, well below the value of
100% that is considered to interfere with the
multivariate analysis (McCune and Grace 2002). This
adjustment tends to equalize rare and abundant
species, giving more weight to rare species that are
frequently encountered in tropical forests. An additional two-way samples-by-variables matrix was
constructed, which contained the values of the spatial
variables for each forest patch (Table 1). The ordination was done using the Sørensen dissimilarity
coefficient (Bray and Curtis 1957).
The NMS analysis was run with a random starting
configuration and stepping down in dimensionality
from six axes to one axis. The number of runs with
real data was 40 and with randomized runs was 50.
The maximum number of iterations was 400 and the
instability criterion was set to 0.00001. Based on the
preliminary runs, the best starting configuration and
dimensionality was selected for the final run; the
number of real runs was set to 1 and the maximum
number of iterations was set to 500.
A Mantel test (Mantel 1967) was performed to test
the correspondence between the geographic distances
(km) among the 12 forest patches and the Sørensen
dissimilarity measures among the patches’ woody
plant species composition. The species abundance
matrix used for this analysis was the same one used for
the NMS analysis. An additional two-way samples-bydistance matrix was constructed, which contained the
geographic distances (km) among the 12 forest
patches. Mantel’s asymptotic approximation (Douglas
Table 1 Forest patch spatial variables: area (ha), shape
(SLD), distance (km) to Luquillo Experimental Forest
(LEF), elevation (meters above sea level), isolation index
(600 m radius), percent of forest area in the matrix (polygons
[33% of forest cover, in a 50 m radius), and number of
vegetation corridors connected to other forest patches;
dependent variables: plant species richness (S) and Shannon
diversity index (H0 )
Forest
patch no.
Area (ha)
Shape SLD
index
Dist. LEF
(km)
Elevation (m)
Isolation
index
1
33.3
1.5
11.9
45
3.33E-05
10
16.0
1.7
2.7
90
1.23E-04
9
15.6
2.2
2.8
95
3.69E-05
2
15.3
2.0
11.5
75
2.97E-05
6
15.1
2.1
3.9
25
2.17E-03
3
13.9
1.5
9.2
35
3.68E-04
4
11.4
No. of corridors
S
H0
0
33
2.69
23.5
2
25
2.03
19.7
1
29
2.03
34.0
0
23
2.26
0
27
2.37
55.5
0
16
1.51
Forest
matrix (%)
3.19
0.23
1.5
7.8
50
7.14E-04
18.2
0
12
1.52
11
7.58
1.7
2.3
75
3.12E-05
21.0
4
17
1.92
7
7.26
2.1
3.4
45
1.69E-04
10.2
2
22
2.36
5
4.97
1.5
3.9
105
4.74E-03
27.4
1
16
1.63
12
4.05
1.6
2.2
80
7.89E-04
29.9
1
18
1.91
8
1.60
1.5
2.8
85
4.03E-03
0
0
13
1.48
106
A.G. Van der Valk (ed.)
and Endler 1982) was the method used to test the
significance of the correlation between these two
matrices.
Results
The total number of woody plant species found in the
12 forest patches (c diversity) was 69 species from
4,713 individuals sampled. The percentage of native
species was 80%, with an abundance of 92%. The three
most abundant species were Casearia guianensis
(Aubl.) Urb., Tabebuia heterophylla (DC.) Britton,
and Casearia sylvestris Sw. (Appendix A).
The average species richness for each patch
(a diversity) was 21 species. The amount of compositional variation in the 12 patches, or bw diversity
(Whittaker 1972), was 3.3. Values of bw \ 1 are
considered low and bw [ 5 are considered high
(McCune and Grace 2002). In this study each patch
had, on average, 30% of the total number of species
found.
The area of the forest patches ranged from 1.6 to
33.3 ha; the SLD index was from 1.5 to 2.2; the
distance to the LEF varied from 2.2 to 11.9 km; the
isolation index ranged from 2.23E-05 to 4.74E-03;
the percentage of forest area in the matrix (50 m
radius) ranged from 0% to 55.5%; and the number of
vegetation corridors connected to other dense forest
patches ranged from 0 to 4 (Table 1).
Pearson correlation analysis showed no significant
correlations among five of the seven spatial variables
Table 2 Results from
stepwise regression analyses
for (a) Woody-plant Shannon
diversity index and (b)
Woody-plant species richness
in relation to seven spatial
variables of 12 forest patches
Variable
studied. Area and distance to the LEF correlated
positively (r = 0.67, P = 0.02), while the number of
vegetation corridors and distance to the LEF correlated negatively, though it was marginally significant
(r = - 0.58, P = 0.05).
The Shannon diversity index (H0 ) ranged from 1.48
to 2.69 (Table 1). The significant variables included in
the stepwise regression model were area (P = 0.01)
and shape (P = 0.03), both of which were positively
correlated with H0 (Table 2a). Patch area (P = 0.0002)
and shape (P = 0.01) were also positively correlated
with woody plant species richness, and distance to the
LEF correlated negatively (P = 0.04) (Table 2b).
There was no correlation with plant species number
nor H0 and the other spatial variables (number of
vegetation corridors, degree of isolation, elevation,
and percent of forest in the matrix in a 50-m radius).
In the NMS analysis, the final stress for a threedimensional solution was 7.15. A value between 5
and 10 indicates a good ordination with no risk of
drawing false inferences (Clarke 1993). In addition,
the final instability criterion, which should be 0.0005
or less, was 0.00006. The cumulative coefficient of
determination, R2, for the three axes in the NMS
ordination was 88%, and the orthogonality for axes 1
vs. 3 was 98%.
The NMS analysis (Fig. 2) showed that forest
patches with similar woody plant species composition
correlated with the area (r2 = 0.52, s = 0.55) and
shape (r2 = 0.28, s = -0.50) of the forest patch, as
well as distance to the LEF (r2 = 0.60, s = 0.55).
The patch elevation (r2 = 0.39, s = -0.40) and the
Coefficient
STD error
Student’s T
P
VIF
(a) Woody-plant Shannon diversity index vs. 7 spatial variables
Constant
0.41338
0.48775
0.85
Area
2.906E-06
9.113E-07
3.19
0.4187
0.0110
1.0
Shape
0.69314
0.27400
2.53
0.0323
1.0
-0.20
0.8444
R2 = 0.6640
Adjusted R2 = 0.5894
(b) Woody-plant species richness vs. 7 spatial variables
Constant
-1.02108
5.03646
Area
8.019E-05
1.199E-05
6.69
0.0002
2.0
Shape
9.12325
2.73923
3.33
0.0104
1.1
Distance to LEF
-6.907E-04
2.764E-04
-2.50
0.0370
2.0
2
R = 0.9070
See Table 1 for definitions of
the spatial variables
Adjusted R2 = 0.8721
Forest Ecology
Fig. 2 NMS ordination joint plot of axes 1 and 3, for the 12
forest patches-by-plant species abundance matrix (trees and
shrubs), in relation to seven spatial variables. The angles and
lengths of the radiating vectors indicate the direction and
strength of the relationships of the spatial variables with the
ordination scores. See Table 1 for the definitions and values of
the spatial variables
isolation index (r2 = 0.43, s = -0.42) also correlated with the woody plant species composition. On
the other hand, the woody plant species composition
correlated weakly with the percent of forest area in
the matrix around the patch (50 m radius) (r2 = 0.17,
s = 0.27) as well as the number of vegetation
corridors of each forest patch that connected to other
dense forest areas (r2 = 0.22, s = 0.27).
The Mantel test demonstrated a strong positive
correlation between the geographic distances (km)
among the 12 forest patches and the Sørensen
dissimilarity measures among the patches’ woody
plant species composition (P = 0.000016, Standardize Mantel statistic: r = 0.57).
Discussion
Our results show a link between the spatial configuration of tropical forest patches and the diversity and
composition of woody plants. The area, shape, and
distance to the LEF were the spatial variables that
best explained the differences in plant species diversity and composition of the 12 forest patches studied.
The area of the forest patches showed the strongest
correlation with the species diversity and/or richness
followed by patch shape and the distance to the LEF.
On the other hand, the distance of forest patches to
the LEF showed the strongest correlation with similar
107
species composition followed by the patch area and
shape.
The woody plant Shannon diversity and species
richness were higher in larger patches, consistent with
results from European landscapes (Honnay et al. 1999;
Krauss et al. 2004; Økland et al. 2006). In addition, we
found that patches near the LEF (a rich source of
propagules) had higher woody plant species richness
than more distant forest patches. However, we found
no correlation between patch distance to the LEF and
plant species diversity (Shannon index), which considers not only the number of species but also their
relative abundance or species evenness. Similar results
were found in the study of Chinea and Helmer (2003)
in Puerto Rico, where species richness of trees in
tropical forest patches was related to the distance to
larger forest patches while Shannon index was not.
One possible reason for these results is that while the
LEF is a rich source of species, some of those species
may have a low probability of recruitment in the forest
patches studied, resulting in low evenness.
The literature contains few studies that consider the
ecological effects of the shape of a patch (Forman
1999; Hernandez-Stefanoni 2006; Økland et al. 2006).
Patch shape determines the proportion of edge habitat;
i.e., more convoluted patches have more edge habitat
than more compact patches of the same area. However,
not all edge types may be suitable for species survival.
Ratti (1988) observed that feathered edges provided
greater vegetative complexity to birds than abruptedge habitats. He also found that birds were attracted to
the vegetative diversity of edge habitats, but they
experienced greater predator activity in the abruptedge habitats than in feathered edges.
Our study found that the more convoluted a forest
patch is, the higher the woody plant species diversity
and species number. These results agree with the
results found by Økland et al. (2006) in Norway,
where vascular plant species richness was positively
related to patch shape complexity and patch area.
Hernandez-Stefanoni (2006) also found a positive
correlation with patch shape complexity and the
diversity of tropical trees and shrubs. One plausible
explanation for these observations is that a higher
degree of convolution of a forest patch causes greater
spatial heterogeneity due to the junction of two
different habitats and to the complex patterns of
turbulence and wind flow created by the lobes
(Forman 1999; Turner 2005).
108
Habitat heterogeneity is strongly correlated to plant
species richness, since it allows the coexistence of
competing species (Burnett et al. 1998; Dufour et al.
2006). This environmental heterogeneity may have a
positive feedback on diversity, since richer vegetation
provides greater habitat heterogeneity and resources
for other organisms, such as plant pollinators and seed
dispersers, which may further increase the diversity of
the vegetation (Ratti 1988; Wunderle 1997; Turner
2005; Dufour et al. 2006). Patch convolution also
produces drift-fence and cove effects where the form
of the patch enhances the interception of seeds, plant
pollinators, and seed dispersers, increasing the species
diversity of the patch (Forman 1999).
There was no correlation between other spatial
variables (number of vegetation corridors, percent of
forest in the matrix in a 50-m radius, and degree of
isolation in a 600 m radius) and plant species richness
or diversity. This is probably because woody plant
dispersal is not affected significantly by a lack of
nearby vegetation adjacent to a patch, since woody
plants in Puerto Rico are dispersed mostly by flying
animals (e.g., birds, bats) or by wind (Francis and
Lowe 2002). Approximately 74% of the species
sampled in this study are known to be dispersed by
flying animals (mostly birds and, to a lesser extent,
bats) and 10% by wind (Appendix A; T. Carlo 2007,
personal communication, University of Washington,
Seatle, WA; Francis and Lowe 2002).
In this study the dispersal limitation within a
distance of 600 m was low, as was indicated by the
isolation values and the bw diversity obtained. These
results suggest that the 12 forest patches, considered as
a whole, were not sufficiently isolated from wooded
areas to exhibit major limitations of seed dispersal at a
scale of 600 m. Studies reviewed by Sork and Smouse
(2006) showed that physical isolation does not prevent
pollen flow at a distance of 1 km in tropical tree
species, and seed flow into fragments was high even
though there were few seed donors.
However, at a scale of 10 km some species do
appear to be dispersal limited, because we found that
forest patches near LEF had higher plant species
richness than patches farther away. In addition, the
Mantel test demonstrated that forest patches that were
closer had more similarity in woody plant species
composition, and as the geographic distance increased
to a maximum of 10 km, the dissimilarity in species
composition became greater. These results showed
A.G. Van der Valk (ed.)
that distant sites (10 km apart) are less likely to share
species than nearby sites, probably because seed
dispersal for some species is limited at this scale
(Hubbell 1979, 2001; Condit et al. 2002). Although in
this study we selected forest patches with similar
abiotic conditions, the differences in floristic composition as patches are more distant could also be due to
slight differences in soil, temperature, or precipitation. However, Pyke et al. (2001) looked at samples in
a lowland neotropical forest and found that distance
affects plant species variation over relatively small
scales (\5 km), which is close to the scale of our
study, while climate and geology predicted differences in floristic composition at broader scales.
Patches that had similar area and shape had similar
woody plant species composition (Fig. 2). Thus, the
size and shape of a forest patch were important in
providing conditions for the existence of particular
species. Smaller or more convoluted patches have more
habitat edge, proportional to the area, when compared
to larger or less convoluted patches. These smaller
patches may lack the interior conditions necessary for
the survival of some species that require remoteness
from their surroundings or particular microclimate
conditions, for example, low temperature or high
humidity (Kremen et al. 1994; Grashof-Bokdam
1997; Bender et al. 1998; Forman 1999). Normally,
rare and specialist species are found in the interior of a
forest patch, while most species that live in edge
habitats are generalists and common. There are,
however, indifferent species that can thrive in both
interior and edge conditions (Grashof-Bokdam 1997;
Bender et al. 1998; Krauss et al. 2004; Turner 2005).
Forest patches with similar distances to the LEF
had similar woody plant species composition. Forest
patches closer to the LEF are also closer to each
other, allowing the sharing of species. On the other
hand, forest patches near the LEF probably will have
a different source of propagules than patches farther
from the LEF. This can cause differences in shaping
the composition of species that will be established in
a forest patch.
The isolation index in a 600-m radius also correlated
with similar woody plant species composition in the
NMS analysis suggesting that the degree of isolation
may cause differences in the kind of propagules that
will arrive in these patches. The effectiveness of
animal-mediated seed dispersal, which is the major
form of dispersal in our study area, can be limited by the
Forest Ecology
109
degree of isolation (Butaye et al. 2001; Wunderle
1997). Woody plant species composition also correlated with elevation in the NMS analysis. The mean
elevation of the 12 forest patches varied from 25 to
105 m, a range of 80 m. Past studies have shown that
tree species composition varies according to elevation,
especially when the increments in altitude result in
temperature decreases and increases in rainfall and
cloud cover (White 1963; Crow and Grigal 1979;
Weaver 1991; Garcı́a-Martinó et al. 1996). However,
these other studies examined increments in elevation
over ranges of hundreds of meters, where noticeable
differences in rainfall and temperature occur.
The number of vegetation corridors connecting the
forest patches and the percentage of forest in the matrix
(50 m radius surrounding the patch) correlated weakly
with woody plant species composition, indicating that
these factors did not significantly affect the woody
plant species composition of the forest patches.
Conclusions
There are few studies that have examined the relationships between woody plant diversity and species
composition in tropical forest patches and spatial
variables at a landscape level. In this study we found
that there was a strong correlation with the spatial
configuration of tropical forest patches and the woody
plant species diversity and composition in northeastern
Puerto Rico. The variables that best explained the
species diversity and/or richness were patch area and
shape and the distance from the patch to the LEF
reserve. Forest patch area, shape, and distance to the
LEF also correlated with similar woody plant species
composition. We also found that species composition
dissimilarity increased with the geographic distance
between patches at a scale of 10 km. These findings
support the premise that there are strong associations
between landscape elements and the ecological processes involved in shaping patterns of species
diversity. The spatial configuration of forest patches
(patch area and shape and distance among patches)
contributes to the existence of a particular plant species
composition and diversity. The results of this study
shed light on possible explanations for the observed
spatial patterns of woody plant species diversity and
composition at landscape levels. These findings can be
used for the effective management or restoration of
forest patches and for the conservation of woody plant
species, especially in highly fragmented landscapes.
Acknowledgments We thank Mitchell Aide, Nicholas
Brokaw, and Eugenio Santiago for assisting with their advice
and expertise and Jess Zimmerman for his valuable help with the
data analyses. We also thank two anonymous referees for their
constructive reviews that greatly improved this article. Our
profound gratitude goes to Marcos Caraballo for his assistance in
the identification of the species. We also thank Frank Axelrod for
revising all the voucher specimens deposited in the University of
Puerto Rico Rı́o Piedras herbarium. Special thanks are extended
to Pedro J. Rodrı́guez Esquerdo for his valuable insights with the
statistics. We are very grateful to many undergraduate students
from the College of Natural Sciences who made possible this
investigation. We also thank Edwin T. Pérez for his help with the
graphic art of Fig. 1. Our gratitude also goes to the Puerto Rico
Louis Stokes Alliance for Minority Participation for providing
part of the funds for this research and to the Institute for Tropical
Ecosystem Studies for providing the facilities and the support of
their personnel. This research was supported by grants DEB
0080538 and DEB 0218039 from NSF to the Institute for
Tropical Ecosystem Studies, University of Puerto Rico and to the
International Institute of Tropical Forestry USDA Forest
Service, as part of the Long-Term Ecological Research
Program in the Luquillo Experimental Forest.
Appendix A
List of the 69 woody plant species, sampled in the 12 forest patches examined, with species abundance values, species origin, and
mode of dispersal
Species (total: 69)
Origin
Abundance (total: 4,713)
Percent of abundance (%)
Mode of dispersal
Casearia guianensis
Tabebuia heterophylla
Native
Native
1,458
593
30.94
12.58
Bird
Wind
Casearia sylvestris
Native
566
12.01
Bird
Miconia prasina
Native
203
4.31
Bird
Andira inermis
Native
192
4.07
Bat
Syzygium jambos
Exotic
188
3.99
Bat
110
A.G. Van der Valk (ed.)
Appendix continued
Species (total: 69)
Origin
Abundance (total: 4,713)
Percent of abundance (%)
Mode of dispersal
Guapira fragrans
Native
185
3.93
Bird
Eugenia biflora
Native
144
3.06
Bird
Calophyllum antillanum
Myrcia splendens
Native
Native
142
96
3.01
2.04
Bat
Bird
Citharexylum spinosum
Native
89
1.89
Bird
Casearia decandra
Native
85
1.80
Bird
Inga laurina
Native
81
1.72
Gravity and bat
Spathodea campanulata
Exotic
76
1.61
Wind
Guarea guidonia
Native
71
1.51
Bird
Peltophorum pterocarpum
Exotic
57
1.21
Wind
Hura crepitans
Native
40
0.85
Capsule blow up
Cupania americana
Native
36
0.76
Bird
Ocotea leucoxylon
Native
36
0.76
Bird
Faramea occidentalis
Native
35
0.74
Bird
Zanthoxylum martinicense
Native
27
0.57
Bird
Ardisia obovata
Native
21
0.45
Bird
Hymenaea courbaril
Native
21
0.45
Gravity
Trichilia pallida
Native
21
0.45
Bird
Trichilia hirta
Nectranda turbacensis
Native
Native
20
20
0.42
0.42
Bird
Bird
Albizia procera
Exotic
19
0.40
Wind
Randia aculeata
Native
18
0.38
Bird
Chrysophyllum argenteum
Native
16
0.34
Bird
Miconia impetiolaris
Native
15
0.32
Bird
Cordia sulcata
Native
15
0.32
Bird and bat
Picramnia pentandra
Native
13
0.28
Bird
Cinnamomum elongatum
Native
9
0.19
Bird
Cordia collococca
Native
8
0.17
Bird
Petitia domingensis
Native
8
0.17
Bird
Artocarpus altilis
Exotic
8
0.17
Gravity and human
Mangifera indica
Exotic
7
0.15
Gravity and human
Tabernaemontana citrifolia
Native
6
0.13
Bird
Calophyllum inophyllum
Exotic
5
0.11
Bat
Ardisia solanacea
Inga vera
Exotic
Native
4
4
0.08
0.08
Bird
Bat
Clusia rosea
Native
4
0.08
Bird
Piper amalago
Native
3
0.06
Bat
Palicourea crocea
Native
3
0.06
Bird
Leucaena leucocephala
Native
3
0.06
Wind
Swietenia mahagoni 9 macrophylla
Exotic
3
0.06
Wind
Senna siamea
Native
3
0.06
Wind
Miconia racemosa
Native
3
0.06
Bird
Roystonea borinquena
Native
3
0.06
Bird
Adenanthera pavonina
Exotic
3
0.06
Gravity
Bursera simaruba
Native
3
0.06
Bird
Forest Ecology
111
continued
Species (total: 69)
Origin
Abundance (total: 4,713)
Percent of abundance (%)
Mode of dispersal
Coccoloba venosa
Native
3
0.06
Bird
Cecropia schreberiana
Native
2
0.04
Bird and bat
Gonzalagunia hirsuta
Ficus citrifolia
Native
Native
2
2
0.04
0.04
Bird
Bird and bat
Thespesia grandiflora
Endemic
2
0.04
Gravity
Spondias mombin
Native
1
0.02
Bird and bat
Margaritaria nobilis
Native
1
0.02
Capsule blow up
Coccoloba diversifolia
Native
1
0.02
Bird
Annona reticulata
Native
1
0.02
Gravity and human
Delonix regia
Exotic
1
0.02
Gravity
Tournefortia hirsutissima
Native
1
0.02
Bird
Psidium guajava
Native
1
0.02
Bird, bat, and human
Terminalia catappa
Exotic
1
0.02
Bat
Trema micrantha
Native
1
0.02
Bird
Piper jacquemontianum
Native
1
0.02
Bat
Vitex divaricata
Native
1
0.02
Bird
Annona muricata
Native
1
0.02
Gravity and human
Citrus 9 aurantium
Exotic
1
0.02
Gravity and human
Species origin
Abundance
Mode of dispersal of the 69 species examined
Native species: 80%
Native: 92%
Bird: 61%
Endemic species: 1.4%
Endemic: 0.04%
Bat: 13%
Exotic species: 8.8%
Exotic: 7.9%
Wind: 10%
Other (e.g., gravity, human): 16%
References
Achard F, Eva HD, Stibig H, Mayaux P, Gallego J, Richards T
et al (2002) Determination of deforestation rates of the
world’s humid tropical forests. Science 297:999–1002.
doi:10.1126/science.1070656
Aide TM, Zimmerman JK, Herrera L, Rosario M, Serrano M
(1995) Forest recovery in abandoned tropical pastures in
Puerto Rico. For Ecol Manag 77:77–86. doi:10.1016/
0378-1127(95)03576-V
Beier P, Noss RF (1998) Do habitat corridors provide connectivity? Conserv Biol 12:1241–1252. doi:10.1046/j.
1523-1739.1998.98036.x
Bender DJ, Contreras TA, Fahrig L (1998) Habitat loss and
population decline: a meta-analysis of the patch size
effect. Ecology 79:517–533
Bierregaard RO Jr, Lovejoy TE, Kapos V, DosSantos AA,
Hutchings RW (1992) The biological dynamics of tropical
rainforest fragments. Bioscience 42:859–866. doi:10.2307/
1312085
Bray JR, Curtis JT (1957) An ordination of the upland forest
communities in southern Wisconsin. Ecol Monogr
27:325–349. doi:10.2307/1942268
Brown S, Lugo A (1990) Tropical secondary forests. J Trop
Ecol 6:1–32
Burkey TV (1988) Extinction in nature reserves: the effect of
fragmentation and of migration between reserves fragments. Oikos 55:75–81. doi:10.2307/3565875
Burnett MR, August PV, Brown JH Jr, Killingbeck KT (1998)
The influence of geomorphological heterogeneity on
biodiversity I. A patch-scale perspective. Conserv Biol
12:363–370. doi:10.1046/j.1523-1739.1998.96238.x
Butaye J, Jacquemyn H, Hermy M (2001) Differential colonization causing non-random forest plant community
structure in a fragmented agricultural landscape. Ecography 24:369–380. doi:10.1034/j.1600-0587.2001.d01-193.x
Chinea JD, Helmer EH (2003) Diversity and composition of
tropical secondary forests recovering from large-scale
clearing: results from the 1990 inventory in Puerto Rico.
For Ecol Manag 180:227–240. doi:10.1016/S0378-1127
(02)00565-0
Clarke KR (1993) Non-parametric multivariate analyses of
changes in community structure. Aust J Ecol 18:117–143.
doi:10.1111/j.1442-9993.1993.tb00438.x
Coleman WG, Mattice J, Brocksen RW (1996) Soulé’s conservation biology as the foundation for econometric
112
ecosystem management. Conserv Biol 10:1494–1499. doi:
10.1046/j.1523-1739.1996.10061494.x
Colwell RK (2000) EstimateS: statistical estimation of species
richness and shared species from samples. Version 6.
User’s guide and application published at http://purl.oclc.
org/estimates
Condit R, Pitman N, Leigh EG Jr, Chave J, Terborgh J, Foster
RB et al (2002) Beta-diversity in tropical forest trees.
Science 295:666–669. doi:10.1126/science.1066854
Crow TR, Grigal DF (1979) A numerical analysis of arborescent communities in the rain forest of the Luquillo
Mountains, Puerto Rico. Vegetatio 40:135–146
de Blois S, Domon G, Bouchard A (2002) Landscape issues in
plant ecology. Ecography 25:244–256. doi:10.1034/j.
1600-0587.2002.250212.x
Dietz JL (1986) Economic history of Puerto Rico: institutional
change and capitalist development. Princeton University
Press, New Jersey
Douglas ME, Endler JA (1982) Quantitative matrix comparisons in ecological and evolutionary investigations. J Theor
Biol 99:777–795. doi:10.1016/0022-5193(82)90197-7
Draper NR, Smith H (1981) Applied regression analysis, 2nd
edn. Wiley, New York
du Toit JT, Walker BH, Campbell BM (2004) Conserving
tropical nature: current challenges for ecologists. Trends
Ecol Evol 19:12–17. doi:10.1016/j.tree.2003.09.018
Dufour A, Gadallah F, Wagner HH, Guisan A, Buttler A
(2006) Plant species richness and environmental heterogeneity in a mountain landscape: effects of variability and
spatial configuration. Ecography 29:573–584
ESRI (1999) ArcView geographic information system, version
3.2. Environmental Systems Research Institute, Inc,
Redlands
Ewel JJ, Whitmore JL (1973) The ecological life zones of
Puerto Rico and the U.S. Virgin Islands. Research
Paper ITF-018, US Department of Agriculture Forest
Service. International Institute of Tropical Forestry,
Puerto Rico
Forman RT (1999) Land mosaics, the ecology of landscape and
regions. Cambridge University Press, Cambridge
Francis JK, Lowe CA (2002) Bioecologı́a de árboles nativos y
exóticos de Puerto Rico y las Islas Occidentales. US
Department of Agriculture, Forest Service International
Institute of Tropical Forestry, Puerto Rico
Garcı́a-Martinó AR, Warner GS, Scatena FN, Civco DL (1996)
Rainfall, runoff and elevation relationships in the Luquillo
Mountains of Puerto Rico. Caribb J Sci 32:413–424
Gibbs JP (1998) Amphibian movements in response to forest
edges, roads, and streambeds in southern New England. J
Wildl Manag 62:584–589. doi:10.2307/3802333
Grashof-Bokdam C (1997) Forest species in an agricultural
landscape in the Netherlands: effects of habitat fragmentation. J Veg Sci 8:21–28. doi:10.2307/3237238
Grau HR, Aide TM, Zimmerman JK, Thomlinson JR, Helmer
E, Zou X (2003) The ecological consequences of socioeconomic and land-use changes in postagriculture Puerto
Rico. Bioscience 53:1159–1168. doi:10.1641/0006-3568
(2003)053[1159:TECOSA]2.0.CO;2
Gustafson EJ (1998) Quantifying landscape spatial pattern:
what is the state of the art? Ecosystems (NY, Print) 1:143–
156. doi:10.1007/s100219900011
A.G. Van der Valk (ed.)
Haines-Young R, Chopping M (1996) Quantifying landscape
structure of landscape indices and their application to
forested landscapes. Prog Phys Geogr 20:418–445. doi:
10.1177/030913339602000403
Helmer EH (2004) Forest conservation and land development
in Puerto Rico. Landsc Ecol 19:29–40. doi:10.1023/
B:LAND.0000018364.68514.fb
Hernandez-Stefanoni JL (2006) The role of landscape patterns
of habitat types on plant species diversity of a tropical
forest in Mexico. Biodivers Conserv 15:1441–1457. doi:
10.1007/s10531-005-0598-6
Honnay O, Hermy M, Coppin P (1999) Effects of area, age and
diversity of forest patches in Belgium on plant species
richness, and implications for conservation and reforestation. Biol Conserv 87:73–84. doi:10.1016/S0006-3207
(98)00038-X
Hubbell SP (1979) Tree dispersion, abundance, and diversity in
a tropical dry forest. Science 203:1299–1309. doi:
10.1126/science.203.4387.1299
Hubbell SP (2001) The unified neutral theory of biodiversity
and biogeography. Princeton University Press, Princeton
Jeanneret P, Schüpbach B, Luka H (2003) Quantifying the
impact of landscape and habitat features on biodiversity in
cultivated landscapes. Agric Ecosyst Environ 98:311–320.
doi:10.1016/S0167-8809(03)00091-4
Krauss J, Klein AM, Steffan-Dewenter I, Tscharntke T (2004)
Effects of habitat area, isolation, and landscape diversity
on pant species richness of calcareous grasslands. Biodivers Conserv 13:1427–1439. doi:10.1023/B:BIOC.000
0021323.18165.58
Kremen C, Merelender AM, Murphy DD (1994) Ecological
monitoring: a vital need for integrated conservation and
development programs in the tropics. Conserv Biol
8:388–397. doi:10.1046/j.1523-1739.1994.08020388.x
Kruskal JB (1964) Nonmetric multidimensional scaling: a
numerical method. Psychometrika 29:115–129. doi:
10.1007/BF02289694
Laurance WF, Ferreira LV, Rankin-de Merona JM, Laurance
SG (1998) Rain forest fragmentation and the dynamics of
Amazonian tree communities. Ecology 79:2032–2040
Lugo AE (1988) Estimating reductions in the diversity of
tropical forest species. In: Wilson EO (ed) Biodiversity.
National Academy Press, Washington, DC
Lugo AE (2006) Conservación de recursos. In: López Marrero
TM, Villanueva Colón N (eds) Atlas ambiental de Puerto
Rico. La Editorial Universidad de Puerto Rico, San Juan,
pp 124–130
Lugo AE, Helmer E (2004) Emerging forests on abandoned
land: Puerto Rico’s new forests. For Ecol Manag
190:145–161. doi:10.1016/j.foreco.2003.09.012
Mantel N (1967) The detection of disease clustering and generalized regression approach. Cancer Res 27:209–220
Mather PM (1976) Computational methods of multivariate
analysis in physical geography. Wiley, London
Mazerolle MJ, Villard MA (1999) Patch characteristics and
landscape context as predictors of species presence and
abundance: a review. Ecoscience 6:117–124
McCloskey M (1993) Note on the fragmentation of primary
rainforest. Ambio 22:250–251
McCune B, Grace JB (2002) Analysis of ecological communities. MjM Software Design, Oregon
Forest Ecology
McCune B, Mefford MJ (1997) PC-ORD for windows: multivariate analysis of ecological data, version 3.18. MjM
Software, Gleneden Beach
McGarigal K, Cushman SA (2002) Comparative evaluation of
experimental approaches to the study of habitat fragmentation effects. Ecol Appl 12:335–345. doi:10.1890/
1051-0761(2002)012[0335:CEOEAT]2.0.CO;2
Metzger JP (1997) Relationships between landscape structure
and tree species diversity in tropical forests of South-East
Brasil. Landsc Urban Plan 37:29–35. doi:10.1016/S01692046(96)00367-2
Metzger JP (2000) Tree functional group richness and landscape structure in a Brazilian tropical fragmented
landscape. Ecol Appl 10:1147–1161. doi:10.1890/10510761(2000)010[1147:TFGRAL]2.0.CO;2
Murphy HT, Lovett-Doust J (2004) Context and connectivity
in plant metapopulations and landscape mosaics: does the
matrix matter? Oikos 105:3–14. doi:10.1111/j.0030-1299.
2004.12754.x
Økland RH, Bratli H, Dramstad WE, Edvardsen A, Engan G,
Fjellstad W et al (2006) Scale-dependent importance of
environment, land use and landscape structure for species
richness and composition of SE Norwegian modern agricultural landscapes. Landsc Ecol 21:969–987. doi:
10.1007/s10980-006-0005-z
Opdam P, Verboom J, Powels R (2003) Landscape cohesion:
an index for the conservation potential of landscapes for
biodiversity. Landsc Ecol 18:113–126. doi:10.1023/A:
1024429715253
Patton DR (1975) A diversity index for quantifying habitat
edge. Wildl Soc Bull 394:171–173
Pearson SM (1993) The spatial extent and relative influence of
landscape level factors in wintering bird populations.
Landsc Ecol 8:3–18. doi:10.1007/BF00129863
Puerto Rico Station Data Inventory NOAA/NWS Cooperative
Observer Network. http://www.ncdc.noaa.gov/oa/climate/
stationlocator.html. Accessed 17 Dec 2005
Pyke CP, Condit R, Aguilar S, Hernández A (2001) Floristic
composition across a climatic gradient in neotropical lowland forest. J Veg Sci 12:533–566. doi:10.2307/3237007
Ratti JT (1988) Preliminary test of the ecological trap hypothesis. J Wildl Manag 52:484–491. doi:10.2307/3801596
Rosenzweig ML (1995) Species diversity in space and time.
Cambridge University Press, Cambridge
Sala OE, Chapin FSIII, Armesto JJ, Berlow E, Bloomfield J,
Dirzo R et al (2000) Global biodiversity scenarios for the
year 2100. Science 287:1770–1774. doi:10.1126/science.
287.5459.1770
113
Sork VL, Smouse PE (2006) Genetic analysis of landscape
connectivity in tree populations. Landsc Ecol 21:821–836.
doi:10.1007/s10980-005-5415-9
Tewksbury JJ, Levey DJ, Haddad NM (2002) Corridors affect
plants, animals, and their interactions in fragmented
landscape. Proc Natl Acad Sci USA 99:12923–12962. doi:
10.1073/pnas.202242699
Thomlinson JR, Serrano MI, López TM, Aide TM, Zimmerman JK (1996) Land-use dynamics in a post-agricultural
Puerto Rican landscape (1936–1988). Biotropica 28:525–
536. doi:10.2307/2389094
Turner MG (1989) Landscape ecology: the effect of pattern on
process. Annu Rev Ecol Syst 20:171–197. doi:10.1146/
annurev.es.20.110189.001131
Turner MG (2005) Landscape ecology: what is the state of the
science? Ann Rev Ecol Evol 36:319–344. doi:10.1146/
annurev.ecolsys.36.102003.152614
Turner IM, Corlett RT (1996) The conservation value of small,
isolated fragments of lowland tropical rain forest. Trends
Ecol Evol 11:330–333. doi:10.1016/0169-5347(96)10046-X
United States Department of Agriculture Soil Conservation
Service in Cooperation with the University of Puerto Rico
College of Agricultural Sciences (1977) Soil survey of
Humacao area of eastern Puerto Rico. USDA-NRCS,
Humacao Field Office, Humacao
Waldhardt R (2003) Biodiversity and landscape-summary,
conclusions and perspectives. Agric Ecosyst Environ
98:305–309. doi:10.1016/S0167-8809(03)00090-2
Weaver PL (1991) Environmental gradients affect forest
composition in the Luquillo mountains of Puerto Rico.
Interciencia 16:142–151
Whitcomb RF, Robbins CS, Lynch JF, Whitcomb BL,
Klimkiewicz MK, Bystrak D (1981) Effects of forest
fragmentation on avifauna of the eastern deciduous forest.
In: Burgess RL, Sharpe DM (eds) Forest island dynamics in
man-dominated landscapes. Springer-Verlag, New York,
pp 125–205
White HH (1963) Variation of stand structure correlated with
altitude, in the Luquillo mountains. Caribb For 24:46–52
Whittaker RH (1972) Evolution and measurement of species
diversity. Taxon 21:213–251. doi:10.2307/1218190
Wu J, Hobbs R (2002) Key issues and research priorities in
landscape ecology: an idiosyncratic synthesis. Landsc
Ecol 17:355–365. doi:10.1023/A:1020561630963
Wunderle JM Jr (1997) The role of animal seed dispersal in
accelerating native forest regeneration on degraded tropical lands. For Ecol Manag 99:223–235. doi:10.1016/
S0378-1127(97)00208-9
Vascular diversity patterns of forest ecosystem before
and after a 43-year interval under changing climate
conditions in the Changbaishan Nature Reserve,
northeastern China
Weiguo Sang Æ Fan Bai
Originally published in the journal Plant Ecology, Volume 201, No. 1, 115–130.
DOI: 10.1007/s11258-008-9504-0 Springer Science+Business Media B.V. 2008
Abstract The Changbaishan Nature Reserve (CNR)
is the largest protected temperate forest in the world.
It was established in 1960 to protect the virgin
Korean pine (Pinus koraiensis) mixed hardwood
forest, a typical temperate forest of northeast China.
Studies of vascular diversity patterns on the north
slope of the CNR mountainside forest (800–1700 m
a.s.l.) were conducted in 1963 and in 2006. The aim
of this comparison was to assess the long-term effects
of the protected status on plant biodiversity during
the intervening 43 years. The research was carried
out in three forest types: mixed coniferous and broadleaved forest (MCBF), mixed coniferous forest
(MCF), and sub-alpine coniferous forest (SCF),
characterized by different dominant species. The
alpha diversity indicted by species richness and the
Shannon–Wiener index were found to differ for the
same elevations and forest types after the 43-year
interval, while the beta diversity indicated by the
Cody index depicted the altitudinal patterns of plant
W. Sang and F. Bai are contributed equally to this research.
W. Sang (&) F. Bai
Key Laboratory of Vegetation and Environmental
Change, Institute of Botany, The Chinese Academy of
Sciences, Nanxincun 20 , Xiangshan,
Beijing 100093, China
e-mail: swg@ibcas.ac.cn
species gain and loss. The floral compositional
pattern and the diversity of vascular species were
generally similar along altitudinal gradients before
and after the 43-year interval, but some substantial
changes were evident with the altitude gradient. In
the tree layer, the dominant species in 2006 were
similar to those of 1963, though diversity declined
with altitude. The indices in the three forest types did
not differ significantly between 1963 and 2006, and
these values even increased in the MCBF and MCF.
However, originally dominant species, such as Pinus
koraiensis, tended to decline, the proportion of broadleaved trees increased, and the species turnover in the
succession layers showed a trend to shift to higher
altitudes. The diversity pattern of the understory
fluctuated along the altitudinal gradient due to microenvironmental variations. A comparison of the alpha
diversity indices among the three forest types reveals
that the diversity of the shrub and herb layer
decreased, and some rare and medicinal species
disappeared. Meteorological records show that climate has changed significantly in this 43-year
intervening period, and information collected from
another field survey found that the most severe
human disturbances to the CNR forests stemmed
from the exploitation of Ginseng roots and Korean
pine nuts.
Keywords Alpha diversity Beta diversity
Forest ecosystem change Global warming
Human activity Nature preserve
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_10
115
116
Introduction
Biodiversity patterns are the combined result of the
interior processes of the plant community and external
conditions. The hierarchy in forest plant species
diversity is established by the interaction between
the canopy and the understory layers. Gaps in the main
story are initially filled by succession trees growing up
from seedlings, and these gaps may naturally persist
for a long time (Whitmore 1998; Chapin et al. 2002).
Species richness gradients are always determined by
natural environmental gradients, such as altitude,
temperature, seasonal variations, and geological conditions (Rosenzweig 1995; Begon et al. 1996; Körner
2000). Diversity pattern gradients on individual
mountains are simpler to understand and work with
than those in more extensive regional, national,
continental, or global ranges, because of the shorter
geographical distances involved and the similarity of
climatic zones in mountains (Körner 2000; Buker
2003). However, altitudinal patterns of plant species
diversity have long been in contention due to variations in research domains and sampling methods (He
and Chen 1997; Buker 2003; Bhattarai et al. 2004).
Emissions from fossil fuels are increasing, and the
effect of human activity on the environment is
intensifying. Changes in global and local climate,
and in land use, have become the major factors
affecting natural diversity patterns, and such changes
are resulting in noticeable shifts in natural forest
distribution and biodiversity loss (Whitmore 1998;
Dupouey et al. 2002; Carey and Alexander 2003;
Parmesan 2006). Hence, the establishment of nature
reserves with the aim of rescuing critical ecosystems
and endangered species has focused on the relationships between biodiversity, warming climate, and
human impacts (Mouillot et al. 2000; Hooper et al.
2005; Fang et al. 2006). Ecologists continue to study
the long-term maintenance of global biodiversity in
various ecosystems (Halpin 1997; Anderson and
Inouye 2001; Defries et al. 2005; Chapman et al.
2006). Such long-term surveys can assess whether
current ecosystem conservation strategies suitably
address the impact of climate change on biodiversity
and species distribution (Phillips et al. 1998; Scott
et al. 2001; Clark 2002).
The ecological responses of vegetation to recent
climate change have been reviewed by Parmesan
(2006), who pointed out the serious shortage of
A.G. Van der Valk (ed.)
studies on the biological impact of climate change on
mountain plant patterns, especially in Asia. The
summer temperatures in northeastern China have
increased 0.15C per decade during the past 50 years,
which is higher than the global average warming rate
(Li et al. 2005). The existence of so few studies of
floral shift along the elevation gradients of mountainsides has been attributed to the absence of
historical data. Long-term records of field observations are necessary for assessing the trends in
ecosystems affected by environmental change and
human activities, especially within the framework of
surveying permanent plots (Grabherr et al. 1994;
Anderson and Inouye 2001; Clark 2002; Moiseev and
Shiyatov 2003; Pauli et al. 2007).
The Changbaishan Nature Reserve (CNR) was
founded in 1960 to protect the virgin Korean pine
(Pinus koraiensis) mixed hardwood forest, a temperate forest type typical of those found in northeastern
China (Tao 1994; Yang and Xu 2003; Stone 2006).
The CNR is the largest protected temperate forest in
the world and is home to endangered Siberian tigers
(Panthera tigris altaica), valuable species of Ginseng
(Panax ginseng C.A. Mey), and the Korean pine (P.
koraiensis).
Due to its quality, the wood of the Korean pine has
high commercial value, while the Korean pine seeds
are an important food source for the rodent wildlife of
the forest ecosystem. A study conducted 43 years ago
systematically characterized the distribution pattern of
vegetation in the CNR (Chen 1963; Chen et al. 1964;
Zhou and Li 1990) and identified five natural vertical
vegetation zones along the elevation gradient (Chen
et al. 1964; Wang et al. 1980; Zhao et al. 2004): (1) a
mixed coniferous and broad-leaved forest zone
(MCBF) (below 1100 m a.s.l.), dominated by Pinus
koraiensis, Acer mono, Tilia amurensis, Ulmus davidiana var. japonica, Quercus mongolica, etc.; (2) a
mixed coniferous forest zone (MCF) (1100–1500 m
a.s.l.), dominated by P. koraiensis, Picea jezoensis var.
komarovii, Abies nephrolepis, Larix olgensis var.
changpaiensis, etc.; (3) a sub-alpine coniferous forest
zone (SCF) (1500–1800 m a.s.l.), dominated by Picea
jezoensis var. komarovii, Larix olgensis var. changpaiensis, and Abies nephrolepis; (4) a birch forest zone
(BF) (1800–2100 m a.s.l.), dominated by Betula
ermanii; (5) a tundra zone (above 2100 m a.s.l.),
dominated by Rhododendron aureum, R. redowskianum, Vaccinium uliginosum var. alpinum, etc. (Fig. 1).
Forest Ecology
Although tree cutting was prohibited around the time
the natural reserve was founded, the forest is still
threatened by warming temperatures and by human
activities under the canopy, such as tourism and the
collecting of pine nuts (Chen and Wang 1999).
Although the forest has been monitored by field
investigations and satellite imagery during the past
10–28 years (Zhang et al. 1994; Liu 1997; Jin et al.
2005; Liu et al. 2005; Li et al. 2006), no long-term
survey comparing the diversity patterns in the 1960s to
the present status of CNR had been conducted before the
study reported here. In 1963, Chen et al. (1964)
investigated the flora with the aim of characterizing
the major forests in various vertical zones on the north
slope of the CNR. Fortunately, a set of original plot
records from that survey of the CNR has been preserved.
The aim of the this study was to determine whether
significant differences have occurred in terms of the
forest biodiversity patterns of vascular plant species
on the north slope of the CNR by comparing the field
observations made in 1963 with those of 2006.
Analysis and discussion of the possible reasons for
this change, involving global warming and human
activities, is relevant to any assessment of the longterm protective effects of the CNR and to planning
future management strategies.
117
Changbaishan Mountain (800–1700 m a.s.l.). The
map coordinates of the investigation area are
127550 E–128080 E, 42040 N–42230 N (Fig. 1).
The topography is very moderate, ranging from 800
to 1700 m a.s.l., with slopes of \15.
Human activities in these three forest types have
been reduced since the establishment of the CNR,
though some disturbances still occur. The different
elevations with their plant compositions are variously
affected by a range of human activities, such as
gathering pine seeds, collecting herbs, and tourism
(Table 1). Of the three forest types mentioned above,
the MCF has been impacted the most by these
activities, the SCF the least.
According to the climatic records of the Songjang
meteorological station (721.4 m a.s.l.; 42250 N/
128070 E), located on the northern edge of the
CNR (Fig. 2), climate trends have changed during the
45 years from 1960 to 2005. Annual average temperatures have fluctuated and gradually increased
about 0.37C per decade. The temperature values
recorded after 1985 are higher than preceding ones
and than the 45-year average (2.7C). Precipitation
has fluctuated around the 45-year average
(680.5 mm), without any apparent trend.
Field surveying
Methods
Study area
The CNR covers an area of 196,456 ha and is located
in Jilin Province, northeastern China. This region is
situated on the border between China and North
Korea. The major factor affecting the weather is the
monsoon. The area is characterized by a mountain
climate, with dry and windy springs, short and rainy
summers, cool and foggy autumns, and cold and long
winters; there are decreasing temperatures with
increasing elevation, abundant precipitation that
increases along altitudinal gradients, and strong
winds with a prevailing direction to the west–
south–west (WSW) (Chi et al. 1981) (Table 1).
The research was carried out in a mixed coniferous
and broad-leaved forest (MCBF), a mixed coniferous
forest (MCF), and a sub-alpine coniferous forest
(SCF), three of the natural vertical vegetation zones
identified in the earlier study, on the north slope of
Based on the records of Chen et al. (1964), we
repeated the 1963 field survey in 2006, using the
same methods and at the same locations. The plots
were chosen according to the 1963 plot records,
which contained information on elevation, landforms,
slope gradients, slope directions, and dominant
species in each forest type.
The field observations of the 1963 survey were
carried out from June to August and the 2006 survey
from July to August. In order to achieve comparable
results, we utilized the same sampling method as the
1963 inventory. From the 63 plots sampled in 1963,
we chose 52 plots (20 9 20 m) with clear location
records and distinct geographic characteristics. A
minimum of two plots were sampled in each 100-m
altitudinal interval, and ten plots were chosen in each
forest type (Table 1).
Each plot contained four sub-plots (10 9 10 m) for
surveying trees, four shrub sub-plots (10 9 10 m),
and four herb sub-plots (1 9 1 m). The plot positions
were determined by GPS, which also recorded the
118
A.G. Van der Valk (ed.)
Fig. 1 Map of the research area and vegetation zone according to altitude in the Changbaishan Nature Reserve (CNR)
Forest Ecology
119
Table 1 Altitude, temperature, precipitation, and plot characteristics of the survey area along the northern slope of Changbaishan
Mountain
Altitude
(m)
Annual mean
temperaturea
(C)
Annual
precipitationa
(mm)
Forest
typeb
Human
activityc
Number
of plots
MCBF
A, B
13
800
2.32
703.62
900
1.81
728.95
1000
1.29
755.19
1100
0.78
782.37
10
3
MCF
A, B, C
5
1200
0.27
810.53
6
1300
1400
-0.24
-0.75
839.70
869.92
3
2
1500
-1.26
901.23
1600
-1.78
933.67
4
1700
-2.29
967.28
4
a
SCF
B, C
2
Data from Chi et al. 1981
b
MCBF, a mixed coniferous and broad-leaved forest zone; MCF, a mixed coniferous forest zone; SCF, a sub-alpine coniferous
forest zone
c
A, collecting Korean pine nuts; B, collecting herbs and medicinal plant materials; C, tourism
elevation, topography, and other background information. All living trees with height C1.3 m in the tree
sub-plots were tallied by species and DBH (diameter at
breast height). The canopy was divided into a main
story (DBH C 8 cm ) and succession layer (DBH \
8 cm ) (Hao 2000), representing the main composition
and the future trend of the forest ecosystem, respectively. In the shrub and herb sub-plots, records of every
stem were organized by species or species group,
abundance, coverage, and average height.
Data analysis
Alpha diversity index
Species richness (S) and the Shannon–Wiener index
(H), both based on the AIV, were used to measure
plant species diversity and to show the distribution
pattern of plant diversity within plots at the same
altitude. These two indices were calculated for the
canopy (DBH 3 8 cm), succession (DBH \ 8 cm),
shrub, and herb layers. Species richness was determined as the total number of species in each plot. The
values of the Shannon–Wiener index are generally
defined as follows (Ma 1994; Ma and Liu 1994;
Shang 2002):
Altered Important Value
H¼
The Altered Important Value (AVI, Ma et al. 1995)
was used to measure the relative importance of the
plant species in the community. The AIV values were
determined as follows:
s
X
pI ln pI
i¼1
where s is the total number of species in a plot, and pI
is the AIV corresponding to the species.
Relative dominance þ Relative abundance þ Relative height
100%
3
Relative coverage þ Relative abundance þ Relative height
100%
AIVðshrub/herb layerÞ ¼
3
AIVðtree layerÞ ¼
120
A.G. Van der Valk (ed.)
4.5
1000
800
3.5
700
3
600
500
2.5
y = 0.0372x - 71.132
2
2
R = 0.5266
400
300
200
1.5
100
1
Average precipitations(mm)
900
4
Average temperature ( °C)
Fig. 2 Changes in the
annual average temperature
and precipitation at
Songjiang in the CNR from
1960 to 2000
45-year average
temperature
45-year average
precipitation
Annual average
temperature
Annual average
precipitations
Temperature
trend
0
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
The t test was used to assess the differences in the
Shannon–Wiener index (H) mean values for each plot
of the three forest types between the two surveys.
Beta diversity index
Species turnover diversity, indicated by the Cody
index (bc), was calculated as the gain and loss of
species between neighboring altitudes according to the
formula proposed by Wilson and Shmidam (1984):
bc ¼
gðH Þ þ lðH Þ
2
where g(H) and l(H) are the number of species gained
and lost, respectively, between two neighboring
elevation intervals (100 m). In this study, the Cody
index (bc) was used to illustrate the spatial patterns of
species change along the altitudinal gradient.
The researchers who conducted the 1963 survey
did not plan to analyze the beta diversity, so the plot
numbers varied within each altitudinal interval (e.g.
two plots at 1400 m a.s.l., and 13 plots at 800 m a.s.l.).
The situation is different with the general sampling
design using the same sample size at every elevation
(Fosaa 2004; Chao et al. 2005; Harborne et al. 2006).
The minimum total area of each elevation plot
(800 m2) was far larger than the minimum sample
area in the temperate zone. Hao (2000) concluded that
the sampling areas are 64 m2 in the CNR and that the
Cody index (bc) is therefore credible.
To avoid any bias caused by the different sample
areas, we calculated the same year bc to determine
whether there was any significant difference between
the two methods: (1) based on the different areas,
including all plots at each elevation; (2) based on the
same number of plots at each elevation, randomly
selecting two plots at each 100-m interval.
The paired t test results of bc by these two methods
using the 1963 and 2006 data showed that all t test
values were \1 (P [ 0.05); thus, there were nonsignificant differences between varied sample areas.
As in Heegaard’s (2004) research, we demonstrated
that the main factor influencing the beta diversity was
not the sampling area or habitat size, but the
ecological gradients and increased distances. Thus,
we concluded that the method based on the different
sampling areas could be used to measure the beta
diversity in the study.
Results
Altitudinal gradient patterns of plant diversity in
1963 and 2006
Floral composition
The 1963 survey found 202 species, 144 genera, and
61 families of vascular plants, comprising 38 tree
species, 37 shrub species, and 127 herb species,
within the altitude range between 800 and 1700 m
a.s.l. The numbers of species had slightly declined by
2006, when we found 196 species, belonging to141
genera and 58 families, comprising 37 tree species,
37 shrub species, and 122 herb species.
Alpha diversity pattern
The patterns of species richness and the Shannon–
Wiener index along altitudinal gradients in 2006 are
Forest Ecology
similar to those of 1963, except in the succession
layer (Fig. 3).
Figure 3a, c shows that species richness in the tree
layer and the Shannon–Wiener index in the main story
decreased along altitude. The least square linear
regression analysis is given in Table 2. The curve in
2006 has the nearly same shape as that of 1963 but, in
general, there is more richness in 2006 than 1963.
Both results show three peaks in species richness and
Shannon–Wiener index, all of which are located in the
central part of the three forest types. The first is at
900 m a.s.l. for 1963 and at 1000 m a.s.l. for 2006; the
second occurs between 1200 and 1300 m a.s.l. for the
two periods; the third is at 1500 m a.s.l. for 1963 and
1600 m a.s.l. for 2006. Above an altitude of 900 m
a.s.l. for the 1963 survey and 1000 m a.s.l. for the
2006 survey, the Shannon–Wiener index decreases
more or less monotonically up to 1700 m a.s.l.
It can be seen that the diversity pattern of the
understory layer (shrub and herb layers) in 2006 still
has the same shape as in 1963, but is less rich. The
diversity tends to decrease along the altitudinal
gradient (Fig. 3b, e, f). The indices fall from peaks
at 800 m a.s.l. to troughs at 1200 or 1300 m a.s.l.,
Fig. 3 Patterns of change
in forest alpha diversity
along the altitudinal
gradient in 1963 and 2006.
a Species richness in tree
layer, b species richness in
understory, c–f Shannon–
Wiener index in main story
c, success layer d, shrub
layer e, and herb layer f
121
then increase again to peaks at 1400 or 1500 m a.s.l.,
finally decreasing with elevation. A linear regression
between the Shannon–Wiener index values and
altitude was used for the shrub layer (Table 2).
In contrast to the main story, the Shannon–Wiener
index of the succession layer is distinct for the two
surveys (Fig. 3d). The index for 1963 generally
shows a decreasing trend for the main story, with
peaks at 900 and 1500 m a.s.l. In contrast, the
decreasing trend of the under tree layer in 2006 is
similar to that of the succession layer, with a lower
value at 1200 m a.s.l.
Beta diversity pattern
Figure 4 shows the beta diversity patterns (Cody
index) along the altitudinal gradient for 1963 and
2006. All beta diversity declined sharply with
increasing elevation. The beta diversity patterns of
the two surveys differ only for the succession layer,
with similar trends seen for the other canopies and all
of its species.
The beta diversity pattern of the main story shows
two peaks (1000–1100 m a.s.l. and 1400–1500 m
-0.51
-0.80
-0.88
2.480 (1.540, 2.707)
1.875 (0.632, 2.292)
2.512 (2.391, 3.547)
-0.77 -0.0009 (-0.0011, -0.0002)
-0.39 -0.0005 (-0.00098, 0.0003)
-0.90 -0.001 (-0.0016, -0.0007)
2.123 (1.988, 2.971)
1.462 (0.972, 2.779)
2.969 (1.761, 3.263)
-0.0003 (-0.0012, 0.0002)
-0.0011 (-0.0015, -0.0004)
The regression model is y = B 9 x ? A (n = 10, x = elevation m a.s.l., y = alpha biodiversity index). The confidence probability of the regression parameters B and A is
95%. The confidence limits are in parenthesis following the parameters (lower bound, upper bound)
a
Tree layer
Species richness
Succession layer
Shrub layer
A
B
Regression parameters
Shannon-Wiener index Main story (canopy) -0.0007 (-0.0013, -0.0005)
A
B
Regression equationa
r
2006
a
1963
Canopy
Diversity index
Table 2 A comparison of the linear relations between species richness and altitude in 1963 and 2006
-0.0051 (-0.0082, -0.00195 13.164 (9.139, 17.188) -0.80 -0.0062 (-0.0088, -0.0037) 16.503 (13.257, 19.749) -0.89
A.G. Van der Valk (ed.)
r
122
a.s.l.) for 1963 and 2006 (Fig. 4a). In the 800–
1000 m a.s.l. interval, the Cody index increases
gradually to a peak at 1000–1100 m a.s.l., decreases
to its lowest level, and then fluctuates between 1100
and 1400 m a.s.l. until reaching a second peak. The
index value at low elevation is greater than that at
high elevation.
The beta diversity patterns in the succession layer
fluctuate along the altitudinal gradient with three
peaks and troughs, as shown in Fig. 4b. Peaks in the
1963 data occur at 900–1000, 1200–1300, and 1500–
1600 m a.s.l.; troughs occur at 1100–1200, 1300–
1400, and 1600–1700 m a.s.l. Peaks in the 2006 data
occur at 1000–1100, 1400–1500, and 1600–1700 m
a.s.l., while the troughs occur at 900–1100, 1100–
1300, and 1500–1600 m a.s.l.
Figure 4c, d shows a similar trend for the understory layer and all its species. Comparing the trends
of the 1963 and 2006 data, the lower Cody index for
both years are at 1100–1300 m a.s.l., while peaks
appear at 1000–1100 and 1400–1500 m a.s.l. for
1963 and at 900–1000 and 1500–1600 m for 2006.
Changes in the three forest types
Changes in species composition
The dominant tree species in the canopy and the
succession layer are given in Tables 3 and 4, which
show that the dominant species in 2006 were
generally similar to those of 1963. Nevertheless,
some important changes in species status occurred in
the three forest types during the time interval between
studies.
In the MCBF (800–1100 m a.s.l.), Pinus koraiensis was still the predominant tree, but the AIV
declined from 27 to 19%. The proportion of broadleaved tree species increased, so that Quercus mongolica became the newly dominant tree.
In the MCF (1100–1500 m a.s.l.), the AIV of Abies
nephrolepis increased, although those of other needleleaved trees (e.g. Larix olgensis var. changpaiensis
and Pinus koraiensis) lost their once dominant status.
The proportion of broad-leaved tree species increased
(e.g., Acer mono and Betula platyphylla).
In the SCF (1500–1700 m a.s.l.), Picea jezoensis
var. komarovii and A. nephrolepis took each others
place, with the AIV of A. nephrolepis being 44%.
Broad-leaved tree species were also added.
Forest Ecology
123
Fig. 4 Patterns of change
in forest beta diversity
along an altitudinal gradient
in 1963 and 2006. a
Canopy, b succession layer,
c understory, d all plant
species
As shown in Table 3, the AIV of the succession
layer and the status of the broad-leaved tree species in
MCBF and MCF all increased. There were broadleaved trees together with the dominant species in the
succession layer of the MCBF (Table 4).
Alpha diversity change
Table 5 enumerates the species richness and the
Shannon–Wiener index for the different forest types
of the two surveys. The trends of the changes were
nearly the same in the MCBF and MCF, but they
differed in the SCF.
The indices of the tree layer (the canopy and
succession layer) do not differ significantly between
the 1963 and 2006 data sets. The values of the
canopy indices increased 5 and 13% in the MCBF
and MCF, respectively, but are almost equal for the
SCF. The indices of the succession layer increased
more than 20% in the MCBF and MCF and by 5%
in the SCF.
The diversity of the shrub and herb layers
significantly decreased in the three forest types,
except in SCF shrub layer. In the MCBF and MCF,
species richness in the shrub layer decreased
slightly, and the Shannon-Wiener index of both fell
Table 3 Comparisons of the composition of dominant speciesa in the canopy of the three forest types between the 1963 and 2006
surveys
Forest type
MCBF
MCF
SCF
Year
Components of tree species (DBH C 8 cm) and AIV (9100)b
1963
PK 27 ? AM 15 ? AN 13 ?TA 10 ? AP 8 ? Others 27
2006
PK 19 ? AM 13 ? AN 12 ? QM 10 ? TA 8 ? Others 38
1963
LO 29 ? PK 26 ? AN 25 ? PJ 11 ? Others 9
2006
AN 27 ? LO 22 ? PK 22 ? PJ 6 ? Others 23
1963
LO 39 ? PJ 34 ? AN 19 ? Others 8
2006
AN 44 ? PJ 26 ? LO 13 ? Others 17
Only trees with a DBH (diameter at breast height) C 8 cm were counted. AIV, Altered Important Value
b
AN,Abies nephrolepis; LO, Larix olgensis var. changpaiensis; PJ, Picea jezoensis var. komarovii; PK, Pinus koraiensis; QM,
Quercus mongolica; AM, Acer mono; AP, A. pseudo-sieboldianum; TA, Tilia amurensis
124
about 15%, while 1963 species richness had
decreased by over 40% and the Shannon-Wiener
index by over 20% by 2006. There is no statistically
significant decrease in the shrub layer of the SCF,
and the diversity of its herb layer decreased less
than 12%.
Discussion
Effective conservation in the CNR
Floral composition
The establishment of the CNR has prevented the
forest communities from experiencing large disturbances, such as clear cutting and land-use change
(Liu 1997; Yang and Xu 2003; Liu et al. 2005), as
clearly shown by our comparisons of field data
from 1963 and 2006. The total number of species
recorded in 2006 were only slightly less than those
counted in 1963 and, in general, similar patterns
were found in the floral composition of the tree
layer. As seen in this analysis, the species compositions of the dominant trees in the tree layer were
generally the same in both surveys. The succession
layer represents the future canopy of the forest
following long-term growth (Lertzman 1992; Whitmore 1998; Chapman et al. 2006). Consequently,
the results of this investigation indicate that
biodiversity conservation has been effective in the
CNR.
Other investigators of the CNR have reported
similar results in terms of these three forest types.
Zhao et al. (2004) described the composition and
structure of communities on the northern slope of
Changbai Mountain based on TWINSPAN classification. They found that three community groups
existed within the altitude range from 700 to 1780 m
a.s.l. The MCBF, containing the dominant tree
species, such as Pinus koaiensis, is the most important forest type in China (Wu 1980; Zhou and Li
1990); the MCF, located between two other forest
types, represents a transitional zone (Zhao et al.
2004); the SCF is typical of alpine forests. Thus, the
vegetation zones in this area are suitable for carrying
out the kind of long-term monitoring and research
reported in our study.
A.G. Van der Valk (ed.)
General diversity pattern between 1963 and 2006
The general comparability of altitudinal diversity
patterns between the CNR surveys of 1963 and 2006
is evident, as are the analogous patterns within the
specific elevation intervals (Shao 1999; Hao 2000;
Zhao et al. 2004). The species richness and Shannon–
Wiener indices decreased in the tree layer and
the understory with increasing altitude. A higher
species turnover was apparent at lower elevations.
Hence, the conservative effects of the CNR cannot be
doubted.
The causes of the changes in plant diversity are
complex, involving various ecological gradients and
floral interactions. In lower mountains, the presence
of more broad-leaved tree species is due to the
abundance of growth resources and warmer temperatures (Lopes Valle de Britto Rangel and Felizola
Diniz-Filho 2003) (Bhattarai et al. 2004; Stromberg
2007). Competition resulting from the complex
interactions between plant species and disturbances
caused by animals are also seen to be further reasons
for higher diversity (Vázquez and Givnish 1998;
Pauli et al. 2007). The decline in forest tree diversity
with elevation may be related to decreased temperature gradient, seasonal length, and/or nutritive
material (Körner 2000; Fosaa 2004).
In terms of species–area relationships (Rosenzweig 1995; Hill and Curran 2001), fewer species
were expected in the higher part of Changbai
Mountain because the amount of available habitat
area decreases. Nevertheless, the three peaks of the
diversity curves did occur in the tree layer at the midelevation of each forest type and are consistent with
the hypothesis of a mid-domain effect (MDE)
(Colwell and Lees 2000). There are boundary constraints of species dispersal at the ecotone, but more
overlap of species ranges and species inter-connection occurs in the middle of habitats because they are
the centers of shared geographic domains (Bhattarai
et al. 2004; Pauli et al. 2007).
In contrast to the tree layer, the lower values of the
diversity patterns for the understory appeared at midelevation (1200 or 1300 m a.s.l.) and the peaks at the
boundary between different forest types. These
observations seem to contradict the MDE. Such a
situation has been noted by other researchers and is
interpreted as resulting from the different processes
that influence species richness across an altitudinal
Forest Ecology
125
Table 4 Comparisons of composition of dominant speciesa in the succession layers of the different forest types between 1963 and
2006 surveys
Forest type
Year
Component of tree species (DBH \ 8 cm) and AIV (9 100)b
MCBF
1963
AU 23 ? AP 17 ? AM 16 ? SR 12 ? ATF 6 ? Others 26
2006
AU 23 ? AM 18 ? SR 13 ? AP 11 ? AK 6 ? Others 29
MCF
1963
AN 52 ? PJ 16 ? ATM 11 ? AU 7 ? AB 6 ? Others 8
2006
AN 48 ? ATM 11 ? PJ 10 ? LO 6 ? SP 6 ? Others 19
SCF
1963
AN 41 ? PJ 38 ? AU 7 ? AK 6 ? Others 8
2006
AN 45 ? AU 13 ? BE 8 ? PJ 6 ? Others 14
a
Only trees with a DBH \ 8 cm were counted
b
AN, Abies nephrolepis; LO, Larix olgensis var. changpaiensis; PJ, Picea jezoensis var. komarovii; BE, Betula ermanii; SP, Sorbus
pohuashanensis; AB, Acer buergerianum; AK, A. komarovii; AM, A. mono; AP, A. pseudo-sieboldianum; ATF, A. triflorum; ATM,
A. tegmentosum; AU. A. ukurunduense; SR, Syringa reticulata var. mandshurica
gradient (Ricklefs 1987; Hao 2000; Pearman and
Weber 2007). The environmental patterns of the
understory plants are more sensitive to microenvironmental variations with elevation, which do
not always concur with those of woody plants (Bhattarai
et al. 2004; Bergman et al. 2006; Freestone and Inouye
2006). In the mid-domain, the continuous tree canopies
provided homogeneous habitats for shrubs and herbs,
while at the ecotone between one vegetation zone and
another, greater species diversity is explained as
resulting from more gaps and greater heterogeneity
(Shao 1999; Fosaa 2004; Zhao et al. 2004).
The higher beta diversity at lower elevations,
indicating a higher degree of species turnover, is
generally interpreted to be the result of warmer, more
nutrient-rich conditions, and also supports the conjecture that lowland forests may be more diverse
because they contain more space for species–area
relationships (MacArthur 1972; Rosenzweig 1995;
Vázquez and Givnish 1998). Species turnover peaks
cluster around 1000–1100 and 1400–1500 m a.s.l.
These patterns overlap the boundaries between two
forest types. This could be explained simply as a loss
of species in lower altitudes and a gain in the higher
ones. Beta diversity thus illustrates the degree of
habitat compensation (Ma et al. 1995; Fosaa 2004;
Zhao et al. 2004).
Long-term changes in the CNR
Changes in species composition
Despite effective conservation in the CNR, this forest
ecosystem is under pressure from the impact of
climate change and social economic development,
and a degree of instability is evident in the composition and diversity patterns of these forests. Some
rare herb species have disappeared, such as Adonis
amurensis, Jeffersonia dubia, and Goodyera repens.
The predominance of needle-leaved tree species, such
as Pinus koraiensis, has declined, while the broadleaved tree species have expanded into higher altitude
communities in both the main story and the succession layer. This trend was predicted using
observational data and modeling by Yang and Xu
(2003). The reasons for this change were ascribed to
the combined effect of climate change and local
human activities, an assumption that is supported by
observed changes in the diversity pattern, as discussed below.
Changes in the succession layer
Because it is the main force driving the changes in the
forest ecosystem, the succession layer is arguably the
most important part of a forest (Whitmore 1998).
Unlike the Shannon–Wiener and Cody indices for the
canopy, those for the succession layer were distinct in
the two surveys. The alpha diversity pattern changed
from a decreasing trend in 1963 to a trend that
deceases with elevation in 2006. As the increasing
proportion of broad-leaved trees seems to be a signal
of global warming (Yang and Xu 2003; Thuiller et al.
2006), it may be presumed that the species in the
succession layer became more environmentally sensitive, as did the understory layers. Thus, the spatial
pattern altered with the gradual change of the habitat
gradient and the available resources resulting from
0.89 (0.14) 2.40* (0.08)
Canopy layer: Main story and success, shrub, and herb layers
S, Species richness
H, Shannon-Wiener index, including mean (standard error of mean given in parenthesis) of same forest types
b
c
* Significantly different at P \ 0.05
a
2.46 (0.06)
1.48 (0.08)
1.26* (0.10) 1.77* (0.06) 1.11 (0.09) 1.04 (0.10)
1.68* (0.06) 2.07* (0.05) 1.37 (0.06) 1.17 (0.14)
2006 1.73 (0.05) 1.36 (0.07)
1.21 (0.07) 0.94 (0.11)
2.61 (0.07)
2.01 (0.03)
Hc 1963 1.64 (0.05) 1.12 (0.09)
5
5
5
4
19
11
6
6
5
6
10
8
5
6
Sb 1963 6
2006 8
1.11 (0.06) 0.99 (0.14)
4
5
5
7
26
13
26
20
Main story Success layer Shrub layer Herb layer
Main story Success layer Shrub layer Herb layer
Main story Success layer Shrub layer Herb layer
Canopy
layers:
SCF
MCBF
Forest
type:
MCF
1.14 (0.17) 2.73 (0.10)
A.G. Van der Valk (ed.)
Table 5 Plant diversity dynamics of the three forest types on Changbai Mountain in terms of changes in the canopy layer between the 1963 and 2006 surveys
126
climate change (Nogués-Bravo et al. 2007; Pauli
et al. 2007). The underlying factor for the decreased
resilience in the succession layer could be human
impact from continuous harvesting and tourist activities under the canopy (Whitmore 1998; Buker 2003).
The shift in vegetation belts to higher elevations
due to global warming, as demonstrated in this study,
has also been reported from experimental research
and modeling (Walther et al. 2002; Parmesan and
Yohe 2003; Root et al. 2003; Parmesan 2006), but it
has not been found in field investigations. With the
expansion of broad-leaved tree species into higher
altitude communities, the beta diversity in the
succession layer also showed a trend to move to
higher altitudes. The peaks in beta diversity in 1963
were at 900–1000, 1200–1300, and 1500–1600 m
a.s.l; in 2006, the peaks had shifted upwards 100–
200 m, with the peaks occurring at 1000–1100,
1400–1500, and 1600–1700 m a.s.l. Figure 5 shows
that after moving 100 m a.s.l. upwards from the 1963
original beta diversity pattern (Fig. 5a), the first peak
joints (Fig. 5b); moving up 200 m a.s.l., the second
peak joints. The higher the jointed point, the longer
the time needed. This results illustrates that global
warming is likely to have a more severe impact on
higher altitudes, where the adaptive capacities of
species are lower (Nogués-Bravo et al. 2007; Pauli
et al. 2007). Based on our results, climate change has
already had a profound effect on forest patterns in the
CNR.
Changes in the three forest types
The keystone species, such as dominant species in the
tree layer and highly sensitive herb species, are those
that respond most rapidly to external factors. Such
effects extend to the entire community through the
interactions between plants (Mitchell et al. 1999;
Klanderud and Totland 2005). Hence, the changes in
the three forest types can be studied by analyzing the
unique features of the species, the composition of the
community, the degree of diversity, climate change,
and human impact factors.
The changes that occurred in the MCBF and MCF
are nearly the same. The species richness, Shannon–
Wiener index, and proportion of broad-leaved species
in the tree layer increased from 1963 to the present.
This trend was consistent with the RPMKPF modeling result of a primarily mixed broad-leaved Korean
Forest Ecology
127
status and alpha diversity in the understory would also
support this result. The collecting of pine nuts has
dramatically affected the regeneration of the Korean
pine by depleting the seed bank (Wu and Han 1992),
while collecting herbs for food and medicinal purposes has heavily disturbed the understory layer
(Chen and Wang 1999). Global warming could
exacerbate this situation. Miles et al. (1983) used a
Markov chain to simulate the future status of Korean
pine. The results showed that the broad-leaved trees,
especially Quercus mongolica, which is more adapted
to drought conditions, would replace the Korean pine
with increasing climate change (Zhang 1983; Tao
1994). In our study, we found that Quercus mongolica
had become the newly dominant species by 2006.
Human impact in the SCF is relatively light and
concentrated. There are no disturbances other than
tourism, the effects of which are focused on certain
areas and along the tourist path. Thus, the change in
alpha diversity in the SCF is the lowest of the three
forest types (Huston 2005). In the future, however,
the impact on the SCF may intensify due to economic
development (Ohl et al. 2007). After all, the broadleaved trees have invaded the succession layer, and
the diversity of the herb layer has decreased significantly. Human impact on the shrub and herb layers
of forests often escape the notice of forest managers,
because of the delayed or long-term effects (Hooper
et al. 2005; Bergman et al. 2006).
Conclusions
Fig. 5 Upward-shifting trends in forest beta diversity along an
altitudinal gradient in 1963 and 2006. a Original status, b the
100-m upward shift from the 1963 beta diversity pattern, c the
200-m upward shift from the 1963 beta diversity pattern
pine forest under pressure from a combination of
climate change and human activities (Chen and Li
2004). The reduction of Pinus koraiensis’ dominant
The absence of significant changes in the diversity
patterns of vascular plant species with altitude
indicates the effectiveness of the conservation measures that have been in force in the CNR during the
43 years since the 1963 survey. The prohibition of
tree cutting helped to preserve the tree layer.
However, global warming has caused an increase in
the proportion of broad-leaved trees, while human
activity in the forest, such as tourism and the
collecting of pine nuts, has disturbed the understory
plants and the succession layer of trees. The goal of
our study was to provide guiding principles for the
preservation of biodiversity (Hill and Curran 2001).
Regulations and policies for conservation should be
strengthened in the interest of reversing the deteriorating trends in the forest communities of the CNR.
128
Due to the limits of the 1963 survey, we were not
able to cover the higher vegetation zones of the birch
forest and tundra zone. These two tree line ecotones
should show close relationships with environmental
variation (Moiseev and Shiyatov 2003). Thus, incomplete sampling may result in some spurious diversity
patterns (Lomolino 2001; Bhattarai et al. 2004) and
make it difficult to confirm the shift to higher
altitudes of the plant life in those zones. However,
we have made some inferences as to what may
happen and have proposed a number of issues that
should be studied in future investigations.
Acknowledgements This research project was supported by
the National Natural Science Foundation of China (NSFC)
(30590382/C011108) and ‘‘111 Program’’ from Bureau of
China Foreign Expert and Ministry of Education (contract no.
2008-B08044). We are grateful to Prof. Lingzhi Chen for
providing the field observation data of 1963, and to Mr. Kun
Wang and Haicheng Zhou for information on collecting in the
CNR. The Administrative Committee of the Changbai
Mountain Reserve Development Zone, Beijing Forest
University, Mr. Jie Wang, Miao Sun, Liwei Wei, and
Minggang Yin graciously helped with the field investigation
in 2006. We are grateful to two anonymous reviewers for their
constructive criticism, suggestions, and comments, which
resulted in a significantly improved manuscript.
References
Anderson JE, Inouye RS (2001) Landscape-scale changes in
plant species abundance and biodiversity of a sagebrush
steppe over 45 years. Ecol Monogr 71(4):531–556
Begon M, Harper JL, Townsend CR (1996) Ecology: individuals populations and communities. Blackwell Scientific,
Oxford
Bergman E, Ackerman JD, Thompson J, Zimmerman JK
(2006) Land-use history affects the distribution of the
saprophytic Orchid Wullschlaegelia calcarata in Puerto
Rico’s Tabonuco Forest. Biotropica 38(4):492–499. doi:
10.1111/j.1744-7429.2006.00167.x
Bhattarai KR, Vetaas OR, Grytnes JA (2004) Fern species
richness along a central Himalayan elevational gradient,
Nepal. J Biogeogr 31:389–400
Buker A (2003) Inselbergs in a changing world-global trend.
Divers Distrib 9:375–385. doi:10.1046/j.1472-4642.2003.
00035.x
Carey C, Alexander MA (2003) Climate change and amphibian
decline: is there a link? Divers Distrib 9:111–121. doi:
10.1046/j.1472-4642.2003.00011.x
Chao A, Chazdon RL, Colwell RK, Shen T-J (2005) A new
statistical approach for assessing similarity of species
composition with incidence and abundance data. Ecol Lett
8:148–159. doi:10.1111/j.1461-0248.2004.00707.x
Chapin S III, Matson PA et al (2002) Principles of terrestrial
ecosystem ecology. Springer, New York
A.G. Van der Valk (ed.)
Chapman RA, Heitzman E, Shelton MGD (2006) Long-term
change in forest structure and species composition of an
upland oak forest in Arkansas. For Ecol Manage 236:85–
92. doi:10.1016/j.foreco.2006.08.341
Chen LZ (1963) Study on the structure of Picea jezoensis
forest on the western slope of Changbai Mountains. Acta
Phytoecol Geobot Sin 1(1–2):69–80
Chen LZ, Wang ZW (1999) The impact of human alteration on
ecosystem diversity principles. Zhejiang Science and
Technology Press, Zhejiang
Chen LZ, Bao XC et al (1964) Major forests in various vertical
zones on northern slope of Changbai Mountains of Jilin
Province. Acta Phytoecol Geobot Sin 2(2):207–225
Chen XW, Li BL (2004) Tree diversity change in
remaining primary mixed-broadleaved Korean pine
forest under climate change and human activities.
Biodivers Conserv 13:563–577. doi:10.1023/B:BIOC.
0000009490.57334.0a
Chi ZW, Zhang FS, Li XY (1981) Preliminary analysis on the
climate of the Changbai Mountain area. Res For Ecosyst
2:179–186
Clark DA (2002) Are tropical forests an important carbon sink?
Reanalysis of the long-term plot data. Ecol Appl 12(1):3–7.
doi:10.1890/1051-0761(2002)012[0003:ATFAIC]2.0.CO;2
Colwell RK, Lees DC (2000) The mid-domain effect: geometric constrains on the geography of species richness.
Trends Ecol Evol 15:70–76. doi:10.1016/S0169-5347(99)
01767-X
Defries R, Hansen A, Newton A, Hansen M (2005) Increasing
isolation of protected areas in tropical forests over the past
twenty years. Ecol Appl 15(1):19–26. doi:10.1890/03-5258
Dupouey JL, Dambrine E, Laffite JD, Moares C (2002) Irreversible impact of past land on forest land use on forest
soils and biodiversity. Ecology 83(11):2978–2984
Fang JY, Wang ZH, Zhao SQ, Li YK, Yu D, Ni LY, Liu HZ,
Da LJ, Tang ZY, Zheng CY (2006) Biodiversity changes
in the lakes of the Central Yangtze. Front Ecol Environ
4(7):369–377.
doi:10.1890/1540-9295(2006)004[0369:
BCITLO]2.0.CO;2
Fosaa AM (2004) Biodiversity pattern of vascular plant species
in mountain vegetation in the Faroe Islands. Divers Distrib
10(3):217–223. doi:10.1111/j.1366-9516.2004.00080.x
Freestone AL, Inouye BD (2006) Dispersal limitation and
environmental heterogeneity shape scale-dependent
diversity patterns in plant communities. Ecology
87(10):2425–2432. doi:10.1890/0012-9658(2006)87[2425:
DLAEHS]2.0.CO;2
Grabherr G, Gottfried M, Pauli H (1994) Climate effects on
mountain plants. Nature 369:448. doi:10.1038/369448a0
Halpin PN (1997) Global climate change and natural-area
protection: management responses and research directions.
Ecol Appl 7(3):828–843. doi:10.1890/1051-0761(1997)
007[0828:GCCANA]2.0.CO;2
Hao ZQ (2000) Analysis of plant community diversity and
their gradient patterns on the northern slope of Changbai
Mountain, Northeast China. PhD thesis. Chinese Academy of Sciences, P.R. China
Harborne AR, Mumby PJ, Hedley JD, Zychaluk K, Blackwell
PG (2006) Modeling the beta diversity of coral reefs.
Ecology 87(11):2871–2881. doi:10.1890/0012-9658
(2006)87[2871:MTBDOC]2.0.CO;2
Forest Ecology
He JS, Chen WL (1997) A review of gradient changes in
species diversity of land plant communities. Acta Ecol Sin
17(1):91–99
Heegaard E (2004) Trends in aquatic macrophyte species
turnover in Northern Ireland—which factors determine
the spatial distribution of local species turnover? Glob
Ecol Biogeogr 13:97–408. doi:10.1111/j.1466-822X.
2004.00119.x
Hill JL, Curran PJ (2001) Species composition in fragmented
forests: conservation implications of changing forest area.
Appl Geogr 21:157–174. doi:10.1016/S0143-6228(01)
00002-9
Hooper DU, Chapin FSIII, Ewel JJ, Hector A, Inchausti P,
Lavorel S, Lawton JH, Lodge DM, Loreau M, Naeem S,
Schmid B, SetÄlÄ H, Symstad AJ, Vandermeer J, Wardle
DA (2005) Effects of biodiversity on ecosystem functioning: a consensus of current knowledge. Ecol Monogr
75(1):3–35. doi:10.1890/04-0922
Huston MA (2005) The three phases of land-use change:
implications for biodiversity. Ecol Appl 15(6):1864–1878.
doi:10.1890/03-5281
Jin YH, Li DQ et al (2005) Quantitative dynamics on natural
regeneration of secondary forest during the restoration
period in Changbai Mountain area. Natural Sciences
Edition. J Nanjing For Univ 29(5):65–68
Klanderud K, Totland Ø (2005) Simulated climate change
altered dominance hierarchies and diversity of an alpine
biodiversity hotspot. Ecology 86(8):2047–2054. doi:
10.1890/04-1563
Körner C (2000) Why are there global gradients in species
richness? Mountains might hold the answer. Trends Ecol
Evol 15:513–514. doi:10.1016/S0169-5347(00)02004-8
Lertzman KP (1992) Patterns of gap-phase replacement in a
subalpine, old-growth forest. Ecology 73:657–669. doi:
10.2307/1940772
Li J, Gong Q et al (2005) Climatic features of summer temperature in northeast China under warming background.
Chin Geogr Sci 15(4):337–342. doi:10.1007/s11769-0050022-x
Li XB, Wang XP et al (2006) Influence of human disturbance
on community structure and tree species diversity of
secondary forest on northern slope of Mt. Changbai. Sci
Silae Sin 42(2):105–110
Liu QJ (1997) Structure and dynamics of the subalpine
coniferous forest on Changbai Mountain, China. Plant
Ecol 132:97–105. doi:10.1023/A:1009705319128
Liu QJ, Li XR et al (2005) Monitoring forest dynamics using
satellite imagery-a case study in the natural reserve of
Changbai Mountain in China. For Ecol Manage 210:25–
37. doi:10.1016/j.foreco.2005.02.025
Lomolino MV (2001) Elevation gradients of species-density:
historical and prospective views. Glob Ecol Biogeogr
10:3–13. doi:10.1046/j.1466-822x.2001.00229.x
Lopes Valle de Britto Rangel TF, Felizola Diniz-Filho JA
(2003) Spatial patterns in species richness and the geometric constraint simulation model: a global analysis of
mid-domain effect in Falconiformes. Ecologica 24(4);
203–207
Ma KP (1994) The measurement of community diversity. I:
The measurement of a diversity. Chin Biodivers 2(3):
162–168
129
Ma KP, Liu YM (1994) The measurement of community
diversity. II: The measurement of a diversity. Chin Biodivers 2(4):231–239
Ma KP, Huang JH et al (1995) Plant community diversity in
Dongling Mountain, Beijing, China. II. Species richness,
evenness, and species diversities. Acta Ecol Sin 15:268–277
MacArthur RH (1972) Geographical ecology. Harper and Row,
New York
Miles J, French DD, Xu Z (1983) The primary researcher on
the forest succession status of mixed conifer and broadleaved forest in north slop of Changbai Mountain. Res
Forest Ecosyst 3:54–72.
Mitchell RJ, Zutter BR, Gjerstad DH, Glover GR, Wood CW
(1999) Competition among secondary–successional pine
communities: A field study of effects and responses.
Ecology 80(3):857–872
Moiseev PA, Shiyatov SG (2003) The use of old landscape
photographs for studying vegetation dynamics at the treeline ecotone in the Ural Highlands, Russia. In: Nagy L,
Grabherr G, KÖrner C, Thompson DBA (eds) Alpine
biodiversity in Europe. Springer, Berlin, pp 423–436
Mouillot D, Titeux A, Migon C, Sandroni V, Frodello, JP,
Viale D (2000) Anthropogenic influences on a Mediterranean Nature Reserve: modelling and forecasting.
Environ Model Assess 5:185–192. doi:10.1023/A:101153
3811237
Nogués-Bravo D, Araújo MB, Martinez-Rica JP, Errea MP
(2007) Exposure of global mountain systems to climate
warming during the 21st Century. Glob Environ Change
11(7):420–428
Ohl C, Krauze K, Grünbühel C (2007) Towards an understanding of long-term ecosystem dynamics by merging
socio-economic and environmental research Criteria for
long-term socio-ecological research sites selection. Ecol
Econ 63:383–391. doi:10.1016/j.ecolecon.2007.03.014
Parmesan C (2006) Ecological and evolutionary responses to
recent climate change. Annu Rev Ecol Evol Syst 37:637–
669. doi:10.1146/annurev.ecolsys.37.091305.110100
Parmesan C, Yohe G (2003) A globally coherent fingerprint of
climate change impacts across natural systems. Nature
421:37–42. doi:10.1038/nature01286
Pauli H, Gottfried M, Reiter K, Klettner C, Grabherr G (2007)
Signals of range expansions and contractions of vascular
plants in the high Alps: observations (1994–2004) at the
GLORIA master site Schrankogel, Tyrol, Austria. Glob
Change Biol 13:147–156. doi:10.1111/j.1365-2486.2006.
01282.x
Pearman PB, Weber D (2007) Common species determine
richness patterns in biodiversity indicator taxa. Biol
Conserv 138:109–119. doi:10.1016/j.biocon.2007.04.005
Phillips OL, Malhi Y, Higuchi N, Laurance WF, Nunez PV
et al (1998) Changes in the carbon balance of tropical
forests: evidence from long-term plots. Science 282:439–
442. doi:10.1126/science.282.5388.439
Ricklefs RE (1987) Community diversity: relative roles of
local and regional processes. Science 235:167–171. doi:
10.1126/science.235.4785.167
Root TL, Price JT, Hall KR, Schneider SH, Rosenzweig C,
Pounds JA (2003) Fingerprints of global warming on wild
animals and plants. Nature 421:57–60. doi:10.1038/nature
01333
130
Rosenzweig ML (1995) Species diversity in space and time.
Cambridge University Press, Cambridge
Scott JM, Davis FW, McGhie G, Wright RG (2001) Nature
reserves: Do they capture the full rang of American’s
biological diversity? Ecol Appl 11(4):999–1007. doi:
10.1890/1051-0761(2001)011[0999:NRDTCT]2.0.CO;2
Shang YC (2002) Ordinary ecology. Beijing University Press,
Beijing
Shao B (1999) Specie diversity in the sub alpine spruce-fir
forest community on the north slope of MT. Changbai.
Acta Phytoecol Sin 23[Suppl]:47–57
Stone R (2006) A threatened nature reserve breaks down Asian
borders. Science 313:1379–1380. doi:10.1126/science.
313.5792.1379
Stromberg JC (2007) Seasonal reversals of upland-riparian
diversity gradients in the Sonoran Desert. Divers Distrib
13:70–83
Tao Y (1994) Revolution and succession trends of forest vegetation in Changbai Mountain. Res For Ecosyst 7:173–184
Thuiller W, Lavorel S, Sykes MT, Araújo MB (2006) Using
niche-based modelling to assess the impact of climate
change on tree functional diversity in Europe. Divers
Distrib 12:49–60. doi:10.1111/j.1366-9516.2006.00216.x
Vázquez GJA, Givnish TJ (1998) Altitudinal gradients in
tropical forest composition, structure, and diversity in the
Sierra de Manantlán. J Ecol 86:999–1020. doi:10.1046/
j.1365-2745.1998.00325.x
Walther G, Post E, Convey P, Menzel A, Parmesan C, Beebee
TJC et al (2002) Ecological responses to recent climate
change. Nature 416:389–395. doi:10.1038/416389a
A.G. Van der Valk (ed.)
Wang Z, Xu ZB, Li X, Peng DS, Tan ZX (1980) The main
forest type and their structure characteristics (I). Res For
Ecosyst 1:25–42
Whitmore TC (1998) Potential impact of climatic change on
tropical rain forest seedlings and forest regeneration. Clim
Change 39:429–438. doi:10.1023/A:1005356906898
Wilson MV, Shmidam A (1984) Measuring beta diversity with
presence–absence data. J Ecol 72:1055–1064. doi:10.2307/
2259551
Wu YG, Han JX (1992) The analysis of community structure
and nature succession pattern of Korea pine. Res For
Ecosyst 6:14–23
Wu ZY (1980) The vegetation of China. Science and Technology Press, Beijing
Yang X, Xu M (2003) Biodiversity conservation in Changbai
Mountain biosphere reserve, northeastern China: status,
problem, and strategy. Biodivers Conserv 12:883–903.
doi:10.1023/A:1022841107685
Zhang SY (1983) Study on succession pattern of secondary
forest in Changbai Mountain. Res For Ecosyst 3:44–53
Zhang YP, Guo XF et al (1994) Study on tree dynamics of
shrub-broadleaved Korean Pine Forest in Changbai
Mountain. Res For Ecosyst 7:48–53
Zhao SQ, Fang JY et al (2004) Composition, structure and
species diversity of plant communities along an altitudinal
gradient on the northern slope of Mt. Changbai, Northeast
China. Chin Biodivers 12(1):164–173
Zhou YL, Li SY (1990) Forests of China. Science Press,
Beijing
Gap-scale disturbance processes in secondary hardwood
stands on the Cumberland Plateau, Tennessee, USA
Justin L. Hart Æ Henri D. Grissino-Mayer
Originally published in the journal Plant Ecology, Volume 201, No. 1, 131–146.
DOI: 10.1007/s11258-008-9488-9 Springer Science+Business Media B.V. 2008
Abstract Disturbance regimes in many temperate,
old growth forests are characterized by gap-scale
events. However, prior to a complex stage of
development, canopy gaps may still serve as mechanisms for canopy tree replacement and stand
structural changes associated with older forests. We
investigated 40 canopy gaps in secondary hardwood
stands on the Cumberland Plateau in Tennessee to
analyze gap-scale disturbance processes in developing forests. Gap origin, age, land fraction, size, shape,
orientation, and gap maker characteristics were
documented to investigate gap formation mechanisms
and physical gap attributes. We also quantified
density and diversity within gaps, gap closure, and
gap-phase replacement to examine the influence of
localized disturbances on forest development. The
majority of canopy gaps were single-treefall events
caused by uprooted or snapped stems. The fraction of
the forest in canopy gaps was within the range
reported from old growth remnants throughout the
J. L. Hart (&)
Department of Geography, University of North Alabama,
Florence, AL 35632, USA
e-mail: jhart13@gmail.com
H. D. Grissino-Mayer
Department of Geography, The University of Tennessee,
Knoxville, TN 37996, USA
region. However, gap size was smaller in the
developing stands, indicating that secondary forests
contain a higher density of smaller gaps. The majority
of canopy gaps were projected to close by lateral
crown expansion rather than height growth of
subcanopy individuals. However, canopy gaps still
provided a means for understory trees to recruit to
larger size classes. This process may allow overtopped trees to reach intermediate positions, and
eventually the canopy, after future disturbance
events. Over half of the trees located in true gaps
with intermediate crown classifications were Acer
saccharum, A. rubrum, or Liriodendron tulipifera.
Because the gaps were relatively small and close by
lateral branch growth of perimeter trees, the most
shade-tolerant A. saccharum has the greatest probability of becoming dominant in the canopy under the
current disturbance regime. Half of the gap maker
trees removed from the canopy were Quercus;
however, Acer species are the most probable replacement trees. These data indicate that canopy gaps are
important drivers of forest change prior to a complex
stage of development. Even in relatively young
forests, gaps provide the mechanisms for stands to
develop a complex structure, and may be used to
explain patterns of shifting species composition in
secondary forests of eastern North America.
Keywords Canopy gaps Disturbance
Forest development Mixed hardwoods
Succession Tennessee
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_11
131
132
Introduction
All forest ecosystems are subject to natural disturbance events that shape species composition and
stand structure. In many forest types, gap-scale
disturbance processes are the dominant disturbance
mechanisms. Thus, canopy gap characteristics and
forest response have been studied in forests throughout eastern North America to elucidate patterns, and
processes of gap-scale disturbances and forest vegetation dynamics. The overwhelming majority of
canopy gap studies, however, have been conducted
in old growth remnants (e.g., Lorimer 1980; Barden
1981; Runkle 1982; Cho and Boerner 1991; Runkle
2000). Throughout the Eastern Deciduous Forest
Region, most forested land supports secondary stands
(secondary referring to all non-primeval forests prior
to a complex stage of development) composed of
mixed hardwood species (Cowell 1998; Rebertus and
Meier 2001). Few studies have analyzed gap-scale
disturbances and forest response in secondary forests
(but see Clebsch and Busing 1989; Dahir and Lorimer
1996; Wilder et al. 1999; Yamamoto and Nishimura
1999), and no such research has been conducted in
mixed hardwood stands on the Cumberland Plateau.
Undoubtedly, forest disturbance dynamics differ
between old growth remnants and mature secondary
stands. Differences in disturbance characteristics are
attributed to variations in species composition, biomass arrangement, and tree-age distribution. As
forests mature, the distance between large individuals
increases. Tree crowns separate into distinct categories, creating a more complex vertical structure, and
species composition shifts to favor later-successional
species (Goebel and Hix 1996; Oliver and Larson
1996; Goebel and Hix 1997). Forest response to
disturbance events likely differs between old growth
and secondary stands, because of differences in stand
structure and species composition, and also because
of the ages of the oldest trees, as older trees are less
able to respond to increase in available resources
resulting from disturbance events (Fritts 2001).
In old growth forests, the spacing between large
individuals is greater than in secondary forests. Thus,
when a canopy tree is removed from an old growth
stand, the size of the canopy gap created should be
larger than a comparable disturbance during earlier
stages of forest development (Clebsch and Busing
1989; Spies et al. 1990; Tyrell and Crow 1994; Runkle
A.G. Van der Valk (ed.)
1998; Yamamoto and Nishimura 1999). Because
canopy gaps are generally larger in old growth
remnants, many of the gaps in these forests close by
the height growth of subcanopy individuals rather than
lateral crown expansion of perimeter trees (Runkle
1982). This gap-replacement process creates forests
with complex age and size structures, and patchy
species composition in the canopy (Lorimer 1980;
Runkle 1982; Yetter and Runkle 1986; Runkle and
Yetter 1987). Although canopy gaps in secondary
forests are hypothesized to be smaller in size, they may
still act as a mechanism for canopy tree replacement,
and stand structural changes associated with older
forests (Clebsch and Busing 1989; Wilder et al. 1999;
Taylor and Lorimer 2003; Cole and Lorimer 2005).
The overarching goal of our study was to document the influence of localized, natural disturbance
events on the development of secondary hardwood
stands during the understory reinitiation stage of
development. Our research was driven by four major
questions. Question 1: What are the patterns and
processes of canopy gap formation prior to a complex
stage of forest development? We hypothesized that
most canopy gaps would be created by uprooted
stems, as windthrow has been widely reported from
many old growth stands and visual observation of the
forest revealed uprooted trees. Question 2: What
percentage of the forest is occupied by canopy gaps
and what are the shape, size, and age distributions for
gaps in developing stands? We hypothesized that the
land fraction of the forest in gaps would be within the
range of variability reported from old growth stands,
but the forest would contain a higher density of
smaller gaps relative to older stands. Question 3: Do
small canopy disturbances influence density and
diversity patterns in secondary stands? We hypothesized that larger gaps would support a higher
number of individuals as well as higher levels of
diversity because they should contain more microsite
heterogeneity, and the likelihood of documenting rare
species should increase by sampling a larger spatial
area. Question 4: How do the gaps close, and what
effects do they have on composition and structure in
developing stands? We hypothesized that most gaps
would close by lateral crown expansion rather than
height growth of subcanopy individuals and would
cause the structure of the forest to move from a high
density of small trees to a lower density of larger
individuals, more typical of older stands.
Forest Ecology
Methods
Study area
The study was conducted in the Pogue Creek Natural
Area (PCNA) located in Fentress County, Tennessee,
in the north-central portion of the state (Fig. 1). The
PCNA is a 1,505 ha reserve managed by the State of
Tennessee, Department of Environment and Conservation, Division of Natural Areas. The PCNA is
located on the Cumberland Plateau section of the
Appalachian Plateaus physiographic province (Fenneman 1938). The underlying geology consists of
Pennsylvanian sandstone, conglomerate, siltstone,
shale, and coal of the Crab Orchard and Crooked
Forked Groups (Smalley 1986). The area has irregular topography (Fenneman 1938) characterized by
long, narrow to moderately broad ridges and narrow
to moderately broad valleys (Smalley 1986). Soils are
Fig. 1 Map of the Pogue
Creek Natural Area,
Fentress County,
Tennessee. Shaded portion
of the Tennessee inset map
is the Cumberland Plateau
physiographic section
133
acidic, highly leached, and low in fertility (Francis
and Loftus 1977; Smalley 1982; USDA 1995; Hart
2007). Depth to bedrock varies from 1 to 1.8 m and
slope gradients range from 15% to 60%. The
elevation of the study plots ranged from 260 to
490 m amsl.
Climate is classified as humid mesothermal with
moderately hot summers and short-mild to moderately cold winters (Thornthwaite 1948). Local
topography strongly influences microclimatic conditions. The average frost-free period is 160 days (from
early-May to late-October) and the mean annual
temperature is 13C. The July average is 23C and
the January average is 2C (USDA 1995). The area
receives steady precipitation during the year with no
distinct dry season. Mean annual precipitation is
137 cm and mean annual snowfall is 50 cm (USDA
1995). Late spring and summer are characterized by
heavy rains that are often accompanied by moderate
134
to severe thunderstorms and strong winds (Smalley
1982).
Braun (1950) classified the area as part of the Cliff
Section of the Mixed Mesophytic Forest Region, but
local topography influences forest composition and
true mesophytic species only dominate on protected
sites. Regionally, forests are intermediate between
mixed mesophytic and Quercus–Carya types (Hinkle
1978; Hinkle 1989; Hinkle et al. 1993). Forest
vegetation patterns of the PCNA were quantified by
Hart and Grissino-Mayer (2008). The forest was
dominated by Carya ovata, Quercus rubra, Q. alba,
and Q. montana. The sparse sapling layer was
dominated by Acer saccharum. The forest was
established in the late 1920s after the cessation of
local logging operations. From field observations and
investigation of 17 tree cross sections from a previous
study, no signs of fire or other large-scale disturbance
events were evident since the anthropogenic disturbances of the 1920s (Hart 2007). Castanea dentata
Marsh was a forest component prior to the arrival of
Cryphonectria parasitica (Murrill) M.E. Barr (chestnut blight). The blight reached the Cumberland
Plateau in the 1920s, and by the end of the 1930s,
most C. dentata in the region were dead. Thus, the loss
of the species roughly coincided with stand initiation.
Field sampling
Canopy gaps (n = 40) were located along transects
throughout the reserve using the line intersect method
(Runkle 1982; Runkle 1985; Veblen 1985; Runkle
1992). Gaps were defined as environments where a
visible void space existed in the main forest canopy,
leaf height of the tallest stems was less than threefourths the height of the adjacent canopy, and gap
makers were present. We did not use a minimum gap
size threshold to document the full range of canopy
gaps. Transects were established parallel to slope
contour beginning at randomly selected points
throughout the forest. All transects were located
along mid-slope positions. We sampled at mid-slope
positions, because the mid-slope forests of the reserve
are indicative of slope forests of the greater Cumberland Plateau region and the majority of forested
land in the reserve occurs along mid-slopes. Total
transect length and transect length in expanded
(boundary defined by the base of surrounding canopy
trees (Runkle 1981)) and true (area unrestricted from
A.G. Van der Valk (ed.)
above) canopy gaps were documented by recording
the number of paces across each. The fraction of land
area in canopy gaps was calculated by dividing the
transect distance in gaps by total transect length
(Runkle 1985; Runkle 1992). At each gap, physical
site characteristics, including percent slope, aspect,
and elevation, were recorded . When walking transects through a forest, large gaps are more likely to be
encountered than relatively small gaps, and sampling
estimators have been created to correct for sampling
bias (see De Vries 1974; Pickford and Hazard 1978).
However, values obtained with the use of estimator
equations and those obtained by simply dividing
transect distance in gaps by total transect length are
similar (Runkle 1985).
Gap area was determined for expanded and true
gaps by, measuring length (largest distance from gap
edge to gap edge) and width (largest distance
perpendicular to the length). These measurements
were fitted to the formula of an ellipse (Runkle 1985;
Runkle 1992). Although gap shapes can be highly
variable (Ferreira de Lima 2005), most gaps at the
PCNA had elliptical shapes, which is common for
forests of the southern Appalachian Highlands (Runkle 1982; Runkle 1992; Clinton et al. 1994). Thus,
fitting the measurements to the formula of an ellipse
was appropriate for this study.
Canopy gaps can be created by several different
mechanisms that remove overstory trees. Biotic and
abiotic forest conditions can be modified differently
by canopy disturbances that are caused by different
gap formation mechanisms. Differences between gap
origins may also influence forest response. In order to
analyze these patterns, gap formation mechanisms
were classified into one of the three categories (snag,
uprooted stem, or snapped stem) according to gap
origin (Clinton et al. 1993). The number of trees
involved in gap formation was also recorded to
document the abundance of single-tree versus multitree events.
Gap maker trees were taxonomically classified to
quantify any species-specific overstory mortality
patterns and possible composition changes associated
with small canopy disturbances. We measured gap
maker diameter at breast height (dbh, ca. 1.4 m above
the surface or root collar for downed individuals) and
length. Basal area (m2) was calculated for all gap
makers that could be accurately measured and totaled
by gap, to determine the amount of basal area lost per
Forest Ecology
disturbance event. This information may be used to
document the amount of biomass naturally removed
from a stand through gap-scale processes. Direction
of gap maker fall relative to slope (i.e., down, across,
or up slope) was also recorded and all gap makers
were placed into one of four decay classes (1–4, with
4 being the most decayed) following criteria adapted
from McCarthy and Bailey (1994).
In each gap, we recorded species, crown class, and
diameter of all trees C5 cm dbh to characterize forest
gap vegetation. Crown class categories (dominant,
codominant, intermediate, and overtopped) were
visually assessed based on the amount and direction
of intercepted light (Oliver and Larson 1996). The
location of each of these individuals was also
recorded as being in either an expanded or true
canopy gap. All saplings (woody stems C1 m height,
\5 cm dbh) in the expanded gap area were tallied by
species to characterize gap regeneration patterns. The
number of perimeter trees with dominant or codominant positions in the canopy was documented for
each gap, to analyze the number of trees required to
complete the canopy surrounding gaps, and the
number of canopy individuals with the potential to
close the void space through lateral crown expansion.
Tree core samples were collected to aid in the
documentation of gap age. A minimum of nine trees
were cored (mean = 18.6) per gap resulting in the
collection of 742 cores. Tree core samples or cross
sections were also collected from all gap makers that
were not in an advanced state of decay (intact bark
and no sapwood degradation), to aid with gap age
determination and to document the seasonal timing of
gap events, based on the amount of xylem produced
during the last year of growth. Dating the seasonality
of tree death and gap formation illustrates a new
approach in dendroecology.
Data analyses
Tree core and cross section samples were prepared
and processed for dating using the methods outlined
in Stokes and Smiley (1996). The samples were airdried, glued to wooden mounts, and sanded to reveal
the cellular structure of the wood (Orvis and
Grissino-Mayer 2002) before tree rings were dated
with the aid of a stereo zoom microscope. All tree
cores were visually analyzed for radial growth
releases to document gap age. In order to document
135
gap maker death dates, tree rings were measured to
the nearest 0.001 mm using a Velmex measuring
stage interfaced with Measure J2X software for all
sampled gap makers. The measurement series were
visually compared to a reference Quercus chronology
developed by Hart and Grissino-Mayer (2008) for the
site. We confirmed the graphical crossdating of all
gap maker tree-ring series using the computer
software COFECHA, a quality-control program that
uses segmented time series correlation analyses to
confirm the placements of all tree rings (Holmes
1983; Grissino-Mayer 2001). In COFECHA, we
tested consecutive 50-year segments (with 25-year
overlaps) on each undated gap maker series to the
reference Quercus chronology. Once statistically
confirmed, we assigned calendar years to all tree
rings in each individual undated measurement series.
All gap ages were confirmed using gap maker decay
classifications.
Canopy gaps can be caused by the removal of a
single tree or a small cluster of trees. Because singletree gaps may result from the death of a large canopy
tree and multi-tree gaps may result from the deaths of
relatively small trees, the amount of basal area lost
between single- and multi-tree gaps was statistically
analyzed using a two-tailed t-test. This information
may be useful to analyze the quantity of basal area
lost by small canopy disturbance events and applied
to harvesting techniques that may mimic natural
disturbance processes.
The rate of gap formation and closure may be
balanced or may vary through time. Non-parametric
correlation techniques were used to analyze the
relationship between land fraction in gaps and gap
age. Gaps may be caused by a variety of formation
mechanisms that differ in the way overstory vegetation is removed, and the mechanism of canopy tree
removal may influence gap size. In order to determine, if a relationship existed between gap size and
gap origin, data were analyzed using a one-way
ANOVA. A Tukey honestly significant difference
(HSD) test was used to compare mean expanded and
true gap sizes across origin categories to determine if
gap size varied by gap formation mechanism.
Length and width of gaps were measured in the
field. Ratios were calculated for length to width
(L:W) of expanded and true gaps to document gap
shape characteristics. This information is useful to
understand the variation in the shape of gaps created
136
by the disturbance and has implications for forest
response and microenvironmental changes within the
gap environment.
For each gap, density and diversity (H0 ) measures
were calculated for saplings, trees, and total stems
(all woody stems C1 m height) to document forest
response to canopy disturbances. Gap size is believed
to influence stem density and diversity. Correlation
coefficients were calculated to determine if a relationship existed between gap size and density of
individuals in gaps. Regression techniques were then
used to model gap size and density relationships. In
order to analyze the relationship between expanded
gap area and diversity patterns, correlation coefficients were calculated for sapling, tree, and total stem
diversity.
Canopy gaps can close by crown expansion of
perimeter trees at canopy level or by the height
growth of understory individuals. The likely closure
mechanism, either by height growth or lateral crown
expansion, of each gap was recorded in the field to
document changes in forest structure following the
removal of canopy trees. Probable gap successors,
which are the individuals that will likely fill the
canopy void, can often be determined in the field
(Barden 1979; Barden 1980; White et al. 1985;
Yamamoto and Nishimura 1999). The documentation
of replacement trees is useful to project the future
composition of the stand and to analyze the influence
of canopy gaps on forest succession. In order to
quantify recruitment following overstory removal,
crown class distributions were constructed for all
trees located in true gap environments for the 15 most
dominant species with canopy potential. These measures may be used to document future canopy trees
and recruitment patterns associated with gap-scale
disturbance processes.
Results
Gap formation patterns and processes
Of the 40 gaps sampled, 8 (20%) were created by
snags, 16 (40%) were created by uprooted stems, and
16 (40%) were created by snapped stems. Eventually,
snag trees will fall, generally during mild to severe
wind events, possibly causing further disturbance to
the forest. It is possible that a gap created by a snag,
A.G. Van der Valk (ed.)
subsequently blown down, was classified incorrectly.
However, measures were taken to avoid this issue,
such as documenting the decay class of gap makers
and noting the position of the gap maker relative to
other downed logs. The number of gap maker trees
involved with opening the canopy ranged from one to
four. The majority (78%) of the canopy gaps involved
the death of only one individual. Of the nine multitree gaps, six (66%) resulted from uprooted stems
including the gap that consisted of the removal of
four canopy individuals, while the three other multitree gaps resulted from snapped boles.
We identified 50 gap maker trees in the 40 canopy
gaps studied. Most gap makers (n = 36, 72%) could
be identified to the species level; however, 4 (8%)
could only be identified to genus and 10 (20%) were
too decayed to be taxonomically classified. Of the 36
gap makers that could be identified to species, 12
different species were represented. The most common species that caused canopy gap formation was
Quercus montana (n = 8). At the genus level, 50% of
all gap makers were Quercus.
Diameter was measured at ca. 1.4 m above the
surface or root collar for 46 gap makers. Diameter
measurements could not be collected for four gap
makers that were in a state of advanced decay.
Average gap maker diameter at breast height was
38.38 cm ± 11.6 (SD). The minimum diameter of a
gapmaker was 19.5 cm and the maximum was 70 cm.
The gap maker with a diameter of 19.5 cm was
involved in a multi-tree uprooting event that also
included the death of an individual with a diameter of
28 cm. Average basal area lost per gap was
0.16 m2 ± 0.10 (SD). The minimum removed was
0.05 m2 and the maximum was 0.52 m2. Multi-tree
gaps (mean = 0.24 m2 ± 0.13 (SD)) resulted in a
larger amount (P \ 0.01) of basal area lost compared
to single-tree events (mean = 0.14 m2 ± 0.08 (SD)).
Age was determined for all canopy gaps by the
identification of radial growth releases, crossdating
the gap makers to document death dates, field
observation, and gap maker decay classification.
Gap ages ranged from 1 to 17 years with a mean of
7 years. Multiple gaps occurred in 13 years. The
highest frequency of gap events during any one year
was five, which occurred during 3 years (1999, 2002,
and 2003).
Gap seasonality was determined for 17 gaps by
examining the amount of xylem produced during the
Forest Ecology
137
last year of growth. Other gapmakers were too
decayed for this analysis. Of these 17 events, only
one occurred during the dormant season. For the
dormant season gap, the latewood portion of the last
ring was complete and buds were still present on the
tree. All other gap makers had partial rings, indicating that the gap events occurred during the growing
season. Because the majority of these individuals had
already completed the production of earlywood prior
to death, we surmise that these events occurred in the
middle or later part of the growing season.
Gap fraction and physical characteristics
Total transect length was 4.47 km, with 15% of the
total length in expanded gaps and true gaps, and 6% in
true canopy gaps only. When percentage values were
standardized at the hectare level, 1,500 m2/ha were in
expanded gaps and 600 m2/ha were in true gap
environments. Total transect length in true canopy
gaps was plotted by gap age to analyze patterns of gap
formation and closure (Fig. 2). The largest amount of
land area in true canopy gaps occurred in gaps that
were 2 years of age and no gap area occurred in gaps
aged 5, 6, 14, 15, or 16 years. A significant negative
relationship existed, where older gaps occupied a
smaller amount of land area relative to younger gaps.
Average expanded gap area was 213.34 m2 ±
108.44 (SD). The maximum expanded gap area was
587.91 m2 and the minimum was 47.10 m2. Average
true gap area when sampled was 42.78 m2 ± 40.16
(SD), with a maximum of 157.84 m2 and a minimum of
1.14 m2. The size of expanded gaps created by
uprooted stems was significantly larger than that of
gaps created by snags (Fig. 3). No other size differences between gap origins were significant.
Fig. 2 Relationship between land fraction in true canopy gaps
and gap age in the Pogue Creek Natural Area in Tennessee
Fig. 3 Mean sizes of expanded and true canopy gaps by gap
origin with standard deviations. Solid bar and different letter
indicate a significant (P \ 0.05) difference between gap
origins as detected by ANOVA and Tukey’s post-hoc testing
The average L:W ratio of expanded gaps was
1.58:1, with a maximum of 3.60:1 and a minimum of
1.01:1. Thus, the average expanded gap was 58%
longer than it was wide. Similar patterns were
observed for true gap areas, for which the mean ratio
was 2.58:1. The maximum length of true gaps was
475% the width. The minimum L:W patterns of
expanded and true gaps revealed circular over
ellipsoidal shapes.
Density and diversity within gaps
The mean number of canopy trees that bordered gaps
was 6.38 ± 1.79 (SD). The maximum number of
perimeter trees was 12, and minimum number of trees
required to complete the canopy around a gap was 4.
In general, larger canopy gaps were bordered by a
higher number of canopy trees relative to smaller
gaps.
The average number of saplings in expanded gaps
was 54.48 ± 28.47 (SD) with a maximum of 144 and
a minimum of 13 (Fig. 4). The mean number of trees
in expanded gaps was 22.73 ± 7.99 (SD) with a
maximum of 44 and minimum of 11 individuals. The
average number of all stems C1 m height in
expanded gaps was 74.20 ± 34.14 (SD). The highest
number of stems in an expanded gap was 188 and the
138
A.G. Van der Valk (ed.)
Table 1 Density of saplings (C1 m height, \5 cm dbh) in
expanded canopy gaps in the Pogue Creek Natural Area in
Tennessee
Species
Fig. 4 Mean number of saplings (C1 m height, \5 cm dbh),
trees (C5 cm dbh), and total stems (all stems C1 m height)
with standard deviations in expanded canopy gaps in the Pogue
Creek Natural Area in Tennessee
lowest number of individuals was 28. The highest
values for saplings and trees occurred in the same gap
that was 10 years old and caused by the uprooting of
four trees.
The sum of all saplings in all expanded gaps was
calculated by species and standardized at the hectare
level to document sapling establishment, and possible
species recruitment in gap environments. The most
abundant species in the sapling layer of expanded
gaps was Acer saccharum followed by Fagus grandifolia and Acer rubrum (Table 1). Together these
three species comprised almost 69% of all saplings in
expanded gaps.
Acer saccharum represented 29.18% of all trees in
true canopy gaps followed by A. rubrum and
Liriodendron tulipifera (Table 2). Collectively, these
three species represent over half of all trees in true
canopy gaps. Dominance (based on basal area) was
also calculated for all canopy gap trees. The most
dominant species were A. saccharum and A. rubrum
(Table 2). The Acer species were followed by a
second tier of species that included L. tulipifera and
Carya ovata. No other species represented more than
6% of the basal area. Species and diameter of all
snags in true canopy gaps were also recorded. A total
of 40 snags were documented and mean snag
diameter at breast height was 10.89 cm ± 6.21
(SD). Of the 40 snags within true gaps, 12 different
species were represented with A. rubrum, A. saccharum and Q. montana being the most common (n = 8
for all species).
Expanded canopy gaps contained 34 different
species in the sapling layer. Mean sapling diversity
(H0 ) was 1.43 ± 0.42 (SD) (Fig. 5). Maximum
Density/
ha
Relative
density
Acer saccharum Marsh.
863.63
35.70
Fagus grandifolia Ehrh.
474.70
19.62
Acer rubrum L.
327.83
13.55
Asimina triloba (L.) Dunal
168.03
6.95
Magnolia acuminata (L.) L.
158.63
6.56
88.13
49.35
3.64
2.04
Fraxinus americana L.
Liriodendron tulipifera L.
Oxydendrum arboreum (L.) DC.
48.18
1.99
Cornus florida L.
37.60
1.55
Ulmus rubra Muhl.
31.73
1.31
Nyssa sylvatica Marsh.
30.55
1.26
Cercis canadensis L.
29.38
1.21
Tilia heterophylla Vent.
12.93
0.53
Aesculus flava Ait.
11.75
0.49
8.23
0.34
Carpinus caroliniana Walt.
Ilex opaca Ait.
8.23
0.34
Magnolia tripetala L.
8.23
0.34
Quercus montana Willd.
8.23
0.34
Carya ovata (P. Mill.) K. Koch
5.88
0.24
Ostrya virginiana (P. Mill.) K. Koch
5.88
0.24
Sassafras albidum (Nutt.) Nees
5.88
0.24
Ailanthus altissima (Mill.) Swingle
Betula lenta L.
4.70
4.70
0.19
0.19
Diospyros virginiana L.
4.70
0.19
Quercus alba L.
4.70
0.19
Ulmus alata Michx.
3.53
0.15
Amelanchier laevis Weig.
2.35
0.10
Carya tomentosa (Poiret) Nutt.
2.35
0.10
Quercus rubra L.
2.35
0.10
Ulmus americana L.
2.35
0.10
Hamamelis virginiana L.
1.18
0.05
Magnolia macrophylla Michx.
1.18
0.05
Morus rubra L.
1.18
0.05
Quercus velutina Lam.
Total
1.18
0.05
2419.33
100.00
sapling layer diversity was 2.22 and the minimum
was 0.78. Total species richness of the tree layer was
28. Average diversity of all trees in expanded gaps
was 1.90 ± 0.35 (SD) with maximum and minimum
values of 2.44 and 1.20, respectively. Mean total
diversity of all stems C1 m height was 1.95 ± 0.36
Forest Ecology
Table 2 Density and
dominance measures for all
trees (stems C5 cm dbh) in
true canopy gaps in the
Pogue Creek Natural Area
in Tennessee
139
Species
Dominance
(m2/ha)
Relative
dominance
Density/
ha
Relative
density
Acer saccharum
807.30
29.18
0.59
24.34
Acer rubrum
391.95
14.16
0.32
13.15
Liriodendron tulipifera
Carya ovata
251.55
146.25
9.09
5.29
0.23
0.21
9.37
8.68
Oxydendrum arboreum
146.25
5.29
0.14
5.91
Fagus grandifolia
175.50
6.34
0.13
5.48
Tilia heterophylla
122.85
4.44
0.10
4.15
Carya tomentosa
81.90
2.96
0.09
3.58
Carya glabra (P. Mill.) Sweet
64.35
2.33
0.08
3.38
Nyssa sylvatica
99.45
3.59
0.08
3.25
Fraxinus americana
70.20
2.54
0.07
2.95
5.85
0.21
0.07
2.85
111.15
4.02
0.06
2.59
Quercus montana
40.95
1.48
0.05
1.94
Magnolia acuminata
52.65
1.90
0.03
1.44
Quercus rubra
23.40
0.85
0.03
1.40
Ulmus rubra
40.95
1.48
0.03
1.16
Cercis canadensis
35.10
1.27
0.02
0.89
Carya cordiformis (Wangenh.) K. Koch
Ostrya virginiana
17.55
23.40
0.63
0.85
0.02
0.02
0.67
0.66
Diospyros virginiana
11.70
0.42
0.01
0.51
Sassafras albidum
11.70
0.42
0.01
0.49
Prunus serotina Ehrh.
5.85
0.21
0.01
0.31
Aesculus flava
5.85
0.21
0.01
0.25
Ulmus alata
5.85
0.21
0.00
0.21
Betula lenta
5.85
0.21
0.00
0.17
Magnolia tripetala
5.85
0.21
0.00
0.13
Ulmus americana
5.85
0.21
0.00
0.10
2767.05
100.00
2.41
100.00
Quercus alba
Cornus florida
Total
Fig. 5 Mean diversity for saplings (C1 m height,\5 cm dbh),
trees (C5 cm dbh), and total stems (all stems C1 m height)
with standard deviations in expanded canopy gaps in the Pogue
Creek Natural Area in Tennessee
(SD). The highest total diversity value was 2.46 and
the lowest was 1.17. Interestingly, diversity patterns
differed by category (i.e., sapling, tree, and total). For
example, the gap with the lowest sapling diversity
was not the same gap with the lowest tree diversity.
However, the gap with the highest sapling and
highest total woody stem diversity values was an
exception.
Significant positive relationships were found for
the number of saplings (r = 0.54, P = 0.0003), trees
(r = 0.73, P \ 0.0001), and total stems (r = 0.62,
P \ 0.0001) (Fig. 6). However, the largest gap did
not contain the highest number of stems, which
140
A.G. Van der Valk (ed.)
Fig. 6 Relationships between the number of saplings (C1 m
height, \5 cm dbh), trees (C5 cm dbh), and total stems (all
stems C1 m height) and expanded gap area in the Pogue Creek
Natural Area in Tennessee
Fig. 7 Relationships between diversity values for saplings
(C1 m height, \5 cm dbh), trees (C5 cm dbh), and total stems
(all stems C1 m height) and expanded gap area in the Pogue
Creek Natural Area in Tennessee
occurred in a gap of an intermediate size class (188
individuals/231.97 m2). A weak negative relationship
existed between sapling diversity and gap size
(r = -0.33, P = 0.04) (Fig. 7). A similar pattern
was also observed for total stem diversity (r = 0.39, P = 0.01). Tree diversity showed no relationship to expanded gap size. Shannon diversity (H0 ) is a
dimensionless index such that gap size would not bias
the diversity measure.
expanded area for all 40 gaps (213.34 m2). The gap
with the largest expanded area (587.91 m2) was
projected to close by understory height growth.
However, a relatively small gap (153.59 m2) was also
projected to close by height growth of a subcanopy
individual.
Of the 10 successor trees documented, five species
were represented (A. saccharum, A. rubrum, C. ovata,
Q. montana, and Quercus alba). Acer rubrum was the
most common gap successor (n = 3) followed by A.
saccharum (n = 2), C. ovata (n = 2), Q. montana
(n = 2), and Q. alba (n = 1). Acer saccharum
represented 28.7% of trees with intermediate positions of all 15 selected species within true gap
environments (Table 3). Acer saccharum was followed by A. rubrum (13.45%) and L. tulipifera
(13.45%), a noted gap-phase species. Collectively,
these three species represented 55.6% of the intermediate trees from the 15 selected species. A similar
Gap closure and recruitment
Of the 40 gaps studied, 10 were projected to close by
height growth of understory individuals and the
remaining 30 gaps were projected to close by lateral
branch growth of canopy trees surrounding the voids.
Mean expanded area of gaps likely to close via the
height growth of understory trees was 285.13 m2 ±
137.58 (SD), which was ca. 34% greater than the mean
Forest Ecology
141
Table 3 Crown class distributions for all trees (stems C5 cm
dbh) in 40 true canopy gaps in the Pogue Creek Natural Area in
Tennessee
Species
Overtopped
Intermediate
Density Relative Density Relative
density
density
Acer saccharum
72
36.36
64
28.70
Acer rubrum
Liriodendron tulipifera
37
13
18.69
6.57
30
30
13.45
13.45
4
2.02
21
9.42
Tilia heterophylla
11
5.56
10
4.48
Oxydendrum arboreum
Carya ovata
16
8.08
9
4.04
Fraxinus americana
3
1.52
9
4.04
Carya glabra
1
0.51
9
4.04
Fagus grandifolia
21
10.61
8
3.59
Carya tomentosa
5
2.53
8
3.59
Quercus alba
1
0.51
8
3.59
1
0.51
6
2.69
Nyssa sylvatica
12
6.06
5
2.24
Quercus rubra
0
0.00
4
1.79
Carya cordiformis
1
0.51
2
0.90
198
100.00
223
100.00
Quercus montana
Total
pattern was observed for overtopped positions, with
A. saccharum being the most abundant (36.36%)
followed by A. rubrum (18.69%) and F. grandifolia
(10.61%).
Discussion
Gap formation patterns and processes
The majority (80%) of the gaps documented originated from uprooted or snapped stems. Other studies
have also found these mechanisms to be the most
common means of gap formation in the southern
Appalachians (Barden 1979; Barden 1981; Romme
and Martin 1982; Runkle 1982). However, Clinton
et al. (1993) found snag gaps to be more prevalent
following drought in secondary forests of the Appalachian Highlands in North Carolina. Based on the
means by which canopy trees were removed, we
speculate that wind is the dominant disturbance agent
in the forest as strong winds have the potential to
uproot trees and snap boles. Wind also has the
potential to alter forest composition and structure by
blowing down snag trees. Standing dead trees are
often removed by mild to severe wind events, but the
potential for snags to be blown down varies by site
conditions (Jans et al. 1993). Further, snags that
eventually fall likely alter the forest differently than
gaps that are caused rapidly (Franklin et al. 1987;
Krasny and Whitmore 1992; Clinton et al. 1994). The
eventual fall of a snag may cause additional forest
disturbance, possibly with a greater magnitude than
the initial event. Also, the bole and branches of
standing dead trees may block sunlight from reaching
the understory, thereby, facilitating gap closure by
perimeter trees rather than subcanopy individuals.
The percentage of single-tree gaps (78% of gaps
sampled) was within the range of what has been
reported from old growth forests of the eastern USA
(Runkle 1990). Of the multi-tree disturbance events,
most were caused by uprooted stems. Windthrow
gaps have the potential to cause more site modification than gaps caused by other mechanisms, because
as the root network is lifted, microtopography (pits
and mounds) and soil characteristics are also modified (Clinton et al. 1994; Beckage et al. 2000).
Average diameter of gap maker trees was
38.38 cm at breast height and the average diameter
of canopy trees (dominant and codominant crown
classes) that surrounded gaps was 38.83 cm ± 6.04
(SD). This result is contrary to what has been
reported for old growth forests of the southern
Appalachians, where gap makers were significantly
larger than border trees (Runkle 1998). This pattern
may be related to the age of the forest. In second
growth forests, canopy trees are within a narrower
diameter range as their age (and diameter) structure is
not complex. Thus, in mature second growth forests,
size does not indicate that one individual is more
likely to be removed from the canopy than another.
Also, the smallest gap maker was just 19.5 cm dbh,
but was a component of a multi-tree gap with another
individual of 28 cm. Although both of these individuals were below the average size for gap makers, the
removal was enough to open the canopy and modify
the forest.
Of the 17 disturbance events with known seasonality, one occurred during the growing season. Most
growing season deaths occurred after the formation of
earlywood cells but before the completion of latewood cells. We interpret the amount of xylem
produced during the last year of growth to indicate
142
that the majority of the canopy disturbance events
occurred during the middle or later part of the
growing season. The timing of death combined with
the primary origins of formation (uprooted and
snapped stems) indicated that strong winds associated
with frontal and convection storms may be the major
agents that disrupt the forest. Severe wind events in
the region are associated with thunderstorms that
occur ca. 55 days per year, usually during late-spring
and summer (Smalley 1982). Documenting the season of gap formation is important because the time of
year a gap forms may influence the ability of residual
trees to exploit the additional resources (Runkle
1989). Gaps that form during the growing season may
expose shade-developed leaves to changes in environmental conditions. When light levels increase,
expanding leaves and leaves produced in the new
environment may acclimate to high-light conditions.
Fully shade-developed leaves are not able to change
their anatomy to acclimate to modifications in the
light regime. Thus, fully shade-developed leaves may
undergo a period of photoinhibition after gap events
(Kozlowski 1957; Naidu and DeLucia 1997). Plastic
species that periodically flush throughout the growing
season may be best suited to take advantage of the
increased resources of the gap environment
(Kozlowski and Pallardy 1997). Gap seasonality
may be especially important in secondary forests,
where gaps are generally small and relatively shortlived.
Gap fraction and physical characteristics
The fraction of land area in expanded gaps and true
gaps is within the range of what has been reported
elsewhere in the Eastern Deciduous Forest Region
(Runkle 1982; Beckage et al. 2000). Based on gap
fraction and mean gap size, we conclude that the
secondary forest supported a higher number of gaps
compared to older stands throughout the region.
Thus, we propose that during the understory reinitiation stage of development, forests support a higher
density of smaller gaps, but similar total land area in
gap environments compared to stands in a complex
developmental stage. The fraction of land area in
gaps was highest for younger gaps and generally
decreased with increased gap age. This pattern was
expected because older gaps have had a longer time
to be filled. No gaps were documented over 17 years
A.G. Van der Valk (ed.)
of age. From this, we propose that most gaps in the
forest are filled within 20 years of formation, but
many are likely filled within shorter periods. Hart and
Grissino-Mayer (2008) statistically analyzed radial
growth releases attributed to canopy disturbances in
Quercus individuals from the PCNA and found mean
release durations of only 4 years. Thus, we speculate
that most gaps are short-lived in these secondary
hardwood stands, and that the increase in available
resources is generally not sustained for more than
4 years.
In general, L:W patterns were similar for expanded
and true gaps as both had ellipsoidal shapes. The
shape of the disturbed canopy area is largely a
function of the mechanism of tree death and architecture of the gap maker. Circular gap shapes resulted
from canopy disturbances related to snags rather than
uprooted or snapped stems. Trees that remained
standing after their death did not fall, and remove
vegetation in a linear pattern starting at the base of
the tree as is normal for treefall disturbances. The
majority (55%) of the gaps were oriented downslope
from the base of the tree, while 40% were oriented
across slope and 5% of the trees fell up slope. Thus,
most canopy disturbances resulted in ellipsoidal
shaped gaps that were oriented perpendicular to
slope contours. The shape, size, slope, orientation,
height of surrounding forest, residual plants, and
post-treefall debris of canopy gaps, as well as
latitudinal position, are important in determining the
microenvironmental conditions of the disturbed area
(Poulson and Platt 1989; Runkle 1989). These
physical gap characteristics may be useful if forest
management plans have a goal of mimicking natural
disturbance processes.
Density and diversity within gaps
Significant positive relationships were documented
between expanded gap size and the number of
saplings, trees, and total stems. Although, this result
may seem expected, a significant positive relationship
between gap size and stem number does not always
occur (Runkle 1982). Larger gaps may be the result
of high intensity events with few residual trees, may
be characterized by abiotic conditions (e.g., full
sunlight and high temperatures) not conducive to the
growth of forest interior species, and may support
increased herbivory. Interestingly, in our study, the
Forest Ecology
largest gap did not support the highest number of
stems, which occurred in an intermediate size gap.
The density measures analyzed were for expanded
gap areas. Individuals in the entire area of an
expanded gap do not benefit from increased resources
such as light. Because the true gap area gets smaller
with time since the disturbance, gap age may play an
important role in the number of individuals that
inhabit a gap site (Runkle 1982; Clinton et al. 1994).
Also, the number of individuals within the gap should
decrease through self-thinning processes as they
compete to reach the canopy.
We hypothesized that larger gaps would support
higher diversity values. By containing more surface
area, larger gaps have the potential to contain more
site heterogeneity and microsites that may favor
certain species over others. However, only weak
relationships existed between diversity and gap size
and two of the relationships (saplings and total stems)
were negative. Perhaps, gap size is not as important
to diversity as the physical site characteristics of the
gap or the biotic assemblage of the gap prior to
formation. It is also possible that the gaps sampled in
this study (and those in other secondary forests of
similar age) were not large enough for the pattern to
manifest. Although canopy gaps should increase
biological diversity, this pattern does not necessarily
occur at the gap-level, but at the stand-level, where a
collection of different sized and aged canopy gaps
across a variety of sites may support species that are
otherwise absent or sparse under the closed forest
canopy.
Gap closure, recruitment, and succession
No clear species-specific patterns were observed with
gap successors, indicating that the location of an
individual within the gap and its vertical crown
position prior to the disturbance may be the most
important factors that determine how the gap is
closed, and by what species. As further evidence of
this point, radial-growth response of understory
individuals has been shown to be related to position
within a canopy gap (Tryon et al. 1992).
Species composition of gaps is a good predictor of
future forest composition (Runkle and Yetter 1987).
Three species (A. saccharum, F. grandifolia, and
A. rubrum) represented 69% of all saplings in
expanded gaps. Because saplings represent the pool
143
of individuals that may be recruited to larger size
classes following disturbance events, we hypothesize
that Acer species and F. grandifolia will become
more abundant in intermediate, and eventually,
canopy positions under the current disturbance
regime. There is a greater likelihood that individuals
of these species will be able to exploit current and
future gap events, because they are so abundant in the
sapling layer.
We projected that the majority of the canopy gaps
would close by lateral crown expansion rather than
height growth of understory individuals. However,
even gaps that close by lateral branch growth still
provide a means for understory trees to recruit to
larger size classes. This process may allow overtopped trees to reach intermediate positions, and
eventually, the canopy following future disturbance
events. Trees already in intermediate positions may
expand their crowns to become dominant or codominant in the canopy. Over half of all trees located in
true gaps with intermediate crown classifications
were A. saccharum, A. rubrum, or L. tulipifera.
Acer species and L. tulipifera have the greatest
potential to recruit in gaps based on density, dominance, and crown class measures. It is interesting that
species with such different life history characteristics
were well represented in canopy gaps, and employ
different strategies to reach canopy level. Acer
saccharum is very shade-tolerant and has the ability
to persist in the understory of a closed canopy while
maintaining the ability to rapidly respond to
increased light (Canham 1988; Tryon et al. 1992).
Acer rubrum is classed as moderately shade-tolerant
and can exist in the understory of a relatively closed
canopy until the formation of gaps when the species
has also been shown to quickly respond to increased
resources (Wallace and Dunn 1980). In general, the
life history characteristics of the Acer species may be
classed as conservative. Both A. saccharum and A.
rubrum can establish in at least relatively shaded
conditions and wait for the formation of small canopy
gaps to recruit to larger size classes and higher
canopy positions. In contrast to the Acer species, L.
tulipifera is disturbance obligate. The species is
shade-intolerant and cannot exist under a closed
forest canopy. However, L. tulipifera is capable of
quickly responding to increased resources when they
become available and is a common component in
forests with disturbance regimes that consist of small
144
localized events (Buckner and McCracken 1978;
Wallace and Dunn 1980; Orwig and Abrams 1994;
Busing 1995; Lafon 2004).
Shade-intolerant species, such as L. tulipifera,
must reach the canopy in one gap event, because
once the gap closes, they will not be able to survive
under the canopy (Hibbs 1982; Runkle and Yetter
1987; Cole and Lorimer 2005; Webster and Lorimer
2005). Thus, in forests with a disturbance regime
characterized by small localized events, there are
few opportunities for shade-intolerant species to
exist (Runkle 1998). Shade-tolerant species are more
likely to be present in a gap when they form; thus,
they are generally more likely to recruit or reach the
canopy in small gaps (Henry and Swan 1974; Dahir
and Lorimer 1996; McClure et al. 2000; Taylor and
Lorimer 2003). Gaps of a larger size, however, may
allow time for germinants to establish and recruit.
Because of its rapid growth, L. tulipifera can reach
the canopy in gaps of ca. 400 m2 and larger (Busing
1994; Busing 1995). Although gaps of that size were
documented in this study, we do not think L.
tulipifera will reach the canopy at any of the gap
sites before closure. The majority of the gaps
documented will close by lateral branch growth
before understory individuals can reach the canopy.
Although individuals are recruited to larger size
classes and higher vertical positions in these gaps, it
will take multiple disturbance events for most
subcanopy individuals to reach the main canopy
level. Because it will generally take multiple events
for individuals to be recruited to the canopy, shadetolerant A. saccharum and moderately-tolerant A.
rubrum are the species most likely to attain canopy
dominance under the current disturbance regime.
Interestingly, half of the canopy gaps were caused
by the removal of a Quercus individual, but Acer
species represented a large proportion of trees likely
to either reach the canopy or recruit to larger size
classes. These data indicate a likely shift in composition if gap processes continue to remove
Quercus from the canopy and provide the means
for Acer recruitment.
Conclusions
Canopy gaps obviously have an important influence
on forest composition and structure. However, little
A.G. Van der Valk (ed.)
information is available on natural gap-scale disturbances in secondary hardwood forests. By analyzing
gap formation mechanisms, physical gap characteristics, and forest response to canopy gaps, we can
gain a better understanding of the role of gap-scale
disturbance processes in the development of hardwood forests. This study showed that disturbances
that involved the death of a single tree or a small
cluster of trees were common events throughout
secondary stands on the Cumberland Plateau in
Tennessee. The fraction of the forest in canopy gaps
was within the range reported from old growth
remnants throughout the region. However, gap size
was smaller in the developing stands indicating that
secondary forests contain a higher density of smaller
gaps. These localized disturbances modified biomass
arrangement and tree-age distribution patterns as they
allowed for crown expansion of canopy trees,
recruitment of understory individuals, and in some
instances, establishment of new germinants. Thus,
canopy gaps provide the mechanism for forests to
develop a complex size and age structure indicative
of older stands. Gap-scale processes may also be used
to help explain shifting species composition that has
been widely reported throughout the Central Hardwood Forest Region of the eastern US. Half of the
canopy gaps documented in this study were caused by
the removal of a Quercus individual, but A. saccharum, A. rubrum, and L. tulipifera were the most likely
species to capture canopy gaps. The gaps documented
favored the very shade-tolerant A. saccharum
because most gaps were small, and multiple overstory
removal events would be required for trees to reach
the main canopy level. In conclusion, this study
demonstrated that natural disturbance processes have
significant influences on forest development and
successional patterns. Thus, small-scale disturbance
events must be considered when developing longterm forest management plans.
Acknowledgments This research was funded by National
Science Foundation grant BCS-0602445. Justin Hart was
supported by the National Science Foundation under grant
DGE-0538420. We thank Ryan Foster and Kevin Horn for
assistance in the field, and Matthew McConnell and Saskia van
de Gevel for assistance in the laboratory. We also thank the
Tennessee Department of Environment and Conservation for
sampling permission and access to the reserve. Wayne
Clatterbuck, Carol Harden, Sara Hart, and Sally Horn
provided useful comments that greatly improved the
manuscript.
Forest Ecology
References
Barden LS (1979) Tree replacement in small canopy gaps of a
Tsuga canadensis forest in the southern Appalachians,
Tennessee. Oecologia 44:141–142. doi:10.1007/BF0034
6412
Barden LS (1980) Tree replacement in a cove hardwood forest of
the southern Appalachians. Oikos 35:16–19. doi:10.2307/
3544722
Barden LS (1981) Forest development in canopy gaps of a
diverse hardwood forest of the southern Appalachian
Mountains. Oikos 37:205–209. doi:10.2307/3544466
Beckage B, Clark JS, Clinton BD et al (2000) A long-term study
of tree seedling recruitment in southern Appalachian forests: the effects of canopy gaps and shrub understories.
Can J For Res 30:1617–1631. doi:10.1139/cjfr-30-10-1617
Braun EL (1950) Eastern deciduous forests of North America.
Blakiston, Philadelphia
Buckner E, McCracken W (1978) Yellow-poplar: a component
of climax forests? J For 76:421–423
Busing RT (1994) Canopy cover and tree regeneration in oldgrowth cove forests of the Appalachian Mountains. Vegetatio 115:19–27
Busing RT (1995) Disturbance and the population dynamics of
Liriodendron tulipifera: simulations with a spatial model of
forest succession. J Ecol 83:45–53. doi:10.2307/2261149
Canham CD (1988) Growth and canopy architecture of shadetolerant trees: response to canopy gaps. Ecology 69:786–
795. doi:10.2307/1941027
Cho D, Boerner REJ (1991) Canopy disturbance patterns and
regeneration of Quercus species in two old-growth forests.
Vegetatio 93:9–18
Clebsch EEC, Busing RT (1989) Secondary succession, gap
dynamics, and community structure in a southern Appalachian cove forest. Ecology 70:728–735. doi:10.2307/194
0223
Clinton BD, Boring LR, Swank WT (1993) Canopy gap
characteristics and drought influence in oak forests of the
Coweeta Basin. Ecology 74:1551–1558. doi:10.2307/
1940082
Clinton BD, Boring LR, Swank WT (1994) Regeneration
patterns in canopy gaps of mixed-oak forests of the
southern Appalachians: influence of topographic position
and evergreen understory. Am Midl Nat 132:308–319.
doi:10.2307/2426587
Cole WG, Lorimer CG (2005) Probabilities of small-gap capture by sugar maple saplings based on height and crown
growth data from felled trees. Can J For Res 35:643–655.
doi:10.1139/x04-210
Cowell CM (1998) Historical change in vegetation and disturbance on the Georgia Piedmont. Am Midl Nat 140:78–
89. doi:10.1674/0003-0031(1998)140[0078:HCIVAD]2.0.
CO;2
Dahir SE, Lorimer CG (1996) Variation in canopy gap formation among developmental stages of northern hardwood
stands. Can J For Res 26:1875–1892. doi:10.1139/x26-212
De Vries PG (1974) Multi-stage line intersect sampling. For
Sci 20:129–133
Fenneman NM (1938) Physiography of Eastern United States.
McGraw-Hill Book Company, New York
145
Ferreira de Lima RA (2005) Gap size measurement: the proposal of a new field method. For Ecol Manag 214:413–
419. doi:10.1016/j.foreco.2005.04.011
Francis JK, Loftus JS (1977) Chemical and physical properties
of the Cumberland Plateau and Highland Rim forest soils.
USDA, Forest Service, Research Paper SO-138
Franklin JF, Shugart HH, Harmon ME (1987) Tree death as an
ecosystem process. Bioscience 37:550–556. doi:10.2307/
1310665
Fritts HC (2001) Tree rings and climate. Blackburn Press,
Caldwell, New Jersey
Goebel PC, Hix DM (1996) Development of mixed-oak forests
in southeastern Ohio: a comparison of second and oldgrowth forests. For Ecol Manag 84:1–21. doi:10.1016/
0378-1127(96)03772-3
Goebel PC, Hix DM (1997) Changes in the composition and
structure of mixed-oak, second-growth forest ecosystems
during the understory reinitiation stage of stand development. Ecoscience 4:327–339
Grissino-Mayer HD (2001) Evaluating crossdating accuracy: a
manual and tutorial for the computer program COFECHA. Tree Ring Res 57:205–221
Hart JL (2007) A multi-scale analysis of disturbance dynamics in
hardwood forest communities on the Cumberland Plateau,
USA. Dissertation, University of Tennessee, Knoxville
Hart JL, Grissino-Mayer HD (2008) Vegetation patterns and
dendroecology of a mixed hardwood forest on the Cumberland Plateau: implications for stand development. For
Ecol Manag 255:1960–1975
Henry JD, Swan JMA (1974) Reconstructing forest history
from live and dead plant material—an approach to the
study of forest succession in southwest New Hampshire.
Ecology 55:772–783. doi:10.2307/1934413
Hibbs DE (1982) Gap dynamics in a hemlock-hardwood forest.
Can J For Res 12:522–527. doi:10.1139/x82-081
Hinkle CR (1978) The relationship of forest communities and
selected species to edaphic and topographic factors on the
Cumberland Plateau of Tennessee. Dissertation, University of Tennessee, Knoxville
Hinkle CR (1989) Forest communities of the Cumberland
Plateau of Tennessee. J Tenn Acad Sci 64:123–129
Hinkle CR, McComb WC, Safely JM Jr, Schmalzer PA (1993)
Mixed mesophytic forests. In: Martin WH, Boyce SG,
Echternacht AC (eds) Biodiversity of the Southeastern
United States: upland terrestrial communities. Wiley, New
York, pp 203–253
Holmes RL (1983) Computer assisted quality control in treering dating and measurement. Tree Ring Bull 43:69–78
Jans L, Porter L, van Rompaey RSAR et al (1993) Gaps and
forest zones in tropical moist forest in Ivory Coast. Biotropica 25:258–269. doi:10.2307/2388784
Kozlowski TT (1957) Effect of continuous high light intensity on
photosynthesis of forest tree seedlings. For Sci 3:220–224
Kozlowski TT, Pallardy SG (1997) Physiology of woody
plants, 2nd edn. Academic Press, San Diego
Krasny ME, Whitmore MC (1992) Gradual and sudden forest
canopy gaps in Allegheny northern hardwood forests. Can
J For Res 22:139–143. doi:10.1139/x92-019
Lafon CW (2004) Stand dynamics of a yellow-poplar (Liriodendron tulipifera L.) forest in the Appalachian
146
Mountains, Virginia, USA. Dendrochronologia 22:43–52.
doi:10.1016/j.dendro.2004.09.002
Lorimer CG (1980) Age structure and disturbance history of a
southern Appalachian virgin forest. Ecology 61:1169–
1184. doi:10.2307/1936836
McCarthy BC, Bailey DR (1994) Distribution and abundance
of coarse woody debris in a managed forest landscape.
Can J For Res 24:1317–1329
McClure JW, Lee TD, Leak WB (2000) Gap capture in
northern hardwoods: patterns of establishment and height
growth in four species. For Ecol Manag 127:181–189. doi:
10.1016/S0378-1127(99)00129-2
Naidu SL, DeLucia EH (1997) Acclimation of shade-developed leaves on saplings exposed to late-season canopy
gaps. Tree Physiol 17:367–376
Oliver CD, Larson BC (1996) Forest stand dynamics, update
edition. Wiley, New York
Orvis KH, Grissino-Mayer HD (2002) Standardizing the
reporting of abrasive papers used to surface tree-ring
samples. Tree Ring Res 58:47–50
Orwig DA, Abrams MD (1994) Contrasting radial growth and
canopy recruitment patterns in Liriodendron tulipifera and
Nyssa sylvatica: a gap-obligate versus gap-facultative tree
species. Can J For Res 24:2141–2149. doi:10.1139/x94-276
Pickford SG, Hazard JW (1978) Simulation studies on line
intersect sampling of forest residue. For Sci 24:469–483
Poulson TL, Platt WJ (1989) Gap light regimes influence
canopy tree diversity. Ecology 70:553–555. doi:10.2307/
1940202
Rebertus AJ, Meier AJ (2001) Blowdown dynamics in oakhickory forests of the Missouri Ozarks. J Torrey Bot Soc
128:362–369. doi:10.2307/3088668
Romme WH, Martin WH (1982) Natural disturbance by treefalls in old-growth mixed mesophytic forest: Lilley Cornett Woods, Kentucky. In: Muller RN (ed) In:
Proceedings of the central hardwood forest conference.
University of Kentucky, Lexington, pp 367–383
Runkle JR (1981) Gap regeneration in some old-growth forests
of the eastern United States. Ecology 62:1041–1051. doi:
10.2307/1937003
Runkle JR (1982) Patterns of disturbance in some old-growth
mesic forests of the eastern United States. Ecology
63:1533–1546. doi:10.2307/1938878
Runkle JR (1985) Comparison of methods for determining
fraction of land area in treefall gaps. For Sci 31:15–19
Runkle JR (1989) Synchrony of regeneration, gaps, and latitudinal differences in tree species diversity. Ecology
70:546–547. doi:10.2307/1940199
Runkle JR (1990) Gap dynamics in an Ohio Acer-Fagus forest
and speculations on the geography of disturbance. Can J
For Res 20:632–641. doi:10.1139/x90-085
Runkle JR (1992) Guidelines and sample protocol for sampling
forest gaps. USDA, Forest Service, Pacific Northwest
Research Station, GTR 283, Portland
Runkle JR (1998) Changes in southern Appalachian canopy
tree gaps sampled thrice. Ecology 79:1768–1780
Runkle JR (2000) Canopy tree turnover in old-growth mesic
forests of eastern North America. Ecology 81:554–576
A.G. Van der Valk (ed.)
Runkle JR, Yetter TC (1987) Treefalls revisited: gap dynamics
in the southern Appalachians. Ecology 68:417–424. doi:
10.2307/1939273
Smalley GW (1982) Classification and evaluation of forest
sites on the mid-Cumberland Plateau. USDA, Forest
Service, Southern Research Experiment Station, GTR SO38, New Orleans, Louisiana
Smalley GW (1986) Classification and evaluation of forest
sites on the northern Cumberland Plateau. USDA, Forest
Service, Southern Research Experiment Station, GTR SO60, New Orleans, Louisiana
Spies TA, Franklin JF, Klopsch M (1990) Canopy gaps in
Douglas-fir forests of the Cascade Mountains. Can J For
Res 20:649–658. doi:10.1139/x90-087
Stokes MA, Smiley TL (1996) An introduction to tree-ring
dating. University of Arizona Press, Tucson
Taylor SO, Lorimer CG (2003) Loss of oak dominance in drymesic deciduous forests predicted by gap capture methods.
Plant Ecol 167:71–88. doi:10.1023/A:1023975026261
Thornthwaite CW (1948) An approach toward rational classification of climate. Geogr Rev 38:55–94. doi:10.2307/
210739
Tryon EH, Lanasa M, Townsend EC (1992) Radial growth
response of understory sugar maple (Acer saccharum)
surrounding openings. For Ecol Manag 55:249–257. doi:
10.1016/0378-1127(92)90104-H
Tyrell LE, Crow TR (1994) Structural characteristics of oldgrowth hemlock-hardwood forests in relation to age.
Ecology 75:370–386. doi:10.2307/1939541
USDA (1995) Soil survey of Fentress and Pickett Counties.
USDA, Soil Conservation Service, Tennessee
Veblen TT (1985) Forest development in tree-fall gaps in the
temperate rain forest of Chile. Natl Geogr Res 1:162–183
Wallace LL, Dunn EL (1980) Comparative photosynthesis of
three gap phase successional tree species. Oecologia
45:331–340. doi:10.1007/BF00540201
Webster CR, Lorimer CG (2005) Minimum opening sizes for
canopy recruitment of midtolerant tree species: a retrospective approach. Ecol Appl 15:1245–1262. doi:10.1890/
04-0763
White PS, MacKenzie MD, Busing RT (1985) Natural disturbance and gap-phase dynamics in southern Appalachian
spruce-fir forests. Can J For Res 15:233–240. doi:10.1139/
x85-041
Wilder CM, Holtzclaw FW Jr, Clebsch EEC (1999) Succession, sapling density and growth in canopy gaps along a
topographic gradient in a second growth east Tennessee
forest. Am Midl Nat 142:201–212. doi:10.1674/00030031(1999)142[0201:SSDAGI]2.0.CO;2
Yamamoto S, Nishimura N (1999) Canopy gap formation and
replacement pattern of major tree species among development stages of beech (Fagus crenata) stands, Japan.
Plant Ecol 140:167–176. doi:10.1023/A:1009713002039
Yetter TC, Runkle JR (1986) Height growth rates of canopy
gap tree species in southern Appalachian gaps. Castanea
51:157–167
Plurality of tree species responses to drought perturbation
in Bornean tropical rain forest
D. M. Newbery Æ M. Lingenfelder
Originally published in the journal Plant Ecology, Volume 201, No. 1, 147–167.
DOI: 10.1007/s11258-008-9533-8 Springer Science+Business Media B.V. 2008
Abstract Drought perturbation driven by the El
Niño Southern Oscillation (ENSO) is a principal
stochastic variable determining the dynamics of
lowland rain forest in S.E. Asia. Mortality, recruitment and stem growth rates at Danum in Sabah
(Malaysian Borneo) were recorded in two 4-ha plots
(trees C 10 cm gbh) for two periods, 1986–1996 and
1996–2001. Mortality and growth were also recorded
in a sample of subplots for small trees (10 to \50 cm
gbh) in two sub-periods, 1996–1999 and 1999–2001.
Dynamics variables were employed to build indices
of drought response for each of the 34 most abundant
plot-level species (22 at the subplot level), these
being interval-weighted percentage changes between
periods and sub-periods. A significant yet complex
effect of the strong 1997/1998 drought at the forest
community level was shown by randomization procedures followed by multiple hypothesis testing.
Despite a general resistance of the forest to drought,
large and significant differences in short-term
responses were apparent for several species. Using
a diagrammatic form of stability analysis, different
species showed immediate or lagged effects, high or
low degrees of resilience or even oscillatory dynamics. In the context of the local topographic gradient,
species’ responses define the newly termed
D. M. Newbery (&) M. Lingenfelder
Institute of Plant Sciences, University of Bern,
Altenbergrain 21, CH 3013 Bern, Switzerland
e-mail: david.newbery@ips.unibe.ch
perturbation response niche. The largest responses,
particularly for recruitment and growth, were among
the small trees, many of which are members of
understorey taxa. The results bring with them a novel
approach to understanding community dynamics: the
kaleidoscopic complexity of idiosyncratic responses
to stochastic perturbations suggests that plurality,
rather than neutrality, of responses may be essential
to understanding these tropical forests. The basis to
the various responses lies with the mechanisms of
tree-soil water relations which are physiologically
predictable: the timing and intensity of the next
drought, however, is not. To date, environmental
stochasticity has been insufficiently incorporated into
models of tropical forest dynamics, a step that might
considerably improve the reality of theories about
these globally important ecosystems.
Keywords Dynamics Ecosystem El Niño
Resilience Stem growth Tree mortality
Introduction
Tropical rain forests are highly dynamic and responsive ecosystems. Their physical structure and
processes may remain relatively stable over time, but
species composition is thought to constantly fluctuate
around a quasi-equilibrium or change slowly in the
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_12
147
148
long-term (Huston 1979, 1994). Perturbations drive
these dynamics, and earlier ideas of equable tropical
conditions have given way to the view that climate is
indeed variable, particularly on the scale of decades to
centuries (Newbery et al. 1999a). Droughts are the
major cause of perturbation across much of South-East
Asia, and are probably a main determinant of forest
structure and tree species composition, in particular in
Borneo. They are often associated with the El Niño
Southern Oscillation (ENSO) cycle (Trenberth 1997;
Trenberth and Hoar 1997; McPhaden 1999; Cane
2005). The last strong drought in Borneo was in 1997/
1998 and prior to that 1982/1983 (Walsh 1996a, b;
Walsh and Newbery 1999).
Global change scenarios for the tropics expect
climate to become more unstable this century with
more frequent and intense droughts (Hulme and
Viner 1998; Timmermann et al. 1999). Not all
models, however, predict more ENSO activity (Timmermann et al. 2004; Cane 2005; McPhaden et al.
2006; Meehl et al. 2007) although the risk of
droughts is expected to increase during future El
Niño events (Christensen et al. 2007), and this could
have serious implications for forest management.
Tropical rain forests appear to exist in variously
complex and overlapping states of recovery from past
perturbations, whether singular or multiply clustered,
recent or from the distant past (Newbery et al. 1999a,
b; Newbery and Lingenfelder 2004). Measuring how
these forests respond to the perturbations could lead
to valuable models of how changes in drought
frequency and intensity affect their future. It might
also allow a rethinking in tropical rain forest
dynamics with broader considerations for rain forest
conservation.
Forests in parts of Borneo are apparently still
recovering from a very strong drought c. 130 years
ago (Newbery et al. 1992; Walsh and Newbery 1999)
with tree species appearing well adapted to the
several less-severe intervening events (Gibbons 1998;
Newbery et al. 1999b; Gibbons and Newbery 2003;
Newbery and Lingenfelder 2004). Setbacks in forest
biomass temporarily destabilize the ecosystem, but
over centuries a form of dynamic equilibrium is
presumably attained. Uncertain though is how far
from this equilibrium are these forests due to the
1997/1998 and earlier droughts, and whether they
have a high enough resilience to recover before the
next one (Newbery and Lingenfelder 2004).
A.G. Van der Valk (ed.)
In Central America, intensification of the 1982/
1983 dry season by the ENSO that year, together with
a regional trend of decreasing rainfall since 1965,
have been shown to affect tree mortality and forest
population change, and account for species’ geographical distributions (Condit et al. 1995, 1996,
2004; Engelbrecht et al. 2007). This situation contrasts in interesting ways to that of Borneo with its
regionally steady environment punctuated by occasional strong drought. Indeed when generally
comparing drought effects worldwide, it is important
to place these usual periods of large water deficit
within the context of any long-term regular (annual or
supra-annual) patterns characterizing the regional
climate.
This paper reports on the impact of this 1997/1998
drought on lowland dipterocarp rain forest dynamics
in the Danum Valley Conservation Area, Sabah,
Malaysia (Marsh and Greer 1992), a site c. 70 km
inland of the north-eastern coast of Borneo. With
precise enumeration data over a 15-year period,
collected before and after the drought on a large
sample of trees, changes to the most abundant species
and estimated ecosystem resilience are quantified. An
omnibus test of the null model that there were no
species-specific responses at the community level
was performed. An attempt of this kind for tropical
ecosystem dynamics has to our knowledge not
hitherto been made. Finally, a new concept of rain
forest dynamics emerging from this work is
presented.
Climate
Meteorological data have been recorded at Danum
Valley Field Centre (DVFC, 152 m a.s.l.) since July
1985. Monthly mean temperatures ranged 1.8C
about an annual mean temperature of 26.8C, while
the mean daily range was 8.6C. The highest
temperature was recorded in April 1992 with
36.5C, the lowest in January 1993 with 16.5C.
Relative humidity varied between 95.3% at 08:00 and
78.3% at 14:00. These values are typical of equatorial
rainforest locations (Walsh 1996b).
Mean annual rainfall (±SE) across complete years
1986–2003 was 2,825 ± 110 mm with a range from
1,918 mm in 1,997–3,539 mm in 2003. During the
study period of 1986 to 2001 mean annual rainfall was
Forest Ecology
2,787 ± 115 mm. Annual monthly rainfall (±SE)
from July 1985 to December 2003 was 235 ± 13 mm
ranging from 158 mm in April to 312 mm in January.
For the study period the corresponding mean was
232 ± 13 mm. Mean rainfall in the month of April
was significantly different from the annual monthly
mean from 1985 to 2003 (Mann–Whitney U-test:
P = 0.006), varying from 11 mm in 1998—the lowest
monthly value on record—to 433 mm in 1999, the
wettest month of that year. With 701.2 mm of rain,
December 2003 was the wettest month on record at
Danum. Although rainfall in the months of April and
July/August on average was well below the annual
monthly mean, rainfall fluctuated considerably
between years as well as between months. There is
no clear dry season indicating that Danum has a
generally aseasonal tropical climate.
Since the start of meteorological data collection,
Danum experienced 38 droughts, defined as periods
with running 30-day rainfall total (R30) \ 100 mm for
rain forests not short of water (Brünig 1969; Malhi and
Wright 2004). These include two ENSO-related
drought events in 1991/1992(–1994) and in 1997/
1998. Before that, in 1982/1983, a strong ENSO-event
affected Sabah (Beaman et al. 1985; Woods 1989) and
may have had effects at Danum. Very strong droughts
have been recorded in the late 19th and early 20th
centuries at regional scales throughout Sabah (Walsh
1996a, b). In 1997/1998 drought effects on forest
vegetation were reported to be stronger in Sarawak
(Nakagawa et al. 2000; Potts 2003).
149
This approach, however, neglects the rainfall
preceding the 30-day periods. A period of rainfall
below 100 mm that had low rainfall in the months
before would, assuming that the soil water reservoir
was already depleted, likely be more severe for the
trees than such a period with high rainfall preceding
it, in which case water would probably be still
available from storage in the soil (Malhi and Wright
2004; R. P. D. Walsh, pers. comm.). Water stress
caused by a deficit in soil water tends to affect the
forest immediately (in \15 days), but some time is
required (c. 60 days in a central Amazonian rain
forest) for the soil to be recharged with water after a
dry season (Malhi et al. 2002).
Antecedent rainfall history was brought into the
calculation of drought intensity in the following way.
Across the 18.5 years for which data were available,
Julian-day rainfall was averaged to give mean values
of what the vegetation might ‘expect’—the annual
distribution of rainfall to which its species have
generally been subjected. This was termed the mean
daily rainfall, MDR (Fig. 1). The difference between
the actual and mean daily rainfall (ADR–MDR), the
daily rainfall anomaly (DRA), was accumulated
across 90, 180 and 365 days prior to each day (the
accumulated rainfall anomaly, ARA). For any one day
ARA gave the sum of rainfall across the selected
period that was a deficit or a surplus to the expected
average for that period.
Plots and enumerations
Methods
Antecedent rainfall history
For rain forest vegetation that is not short of water,
water stress is assumed to set in when the monthly
mean rainfall drops below 100 mm, the estimated
value for evapotranspiration in the tropics (Brünig
1969, 1971; Walsh 1996b; Walsh and Newbery 1999;
Malhi et al. 2002; Malhi and Wright 2004). For the
daily rainfall data available, this threshold can be
applied to R30, the 30-day running total of rainfall.
Droughts can be assessed by calculating an intensity
index that takes into account the deficit (R30–100) and
the drought duration (Newbery and Lingenfelder
2004).
Two permanent plots were set up and first enumerated in 1985–1986. They lie 0.8 km NW of Danum
Valley Field Centre, 0.3 km apart on gently undulating terrain at c. 200–250 m asl. Each plot has
dimensions 100 m 9 400 m (total area = 8 ha).
Eight 40 m 9 40 m subplots were selected in a
stratified random manner within each plot in 1998
(area = 2.56 ha), half on lower slopes (\12 m elevation) and half on ridges (C25 m). For further
information about the site and the plots see Newbery
et al. (1992, 1996) and Newbery and Lingenfelder
(2004).
Plots were enumerated in (1) August 1985–
December 1986, (2) November 1995–February 1997
and (3) February 2001–February 2002; and subplots
alone in December 1998–March 1999. All trees
150
A.G. Van der Valk (ed.)
Fig. 1 Accumulated rainfall anomalies (ARA) with conditions
applied at Danum, 1985–2003: R30 of mean daily (MDR; black
line) and actual (ADR; red) rainfall, ARA365 (blue) and
accumulation only when R30 \ 232 mm (CARA232; dark
green); dashed reference at 100 mm (see text for explanations).
Smoothing algorithm used a negative exponential function with
sampling proportion equal to 0.02 (smoothing showed not all
of the individual shorter and milder drought events as defined
but was nevertheless preferred over the raw data for clarity).
Intervals between plot measurements are shown as periods P1
and P2, and sub-periods P2a and P2b
C10.0 cm stem gbh (girth at breast height) or
3.18 cm dbh (diameter), were measured for gbh
(above buttresses where these occurred), mapped and
identified at each plot census; deaths were recorded,
and recruits enumerated, in 1996 and 2001. In
subplots trees 10 to \50 cm gbh (small trees) were
remeasured, and deaths recorded: in 1999 limited
resources did not permit the measurement and
identification of recruits. In 1996–2001 a system of
identifying suitable trees for valid growth rate
estimates was introduced. Plots 1 and 2 had 17,942,
17,265 and 16,623 trees C10 stem gbh in 1986, 1996
and 2001, respectively, of correspondingly 450, 466
and 489 species. With 98% of trees identified to
named species, 2% stayed at unnamed but distinct
taxa. Voucher material is held at the Rijksherbarium
(Leiden) and the Sabah Forest Department Herbarium
(Sepilok, near Sandakan).
At the plot level all measured trees were considered.
The intervals 1986–1996 and 1996–2001 were designated periods P1 (10.0 years) and P2 (5.0 years),
respectively. At the subplot level only small trees were
considered at the four dates: recruits in 1996 and 2001
were omitted. Period P2 was subdivided into subperiods P2a (1996–1999, 2.6 years) and P2b (1999–
2001, 2.4 years; see Fig. 1). Further, for all trees three
size classes were defined: small, 10 to\50; medium, 50
to \100 and large C 100 cm gbh; and for small trees
four 10 cm gbh classes.
Species selection
As many species had too few individuals to permit
reliable analysis, those 34 with n C 100 trees, in both
plots together, in either 1986, 1996 or 2001 were
selected. This gave n C 60 valid trees per species for
Forest Ecology
growth estimates. Newbery et al. (1999) give the
basis for these sample sizes. These abundant species
represented on average 60% of all trees enumerated
in 1986–1996–2001. For the subplots, the 22 species
each with n C 50 small trees in 1986 were selected:
they represented on average 53% of the populations.
Dynamics rates
Annualized rates (% year-1) of mortality (ma) and
recruitment (ra) were based on the numbers of trees that
died (nd) or recruited (nr) relative to those at the start of
intervals (nstart) and mean time intervals (t) per species:
1 !
nd t
ma ¼ 1
100
1
nstart
1 !
nr t
1þ
1 100
ra ¼
nstart
were derived (Sheil et al. 1995; Alder 1996). Calculations of mortality and recruitment rates on the plot or
subplot level, or for different species and size classes,
used the mean time intervals of each individual group.
Regressors, trees that decreased in size to \ 10 cm gbh
but remained alive (Lingenfelder 2005), were not
considered in the present analysis. Confidence limits of
means (95%) of ma were found using an approximation
based on the F-distribution (Alder 1996; Nelson 1982).
Recruitment was based on the size of each period’s
starting population because the dynamics of the forest
was in a state of strong disequilibrium over the
15 years recorded. Using final population sizes (i.e.
the numbers of survivors), as have Phillips et al.
(1994), would have led to the estimate of ra being
confounded by the effect of the drought on ma.
Annualized mortality rate, ma, in period P1 was
adjusted to a 5.0-year basis (to equate period P2)
following the correction procedure of Newbery and
Lingenfelder (2004). The approach was based on the
earlier theoretical analysis of Sheil and May (1996).
Correction factors (i.e. the numbers by which the
original values of ma must be multiplied) for the rates
of all trees in the plots, and for their size classes
(small, medium and large) were correspondingly 1.04
(1.03, 1.08 and 1.10). In an analogous way, for small
trees in subplots, the correction factors for period P1
and sub-periods 2a and 2b were 1.11, 0.83 and 0.85,
respectively. These last values were also applied to
151
the 10-cm gbh size classes of small trees without
differentiation. The correction procedure assumes
that the major source of heterogeneity within tree
populations lies with species-specific differences in
ma. A similar correction procedure for ra is unknown.
Relative (rgr; mm m-1 year-1) stem growth rates
were calculated from gbh at the start and the end of
an interval for valid trees, based on time intervals of
each tree.
rgr ¼
ðlnðgbhend Þ
lnðgbhstart ÞÞ
t
103
For a growth rate to be valid between enumerations
the previous point of measurement on the stem should
not have moved (due a problem at the old point) or
been lost (e.g. due to tree breakage), and the stem
was in an optimal condition (i.e. no new buttress
growth; no cracks, splits or embedded lianas in, or
excrescences or termites on, the bark; absence of
deformations such as strong fluting or hollowness, a
pronounced oval cross-section or spiral form; and not
based on relascopic measurement—a few very large
trees). A bootstrapping procedure using N = 2,000
runs found the means and 95% confidence limits to
rgr (with GenStat 7/8, Payne 2000).
Drought response index
Percentage response to drought (RD) of a dynamic rate
variable, v, was calculated as the difference in v
between periods P1 and P2 (v1, v2) relative to the
weighted mean of the rates in these periods. The rate in
P1 received double the weight of that in P2 to reflect the
relative interval lengths of 10.0 and 5.0 years:
RD
1;2
¼ ððv2
v1 Þ 300Þ=ð2v1 þ v2 Þ
RD had a minimum of -150% when v2 was 0, and a
maximum of 300% when v1 was 0. The RD for ma was
multiplied by -1 so that decreases in ma indicated
positive responses, in a similar manner to increases in
ra and rgr. A new composite index cmp was
constructed using squares of loadings on the first
axis of a principal component analysis as linear
coefficients (ma -0.587, ra -0.451 and rgr -0.672):
RD
cmp
¼ 0:345 RD ma þ 0:203:RD
þ 0:452:RD rgr
ra
(correlation-based; k = {0.42, 0.32, 0.26}).
152
A.G. Van der Valk (ed.)
Percentage response to drought was found for P1–
P2a (RD_1,2a), P2a–P2b (RD_2a,2b), and P1–P2b
(RD_1,2b), using ma or rgr as variable v. Sub-periods
P2a and P2b were each taken to be c. 2.5 years in
duration so that the relative weights for P1 and either
P2a or P2b would be 4 to 1:
RD
1;2a
¼ ððv2a
v1 Þ 500Þ=ð4v1 þ v2a Þ
RD
1;2b
¼ ððv2b
v1 Þ 500Þ=ð4v1 þ v2b Þ
RD
2a;2b
¼ ððv2b
v2a Þ 200Þ=ðv2a þ v2b Þ
Of the 22 species, Reinwardtiodendron humile was
outlying because of its highly negative growth rates
which strongly biased the weighted mean across
species. In one case, ma for Lophopetalum becarrianum, both v1 and v2 were 0 and RD_2a,2b was also set to
0. With results for ra lacking, a composite index was
not calculated for the subplot-recorded species.
Randomization and multiple testing
To test whether species differed from one another
significantly, more than would be expected had their
dynamics variables been completely randomly distributed across trees of all species, a Monte Carlo approach
was taken. Randomization simply removed species
identity. For P1 and P2, deaths were re-assigned across
all trees at random, to the same extent as was recorded.
Samples equal in size to those of the 34 species’
populations (with a further all-other-species sample)
were randomly selected without replacement (FORTRAN77 program with NAG20 algorithms), and ma
and RD calculated. The procedure was repeated a
recommended N = 5,000 times (Manly 1997), and
exact probabilities found as twice the percentile of the
tail of the ranked values more extreme than that
observed. Values of ra were simulated in the same way,
but for rgr all values were re-allocated at random
across valid samples sizes. Individual randomized
values of cmp were found as for the recorded data, and
because they were weighted means of j = 3 variables it
was necessary to re-scale them by multiplying by Hj,
and to re-adjust means to their original values.
To derive an overall test for the whole community
individual species’ tests needed to be combined as a
‘family’. Family-wise error rate (FWER) tests of
significance were achieved with the sequentially
rejective procedure of Holm (1979) applying Sidák’s
adjustment to the Bonferroni a-level, and by finding
the Benjamini–Hochberg false discovery rate (FDR)
(Westfall and Young 1993; Benjamini and Hochberg
1995). The Holm procedure was applied to the values
of rgr, but not ma, of the 22 species of small trees in
the subplots. For both 34- and 22-species data sets,
the Bernoulli formula was applied to find the
minimum numbers of species required to reject the
null hypothesis of no family response at a = 0.05,
these being based on the lowest Bonferroni critical Pvalues allowed by the FDR for the four variables
separately (Moran 2003).
Sidak’s adjustment to the level of individual hypothesis rejection (a0 = 1 - (1 - a)1/k =
0.001508; a = 0.05, k = 34) was used because it
more powerful than that of Bonferroni (a0 = 0.05/
k = 0.001471) yet it maintains a strong family-wise
error rate (Westfall and Young 1993). The N = 5,000
randomizations allowed a lowest P-value of 0.0002 to
be detected, which is well below a0 in either case
above. As a consequence, ranking could result in ties
at this lowest level or simple multiples of it. In these
cases, whilst the family-wise P-value at each step was
the maximum of the previous and currently considered step, adjusted a0 -values were averaged across
ties (Appendix 3).
Stability analysis
Building upon the concepts of classical stability
thinking, a diagrammatic approach was developed to
highlight the different species’ modes of response.
Graphing RD1–2b against RD1–2a permitted an evaluation of each species’ trajectory. The four quadrants
(numbered clockwise) showed which species
remained positive (1) or negative (3), and which
switched from being positive to negative (2) or vice
versa (4), between P2a and P2b. Diagonal lines,
where D (= RD1–2b/RD1–2a) was either 1 or -1
represented no change between sub-periods, or a
change in the opposite direction of the same magnitude, respectively. Subdividing, octants defined
regions of stability and instability; numbered again
clockwise they represented four types of response
behaviour: destabilizing non-recovery (10 , 50 ; either
increasingly positive or negative; D [ 1), stabilizing
Forest Ecology
recovery (20 , 60 ; reduced positive and negative;
0 \ D [ 1), stabilizing oscillation (30 , 70 ; positive
switching to negative of less magnitude and the
converse; -1 \ D [ 0) and destabilizing oscillation
(40 , 80 ; positive switching to negative of greater
magnitude and the converse D \ -1).
Results
Antecedent rainfall history
Values of accumulated rainfall anomalies across 90,
180 (not shown—see Lingenfelder 2005) and 365
days (ARA365) ran roughly in parallel, with the 1 year
curve having the strongest amplitudes both in rainfall
deficit and surplus (Fig. 1). Accumulation of anomalies across 1 year is assumed to adequately reflect
the water conditions and to reveal the severity of
drought events by reflecting depletion or saturation of
soil water content quantitatively rather than simply
stating whether it was below or above a certain
threshold value. With this approach, immediate
strong rainfall deficits (R30 \ 100 mm) as well as
the ecologically more meaningful long-term (365day) deficits are being picked up.
153
Drought intensity can thus be expressed as the sum of
all daily rainfall anomalies (total DRA) within an
event, the deficit in rainfall derived from the
antecedent rainfall history (DEFARH).
Neglecting R30 [ 232 mm (CARA232), the 1997–
1999 event was severest with DEFARH =
-1,846 mm, followed by the 1990–1993 one with
DEFARH = -1,567 mm (Appendix 1). In conclusion, between July 1985 and December 2003 three
drought events were shown to be strong. Most severe
was the one centred on 1998, followed by those
centred on 1992 and 1987. The longest drought-free
period by far was between April 1999 and March
2002.
Spectral analysis
The power spectral density function (or ‘spectrum’)
was derived for several of the variables (Chatfield
2004; S-Plus 6 2001 version 7.0). Plots of log10 of
spectral value versus log10 of frequency have characterizing slopes ranging from *0 through -1 to
*-2, these being commonly referred to as white,
pink and brown noise, respectively (Steele 1985;
Vasseur and Yodzis 2004). Results: ADR, -0.161;
DRA, -0.153; R30, -1.91; ARA365, -1.96; CARA100,
-2.19; CARA232, -1.98.
Definition of events
Forest dynamics
Forest on soil that was already water-saturated would
not be able to make use of more rainfall, the excess
running off or draining away. Whilst accumulating
rainfall as DRA when R30 was \ 100 mm would have
been one possibility, the preferred solution took DRAvalues when rainfall was below MDR (the average R30
of MDR being 232.2 mm), i.e. when the forest
received less water than ‘expected’ (Fig. 1). Both
‘conditional accumulations’ (CARA100, CARA232)
highlighted the main droughts at Danum during the
period of climate records.
If a low precipitation event is taken to have
occurred when ARA365 was \ 0 (events were allowed
to be interrupted by up to 8 days without breaking
continuity), 19 such events occurred at Danum
between July 1985 and December 2003 (Appendix
1). Six events were less than 10 days in duration. The
longest-lasting event was that in 1990–1993, followed by 1997–1998, 1986–1988 and 1993–1994.
Overall response
Between periods P1 and P2, for all trees in plots,
annualized mortality (ma) and recruitment (ra), and
stem relative growth rate (rgr), increased by 45%
(interval-corrected 25%), 12% and 12%, respectively
(Table 1). Changes in the weighted means of the 34
most abundant species were very similar. Between
period P1 and sub-period P2a, for small trees in
subplots, ma increased by 41% (interval-corrected
6%) but rgr decreased by 38%; and between period
P1 and sub-period P2b the corresponding changes
were increases of 51% (16%) and 11%. Thus, whilst
ma increased during the drought (P2a) and continued
to rise slightly afterwards (P2b), rgr had a substantial
decrease followed by a larger overcompensating
increase. The weighted means of the 22 most
abundant species at this level showed a similar
response for ma, but a stronger one for rgr (Table 1).
154
A.G. Van der Valk (ed.)
Table 1 Mortality (ma), recruitment (ra) and stem relative
growth (rgr) rates for all (C10 cm gbh) and small (10 to
\50 cm gbh) in plots and subplots, respectively, over all
individuals of all species
ma
(% year-1)
ra
(% year-1)
rgr
(mm m-1 year-1)
1.24
1.39
11.14
12.48
10.85
All trees/plots
P1
P2
1.59 (1.87)
2.30 (2.34)
Small trees/subplots
P1
1.54 (1.70)
1.31
P2a
2.17 (1.81)
–
6.79
P2b
2.32 (1.97)
–
13.36
30–40 cm gbh class where ma increased across
periods and sub-periods (Fig. 2b). By contrast, for
all trees in plots, rgr decreased roughly in parallel
across size classes in periods P1 and P2 (Fig. 2c), but
within the smaller size classes of small trees in
subplots there was a strong change in relative
differences between periods and sub-periods. In the
lowest 10–20 cm gbh size class the decrease in rgr
from period P1 to sub-period 2a, and the increase
from sub-periods 2a to 2b, was much greater than in
the highest 40–50 cm gbh class. The RD1_2a and
RD2a_2b, respectively, increased and decreased in a
linear manner with gbh (Fig. 2d).
Mortality values in parenthesis are 5.0-year interval corrected
rates
Species dynamics
Size class analysis
Plot and period scales
For all trees in the main plots the difference in ma
between periods P1 and P2 increased with tree size,
and the RD1–2 became increasingly negative (Fig. 2a).
Across the smaller size classes, for small trees in the
subplots, trends were not apparent except in the
Between periods P1 and P2, the forest, as shown by
the most abundant 34 species, became more dynamic
(Table 2). Sample sizes are given in Appendix 2.
Thirty-one species increased, and three decreased in
ma between P1 and P2 (Table 2). Species differed
Fig. 2 Annualized
mortality (ma) (a, b) and
relative stem growth (rgr)
(c, d) rates for all trees
within three size classes
(see text for definitions) in
the main plots (a, c) for
periods P1 (open bars) and
P2 (grey bars), and for small
trees within four smaller
size classes in the subplots
(b, d) for period 1 (open
bars), and sub-periods P2a
(light grey bars) and P2b
(dark grey bars) at Danum.
Error bars indicate 95%
confidence limits. Weighted
percent changes (% wresp)
are shown by inverted
triangles: P1–P2 and
P1–P2a (open) and P2a–P2b
(closed)
Forest Ecology
155
Table 2 Mortality (ma; % year-1), recruitment (ra; % year-1) and relative (rgr; mm m-1 year-1) growth rates in periods P1 and P2
for the 34 most abundant species within plots at Danum
Code
Species
ra
ma
rgr
P1
P2
P1
P2
P1
P2
aj
Alangium javanicum Koord.
1.27
3.45
0.86
2.45
7.99
an
Antidesma neurocarpum Miq.
3.20
3.64
1.12
1.17
3.01
6.99
5.56
af
Aporosa falcifera Hook. f.
1.74
2.00
0.74
0.00
12.12
10.61
as
Ardisia sanguinolenta Blume
1.14
1.87
1.79
2.68
11.24
11.33
bt
Baccaurea tetrandra Müll. Arg.
1.06
1.46
0.40
1.06
8.10
7.59
bl
Barringtonia lanceolata (Ridl.) Payens
0.29
1.41
0.68
1.07
5.47
7.11
cs
Chisocheton sarawakanus Harms
1.16
1.48
0.87
0.52
12.06
11.88
cc
Cleistanthus contractus Airy Shaw
1.17
1.26
0.67
1.64
6.89
9.60
dr
Dacryodes rostrata (Blume) H. J. Lam
0.75
1.26
0.26
0.00
7.97
6.64
dm
Dimorphocalyx muricatus (Hook. f.) Airy Shaw
1.05
1.07
0.59
0.85
5.43
8.00
dc
fs
Dysoxylum cyrtobotryum Miq.
Fordia splendidissima (Blume ex Miq.) J. R. M. Buijsen
1.65
1.13
1.43
1.88
0.69
1.76
0.62
1.95
17.63
10.21
15.91
12.22
gk
Gonystylus keithii Airy Shaw
0.77
1.11
1.17
1.51
11.34
14.22
kl
Knema latericia Elmer
0.22
0.49
1.96
0.36
12.24
13.02
ln
Lithocarpus nieuwenhuisii (Seem.) A. Camus
1.01
1.24
0.47
0.68
15.30
17.87
lc
Litsea caulocarpa Merr.
2.45
3.88
2.56
1.88
16.26
20.36
lo
Litsea ochracea Boerl.
1.73
4.33
0.95
1.47
13.54
11.05
lb
Lophopetalum beccarianum Pierre & Ridl.
0.79
0.88
2.01
1.23
15.24
18.95
mk
Madhuca korthalsii H. J. Lam
0.58
1.22
1.11
1.06
10.65
12.21
mp
Mallotus penangensis Müll. Arg.
1.23
1.57
2.35
2.34
11.74
14.94
mw
Mallotus wrayi King ex Hook. f.
1.55
1.99
1.21
1.59
9.10
11.58
mc
Maschalocorymbus corymbosus (Blume) Bremek.
3.67
3.53
1.39
2.17
8.24
9.90
pm
Parashorea malaanonan Merr.
1.91
2.82
0.65
0.59
14.03
14.84
pl
Pentace laxiflora Merr.
1.28
3.45
1.08
0.45
21.03
20.54
pc
Polyalthia cauliflora Hook. f. & Thomson
1.29
1.42
0.56
1.22
5.01
6.58
pr
ps
Polyalthia rumphii Merr.
Polyalthia sumatrana King
0.96
1.14
1.02
1.19
0.82
1.04
0.98
0.88
11.95
15.50
14.03
17.15
10.20
px
Polyalthia xanthopetala Merr.
2.82
4.79
1.76
0.00
9.50
rh
Reinwardtiodendron humile (Hassk.) D. J. Mabberly
2.59
4.13
0.71
1.20
6.19
6.43
sf
Shorea fallax Meijer
2.35
3.25
2.52
1.84
17.97
15.75
sj
Shorea johorensis Foxworthy
4.01
5.12
1.32
1.98
38.35
29.71
sp
Shorea parvifolia Dyer
3.71
4.67
1.32
3.11
43.78
37.41
se
Syzygium elopurae (Ridl.) Merr. & L. M. Perry
1.42
1.77
0.51
0.00
5.66
6.17
st
Syzygium tawaense (Merr.) Masam.
1.37
2.44
1.00
0.00
13.41
15.39
Unweighted
1.60
2.31
1.15
1.19
12.77
13.28
Weighted*
1.55
2.11
1.22
1.38
11.08
12.34
Means
Families are found in Appendix 2(a)
* By the number of trees per species at the start of each period
considerably in annualized mortality rate (ma) with
ranges of 0.22–4.01 widening to 0.49–5.12% year-1
between P1 and P2 (Table 2). Weighted average ma
in P2 was 36% higher than P1. During both periods,
21 species had lower, and 13 species had higher, than
average ma. Across the 34 species, ma was
156
A.G. Van der Valk (ed.)
significantly correlated between periods (r = 0.850,
df = 32, P \ 0.001).
For annualized recruitment rate (ra) species also
ranged widely from 0.26–2.56 in P1 to 0.00–
3.11% year-1 in P2. Fourteen species decreased and
20 increased in ra between P1 and P2. Weighted
mean ra was only slightly (0.17% year-1) higher in
P2 than P1. The ra was also significantly correlated
between periods (r = 0.445, df = 32, P = 0.008).
Mean turnover increased from 1.37 to 1.75% year-1
between periods, and mean difference between ma
and ra widened from 0.46 to 1.12% year-1.
Twenty-three species had higher relative growth
rate (rgr) in P2 than P1, and eleven lower rgr.
Weighted mean rgr increased by 11%, although
ranges in rgr contracted slightly from 3.0–43.8 to
5.6–37.4 mm m-1 year-1 (Table 2). Across species,
ln(rgr) was strongly positively correlated between
periods (r = 0.943, df = 32, P \ 0.001). The ma, ra
and rgr were only weakly to marginally significantly
positively inter-correlated, however (r = 0.214 to
0.348, df = 32, P = 0.043 to 0.225). The mean ma
of the 34 species was lower than that for all trees in
the plots, whilst for ra and rgr the means were closer
(Table 1).
Subplot and sub-period scales
Within the subplots, of the 22 most common species
(Table 3; sample sizes in Appendix 2), individual
species’ mortality rates (ma), which had a range of
0.55 to 3.86% year-1 in P1, became more variable in
Table 3 Annualized mortality (ma; % year-1) and relative growth rates (rgr, mm m-1 year-1) in period P1 and sub-periods P2a
and P2b for the 22 most abundant species within subplots at Danum
Code
Species
rgr
ma
P1
P2a
P2b
P1
P2a
P2b
af
as
Aporosa falcifera
Ardisia sanguinolenta
0.85
1.28
1.17
1.45
1.35
2.59
12.80
12.35
5.42
4.88
9.52
13.59
bt
Baccaurea tetrandra
1.27
1.12
2.07
8.37
3.92
12.49
cc
Cleistanthus contractus
1.24
1.23
1.30
7.81
4.03
14.93
dr
Dacryodes rostrata
0.72
3.54
1.74
7.49
6.99
9.06
dm
Dimorphocalyx muricatus
0.95
1.78
0.37
5.26
3.72
16.54
dc
Dysoxylum cyrtobotryum
2.10
2.90
1.16
20.28
12.56
18.76
fs
Fordia splendidissima
0.99
2.31
3.09
9.52
5.99
12.22
lc
Litsea caulocarpa
2.32
7.39
3.04
18.20
11.58
18.23
lo
Litsea ochracea
1.43
1.58
5.83
16.98
8.08
10.19
lb
Lophopetalum beccarianum
0.57
0.00
0.00
15.11
14.32
24.39
mk
Madhuca korthalsii
0.55
0.72
1.70
10.35
7.71
12.44
mp
Mallotus penangensis
1.47
1.41
0.00
12.10
8.35
13.87
mw
Mallotus wrayi
1.49
2.00
1.99
9.08
5.73
14.83
mc
Maschalocorymbus corymbosus
3.86
3.44
8.28
8.44
4.45
11.80
pl
pc
Pentace laxiflora
Polyalthia cauliflora
2.31
1.28
4.14
1.41
2.72
2.03
16.03
4.77
9.03
2.63
16.41
7.37
pr
Polyalthia rumphii
1.00
0.00
3.59
8.60
6.81
17.91
ps
Polyalthia sumatrana
1.06
2.70
2.31
17.54
15.18
23.57
px
Polyalthia xanthopetala
3.06
9.96
4.01
11.17
9.01
12.50
rh
Reinwardtiodendron humile
3.18
4.06
4.45
4.97
-2.46
7.84
sf
Shorea fallax
2.10
4.57
4.76
17.51
6.80
8.67
Unweighted
1.59
2.68
2.65
11.60
7.03
13.96
Weighted*
1.52
2.26
2.26
10.11
6.08
14.11
Means
* As Table 2
Forest Ecology
subperiods P2a and P2b where two species each had
zero mortality, and maximum values were 10.0 and
8.3% year-1, respectively (Table 3). Sixteen of the
22 species showed increases in ma in P2a compared to
P1. Six species decreased in ma. Between P2a and
P2b, 12 species increased and nine species decreased
in ma; and one species remained at zero mortality.
Weighted mean ma increased between P1 and P2a by
50% but then did not change in P2b. For rgr all
species decreased between P1 and P2a, and all
increased between P2a and P2b (Table 3).
Hypothesis testing of species’ responses
Each set of 34 species’ P-values from the randomization tests of RD_1,2, for ma, ra, rgr and cmp, were
ranked from smallest to largest separately. The
condition of a one-step procedure for multiple
(simultaneous) testing that P-values be uniformly
distributed was tested with the v2-statistic (df = 4) on
frequencies in five bins of 0.2 between 0 and 1
(expected frequency in each = 6.8). The observed
distributions were significantly non-uniform for ra
(v2 = 37.47, P \\0.0001) and cmp (v2 = 21.59,
P \ 0.001), but not for ma (v2 = 4.53, P [ 0.3) and
rgr (v2 = 5.41, P [ 0.2). Holm’s sequential stepdown procedure (Holm 1979) was used accordingly.
There existed a very low degree of logical
interdependence between values when randomizing
(without replacement) across species, i.e. one dead
tree or recruit allocated to any one of the 34 species
cannot be allocated to any of the others, or a high or
low rgr value allocated to one species cannot be
given to another species. The means (±SE) of the 528
pair-wise Pearson correlation coefficients (N = 5,000
simulated values) for each of the four variables, and
the percentage variance accounted for by principal
components analysis (based on r) on the first three
axes, showed that the correlations involved were in
fact negligible (Appendix 3).
For the 22 species of small trees in the subplots
testing of responses for ma was restricted. In an
appreciable percentage of the randomization runs
double-zero cases for the numbers of dead trees
meant that RD could not be found, and substitution of
so many results with RD = 0 would have been
unsatisfactory. Furthermore, many species (15/22
with zero ra -values c. 50–800 times out of 5,000)
had 95%, and often 99%, upper and lower confidence
157
limits at the maximum or minimum of RD possible,
and so rejection of the individual-species’ null
hypotheses was largely impossible.
Species’ response indices
Plot and period scales
Most species had increased RD-values for ma in P2
compared to P1, about equal numbers had increases
and decreases in ra, two-thirds increases in stem rgr,
and two-thirds decreases in cmp (Fig. 3). The
randomization tests followed by family-wise error
rate adjustment highlighted one significant
(P \ 0.01) negative case for ma (Barringtonia lanceolata) and one positive case for rgr
(Dimorphocalyx muricatus) (Fig. 3a, c). For ra, five
species (Aporosa falcifera, Dacryodes rostrata, Polyalthia xanthopetala,
Syzygium elopurae
and
S. tawaense) showed significantly (P \ 0.01) negative responses, one other (Knema latericia) less so
(P \ 0.05); two species (Cleistanthus contractus and
Baccaurea tetrandra) responded significantly and
positively (P \ 0.05) (Fig. 3b). Of 34 species, only
one (D. muricatus) showed a significant cmp
response (positive, P \ 0.01; Fig. 3d).
Allowing for a B 5% false discovery rate (FDR),
the analysis revealed just three further significant
(P \ 0.05) cases for ra (negative—Pentace laxiflora,
positive—Polyalthia cauliflora and Shorea parvifolia), and for rgr (P \ 0.01) one more (positive—
Mallotus wrayi) (Fig. 3b, c). The rate of increase in
RD per ranked-species (linear regression line) was
highest for ra (7.57%), less for ma (3.59%) and lowest
for rgr (1.90%), with cmp between ma and rgr
(2.47%). The RD1_2 of ma, ra and rgr were not
significantly correlated with one another (r = 0.047
to 0.132, df = 32, P [ 0.25). Mean ma of the 34
species was lower than that of all trees in the plots,
whilst for ra and rgr it was close (Fig. 3). Note that
among the positively responding species one, aj, in
Fig. 3b was not significant because its sample size
was the smallest of all species (n = 101), compared
with the four significant species (n = 206 - 324):
see Appendix 2 for full range of sample sizes.
Applying the Bernoulli formula, a minimum of
k = 2, 5, 3 and 2 (out of 34) individually significant
results would have been needed to meet a familywise level of a = 0.05. On this basis ra qualified with
158
A.G. Van der Valk (ed.)
eight strong and three protected cases, but not ma, rgr
and cmp.
balanced decreases (Fig. 4a–c). Despite the wide
range in responses none could be shown to be
significant (P B 0.05). All species had lower rgr in
P2a than P1, all higher rgr in P2b than P2a which led
to three-quarters with higher rgr in P2b than P1
(Fig. 4d–f). Randomization tests showed just one
species with a significantly (P B 0.01) reduced rgr
between P1 and P2a (R. humile), two with similarly
significant increases between P2a and P2b (R. humile, D. muricatus) and P1 and P2b (D. muricatus,
Mallotus wrayi), and one decrease (P B 0.05) in the
last (Shorea fallax). Allowing for the FDR level led
to no further significant cases.
Rates of increase in RD with ranked species
(outlier R. humile excluded) were similar for
RD1–2a, RD2a–2b and RD1–2b using ma (11.8, 12.9 and
12.7 respectively), but increased for rgr (3.10, 4.97
and 6.61). Again, RD for ma and rgr were not
significantly correlated within each of the (sub-)
periods (r = -0.103 to 0.284, df = 20, P C 0.20).
RD1–2a and RD2a–2b were significantly negatively
correlated for ma (r = -0.497, P = 0.019) and rgr
(r = -0.854, P \ 0.001); and conversely RD1–2b and
RD2a–2b significantly positively so for ma (r = 0.735,
P \ 0.001) and rgr (r = 0.428, P = 0.047): RD1–2a
and RD1–2b being insignificantly correlated for ma and
rgr (P [ 0.45). Mean ma of the 22 species was very
close to that for all trees in the plots, although for rgr
it differed slightly (Fig. 4). Applying the Bernoulli
formula again, a minimum of k = 2, 2 and 3 out of 22
individually significant results were needed to qualify
for family-level significance. This requirement was
met for rgr in sub-periods 2a–2b and period 1—subperiod 2b (Appendix 3).
Since ma remained insignificant at the family level
when the less conservative FDR procedure was
applied to the 34 species, and the relationship
between ma and size (gbh) within the small trees
was weak, it may be reasonably inferred that
differences at the 22 species level would be insignificant too. Mortality was also likely to have shown
much less response that rgr when moving from a 5.0year to a 2.5-year period.
Subplot and sub-period scales
Between-scales correlation
Within subplots three-quarters of the 22 species
showed an increase in RD-values for ma in P2a and in
P2b compared to P1: between P2a and P2b increases
The 16 subplots were nested stratified random subsamples of the plots. Selecting the dynamics variables
for the same 22 species studied at the subplot level
Fig. 3 Weighted percent changes between periods P1 (1986–
1996) and P2 (1996–1999) in a mortality (ma, inverted scale), b
recruitment (ra), c relative growth rate (rgr) and d a composite
index (cmp) of the three variables, ranked for the 34 most
common species in the main plots at Danum: solid line,
weighted mean of the 34 species; dashed line, the overall
values for all trees in plots. Codes for species are explained in
Table 2. Species significance, determined by randomization
tests and family-wise adjusted probability levels, is signified by
number of diamonds over/under bars: two, P B 0.01; one,
P B 0.05; and none, ns or P [ 0.05. Crosses above bars
indicate species additionally significant (P B 0.05) after
controlling for false discovery rate
Forest Ecology
159
Fig. 4 Weighted percent
changes in a–c mortality
rate (ma, inverted scale) and
d–f relative growth rate
(rgr) between period P1
(1986–1996) and subperiods P2a (1996–1999)
and P2b (1999–2001), as
(a and d) P1–P2a, (b and e)
P2a–P2b and (c and f)
P1–P2b; ranked for the 22
most common species in the
subplots at Danum. Lines,
significance levels and
species codes as in Fig. 3
from the 34 used at the plot level, the trends across
species were in good agreement. Mortality rate and
rgr for P1 were each strongly correlated between the
subplot and plot levels (r = 0.912 and 0.919 respectively), and correspondingly so were mean rates of
P2a and P2b at the subplot level and those for P2 at
the plot level (r = 0.877 and 0.782). The correlation
between RD1–2 and the average of RD1–2a and RD1–2b
for the 22 of 34 species at the subplot level were
significant for ma (r = 0.468, df = 20, P = 0.028)
and for rgr (r = 0.646, P \ 0.001) indicating that the
subplot sampling was a good representation of the
plot for the species’ responses also. Correlations
between rgr in period P1, and RD1–2 for rgr, for 21
species in the subplots (rh omitted again) with the
subsequent ma in sub-periods P2a and P2b, and
RD1–2a, RD1–2b and RD2a–2b, were all insignificant
(r = -0.332 to 0.224, df = 19, P = 0.13 to 0.94).
Stability analysis
Analysis for ma, showed that octants 20 and 30 were
unoccupied, most species fell in 40 –60 , and a few in 10 ,
70 and 80 (Fig. 5a)—a wide range of trajectory
dynamics. For rgr, 10 –50 were empty, most species
were in 70 and 80 and a few in 60 (Fig. 5b)—a largely
stabilizing response. Octants 10 up to 40 may be
described, respectively, as over-enhanced, enhanced,
not enhanced and under-enhanced, whilst 80 down to
50 as over-recovered, recovered, not recovered and
under-recovered. Thus, in terms of ma most species
either did not recover or they under-recovered, whilst
in terms of rgr most species recovered or overrecovered. The species that ‘benefited’ from, or were
promoted by, the drought were in 10 and 80 , while
those that suffered or were disadvantaged were in 40
and 50 . Four species (if Dimorphocalyx muricatus
very close to the line is allowed) were thus promoted
in terms of ma and eight (including Polyalthia cauliflora on the line) in terms of rgr (Fig. 5). One
species, Cleistanthus contractus, had zero change in
ma and therefore appeared resistant.
Post-drought response in growth was strongest for
understorey species. Across the 16 species which
were [ 0.75 composed of very small (10 to \30 cm
gbh) stems, RD2a_2b and RD1_2b for rgr increased
160
A.G. Van der Valk (ed.)
Fig. 6 Weighted percent changes (RD) in relative stem
growth rate (rgr) between period P1 and sub-period P2a (open
circles), sub-periods P2a and P2b (closed circles) and period P1
and sub-period P2b (open triangles) regressed against the
proportion of very small trees, pvs (10–30 cm gbh) for those 16
species in the subplots at Danum which had C 0.75 of their
small-tree stems in this size class. RD1–2a = -1.5 - 42.5 pvs;
RD2a–2b = -216 ? 325 pvs; and RD1–2b = -265 ? 342 pvs
(see text for statistics). Extrapolated RD1–2a and RD1–2b lines
cross at pvs = 0.685. Codes for species (upper line) are
explained in Table 2; codes for other lines may be found by
down projection
Fig. 5 Weighted percent changes of a mortality rate (ma,
inverted scale) and b relative growth rate (rgr), between period
P1 (1986–1996) and sub-period P2a (1996–1999) plotted
against the same between period P1 and sub-period P2b
(1999–2001) for the 22 most common species in the subplots at
Danum. Codes for species are explained in Table 2; rh has an
x-axis value of -213. Octant numbers 10 –80 are shown without
primes
significantly with this proportion (F = 15.9 and 5.22,
df = 1, 14; P = 0.001 and 0.038; resp.; Fig. 6)
although for RD1_2a the relationship was not significant (F = 0.38, P = 0.55).
Discussion
Resilience of the tree community
During the 1997/1998 ENSO the rainfall in April
reached an exceptionally low value. Based on the
concept of antecedent rainfall history, the conditional
accumulated rainfall deficit also then fell to its lowest
level during the period of recording, highlighting the
importance of the 1997/1998 event (Fig. 1). The
1982/1983 event (Walsh 1996a, b; Walsh and Newbery 1999; Newbery and Lingenfelder 2004) may
have affected Danum to a similar extent and less
extreme lows (relevant to the period of forest
measurements here) were found in 1987 and 1992.
Drought evidently perturbed, but did not disturb,
the forest at Danum: its effect on tree mortality was
quite small overall. The forest community, nevertheless, showed a high resilience to the 1997/1998 event,
with immediate negative rgr responses followed by
recovery. This suggests that interspecific variation in
drought responses may be driving community
dynamics. Recruitment of about a third of the
common species decreased or increased significantly
within 5 years: mortality, however, appeared to be
much less species-specific. Using family-wise statistical procedures, the forest community overall was
shown to have been significantly affected. The
‘family’ is the community in the present context.
The interpretation is that stochastic environmental
variation in accumulated rainfall deficit primarily
controls tree growth rate (through a limitation in
water supply); and recruitment is growth-dependent
Forest Ecology
because saplings advance into the smallest size class
enumerated. Species differ in susceptibility to
increasing deficit because of their differences in
morphology and physiology (Gibbons and Newbery
2002). In this way the dynamics of the forest and its
tree community interactions are largely reducible to,
and understandable in terms of, the plant-environment physical processes in operation.
Drought is a complex factor and the physical
variables behind it can have a distinctive signal.
Spectral analysis showed that raw rainfall values
from Danum displayed white noise (low and high
frequencies evenly distributed) yet the various accumulated drought indices all indicated brown noise
with a higher proportion of contributions from low
frequencies. On this basis realizations of rainfall
depletion are stochastic in nature. Only long-term
measurements that capture forest dynamics before
and after an event for several years are therefore
likely to provide sufficient relevant information.
Because the driving water deficit variable is
stochastic, none of the species can ‘tune in’ to
regular cycles and each is continually experiencing
immediate and lagged effects of the perturbations.
Neighbourhood competitive interactions would also
be expected to be undergoing continual change. This
leads to complex yet predictable dynamics in the
short-to-medium term (10–50 years), which—if stationarity holds—might be expected to average out
towards quasi-constant structural and species composition in the longer term (50–200 years).
The conditional accumulated rainfall anomaly
variable represents what the tree is likely to experience over time. Arguably, relatively high as opposed
to low antecedent rainfall will, respectively, buffer, or
make more susceptible, the forest over those months
preceding a period of strong rainfall deficit. Such a
formulation has not hitherto been made for trees,
let alone tropical ones. This is illustrated by the
inferred weak effect in 1991/1992 versus strong
effect of drought in 1997/1998. Better growing
conditions presumably lead to more stored water
which allows trees to ameliorate the drought effect.
Perturbation response niche
Species-specific responses were very different which
leads here to the idea of a perturbation response
niche. Different species having greater-than-average
161
restrictions in growth and reduced recruitment in
drought periods would be expected to compensate in
the wetter inter-event periods; those enhanced by, or
over-compensating to, the drought are likely at
competitive disadvantage in the wetter periods (currently under test). Further, differentiation of the niche
would be emphasized by the effect of topography
acting through soil water relations, i.e. ridges being
drier than lower slopes. The concept is somewhat
akin to the regeneration niche of Grubb (1977) where
in the former case time is more important, whilst in
the latter it is space.
The new results accord well with the spatial
distributions of the common understorey species
across the two plots with respect to topography.
Dimorphocalyx muricatus, Cleistanthus contractus
and Lophopetalum beccarrianum, in decreasing order
of importance, cluster and are associated on drier ridge
locations at Danum (Newbery et al. 1996). The first
showed strong positive rgr and cmp responses to
drought perturbation (Figs. 3, 4), the second positive
ra, rgr and cmp, and the third reduced mortality
responses—as did Polyalthia rumphii and Mallotus penangensis (Fig. 3). Additionally, Polyalthia
cauliflora showed a positive increase in ra (Fig. 2).
Those mentioned are the six species in the outer sector
or octant 80 in Fig. 5b, and it underlines the likely
central driving role of rgr in species’ dynamics in
response to drought. Three addenda are: (1) for the
seven species with negative ra we have no explanation
at present; (2) Mallotus wrayi, the ubiquitous numerically dominant species, responded positively to
drought in terms of rgr; (3) Reinwardtiodendron humile is a potential drought phytometer. Importantly, all
those species just discussed are understorey taxa
(Newbery et al. 1992, 1996, 1999a, b).
No correlation at the species level between rgr in
period P1 and ma in subperiods P2a and P2b was
detectable, suggesting that faster or slower growing
species were neither more nor less affected by the
drought. Species appeared to respond over time in a
highly idiosyncratic manner, each species with its
own trajectory.
That larger trees died more often than smaller ones
under drought meant that a moderate degree of
random canopy opening followed the defoliation
observed and evidenced by a large increase in small
end branch abscission (Walsh and Newbery 1999).
The drought-adapted small-stemmed understorey
162
species (many in the Euphorbiaceae) were able to
benefit from the temporarily increased light levels.
This in part corroborates the understorey facilitation
hypothesis (Newbery et al. 1999; Newbery and
Lingenfelder 2004), which proposes that droughttolerant or drought-avoiding understorey species in
some Bornean rain forests nurse saplings and small
trees of the drought-sensitive canopy species (particularly of the canopy-forming Dipterocarpaceae),
through the crucial dry periods. The saplings of the
canopy species are thought to be protected from the
direct light and drier conditions caused by canopy
opening.
Complex forest dynamics driven by perturbations
The most remarkable general result from this study is
that the common species at least show highly specific
and different dynamics from one another. Furthermore,
different species responded differently to drought, to
varying degrees, sooner or later after the event, and
with more or less extent of recovery. The oscillating
dynamics of several species is what might be expected
of a system that is moderately perturbed and returning
to an equilibrium (Botkin and Sobel 1975; DeAngelis
and Waterhouse 1987; Ives 1995). The impression of
all these changing patterns of dynamics and their
interactions can be likened to a kaleidoscope. The
challenge is to find the attractor which bounds the
system: this may be the topographic gradient.
The results do not sit well though with the
suppositions of the recently debated neutral theory
of biodiversity for tropical forests (Hubbell 2001,
2005, 2006). In that thesis equivalence of species and
individuals, and random mortality, lead to weakly
diffuse competitive interactions and a slow nonadaptive drift in species composition over time. That
species could be so ecologically and evolutionary
similar has been often challenged (Chave 2004; Bell
2005; Purves and Pacala 2005; Bell et al. 2006): the
theory says any differences that do occur are of little
consequence. Under the neutral theory the patterns of
species’ responses and dynamics recorded at Danum
would presumably be labelled as ‘random’; which
seems highly implausible to us given the supporting
ecological information on forest structure and tree
physiology (Gibbons 1998; Gibbons and Newbery
2002), and the topographic gradient effect on species
patterning (Newbery et al. 1996).
A.G. Van der Valk (ed.)
There is then an underlying structure to the forest
community at Danum, which is determined to a large
degree by the species-specific dynamic responses to
perturbation. This tends to refute neutrality and
species equivalence, rather the forest at Danum
functions on the basis of plurality of species’
responses. A further serious and over-looked problem
with the neutral theory is that it assumes a constant
environment. This too seems not to be the case at
Danum, and is arguably unlikely to be realistically so
anywhere, including the tropics.
A possible alternative to the descriptive neutral
theory (which in any case is very difficult to test
directly, if at all) is to take a dynamic ecosystems
approach (Shugart 1998) in which testable mechanisms and processes may allow understanding of
structural equilibria (or dis-equilibria) in multi-species population dynamics with reference to a
measured stochastically varying environment (May
1974; Ives and Carpenter 2007). It has been shown
theoretically that a stochastic environment can result
in community stability (Chesson 1982; Chesson and
Huntly 1997), and possibly a plurality of responses
might play an important role. One caveat to the
present work, however, is that only one ENSO event
was followed. Predicting and testing for similar
patterns of response after future events will be
valuable.
This new postulate does not attempt to explain
species diversity per se but aims primarily at quantifying, and finding the limits to, complex forest
community dynamics. Rare species, for which data
will be always insufficient to make reliable estimates
of dynamics parameters, could be treated as being
neutral (i.e. as indeterminate), whilst the common
species, for which estimates can be made reliably a
non-neutral way (determinate), could be followed
over time using physical, physiological and statistical
models.
Apart from large historical disturbances, stochastic droughts perturb the forest at Danum on a short
time scale and the forest appears to accommodate
them by being resilient. This may work up to a
certain threshold of frequency and intensity, one to
which the main constituent species are avoidanceor tolerance-adapted. But if, as a result of prognosticated climatic change, droughts were to
increase, then higher tree mortality rates and longer
periods of restricted growth would be expected to
Forest Ecology
163
ensue, the latter inevitably lowering critically the
recruitment of many species to the extent that they
may not recover in the shortened inter-drought
periods; and hence an end-effect change in forest
structure and species composition. With the continuing long-term observations at Danum, it might
soon be possible to model different scenarios with
statistical confidence, and on that basis take the
necessary measures to conserve the lowland tropical rain forests of Borneo in its original and
natural form of a mosaic of perturbed and, it
seems, resilient ecosystems.
Acknowledgements We are grateful to the Danum Valley
Management Committee and the Economic Planning Unit,
Prime Minister’s Office, Malaysia, for permission to undertake
this research; I. and S. Samat, J. Hanapi and N. Majid for recent
field assistance; R. C. Ong (Sabah Forest Department) and G.
Reynolds (Royal Society S.E. Asia Rain Forest Research
Programme) for facilitating the work locally; E. J. F. Campbell,
A. Hämmerli, D. N. Kennedy, G. H. Petol and M. J. Still of
the 1986–1999 enumeration teams; C. E. Ridsdale
(Rijksherbarium, Leiden) and L. Madani (SFD Herbarium,
Sandakan) for tree identifications, especially the 2001 recruits;
and R. P. D. Walsh for access to the Danum climate records.
The research was funded by the Swiss National Science
Foundation (grant nr 31–59088). This paper is a contribution to
the Royal Society S. E. Asian Rain Forest Programme.
Appendix 1
Climate
The low precipitation
events at Danum
1985–2003
a
When ARA365 \ 0
b
(Total DRA) when
ARA365 \ 0 and
R30 \ 232 mm
c
ARA365 \ 0 but
R30 [ 232 mm across all
6 days
Eventa
Start
End
Duration (d)
DEFARH (mm)b
1
8/30/86
5/1/88
610
-905.1
2
11/7/88
12/7/88
31
-38.1
3
2/13/89
2/20/89
8
-7.5
4
3/29/89
5/13/89
44
48.9
5
6/1/89
6/6/89
6
6
10/23/90
3/23/93
883
-1,566.9
7
6/18/93
6/3/94
351
-357.3
8
9
6/15/94
9/15/94
6/23/94
9/23/94
9
5
-65.8
-17.8
10
10/25/94
11/1/94
6
-28.5
11
4/2/95
5/27/95
56
-8.6
12
6/29/95
8/13/95
44
-91.8
13
1/18/97
4/15/99
818
-1,846.0
14
3/18/02
6/26/02
101
-25.4
15
7/10/02
9/2/02
54
21.7
16
11/1/02
11/25/02
23
-73.9
17
12/4/02
3/26/03
112
-126.4
18
6/27/03
6/29/03
3
9.8
19
9/3/03
9/26/03
24
17.0
n.a.c
164
A.G. Van der Valk (ed.)
Appendix 2
Trees
Appendix 2(a) Sample sizes at the start of the periods P1 and
P2 (n86, n96) and corresponding numbers of valid trees (nvP1,
nvP2) for the calculation of annualized mortality (ma;
Species
Family*
% year-1) and recruitment (ra; % year-1) rates, and relative
(rgr; mm m-1 year-1) growth rates, in periods P1 and P2 for
the 34 most abundant species (and their families) at Danum
ma and ra
n86
rgr
n96
nvP1
Alangium javanicum
Alan
101
91
Antidesma neurocarpum
Euph
119
100
77
70
Aporosa falcifera
Euph
261
238
157
143
Ardisia sanguinolenta
Myrs
568
591
430
444
Baccaurea tetrandra
Euph
250
233
189
168
Barringtonia lanceolata
Lecy
141
147
129
120
Chisocheton sarawakanus
Cleistanthus contractus
Meli
Euph
155
289
150
273
116
223
105
212
Dacryodes rostrata
Burs
153
145
130
118
Dimorphocalyx muricatus
Euph
840
801
667
645
Dysoxylum cyrtobotryum
Meli
170
155
129
122
Fordia splendidissima
Legu
520
543
394
414
Gonystylus keithii
Thym
121
126
104
101
Knema latericia
Myri
141
166
128
140
Lithocarpus nieuwenhuisii
Faga
125
115
94
70
Litsea caulocarpa
Laur
322
319
197
215
Litsea ochracea
Laur
163
147
115
95
Lophopetalum beccarianum
Cela
234
267
200
221
Madhuca korthalsii
Sapo
508
532
433
429
Mallotus penangensis
Euph
204
233
172
196
Mallotus wrayi
Euph
2,268
2,207
1,781
1,723
Maschalocorymbus corymbosus
Parashorea malaanonan
Rubi
Dipt
403
149
335
133
245
111
243
93
Pentace laxiflora
Tili
240
214
163
145
Polyalthia cauliflora
Anno
324
302
271
258
Polyalthia rumphii
Anno
141
138
119
119
Polyalthia sumatrana
Anno
222
221
192
186
Polyalthia xanthopetala
Anno
241
223
172
156
Reinwardtiodendron humile
Meli
262
221
166
140
Shorea fallax
Dipt
371
395
264
298
Shorea johorensis
Dipt
197
157
82
72
Shorea parvifolia
Dipt
206
170
124
104
Syzygium elopurae
Myrt
134
120
100
97
Syzygium tawaense
Myrt
Totals
69
nvP2
60
124
120
85
74
10,667
10,328
8,028
7,796
* Family abbreviations; Alan, Alangaceae; Anno, Annonaceae; Burs, Burseraceae; Cela, Celastraceae; Dipt, Dipterocarpaceae; Euph,
Euphorbaceae; Faga, Fagaceae; Laur, Lauraceae; Lecy, Lecythidaceae; Legu, Leguminosae; Meli, Meliaceae; Myrs, Myrsinaceae;
Myrt, Myrtaceae; Rubi, Rubiaceae; Sapo, Sapotaceae; Thym, Thymelaceae; Tili, Tiliaceae
Forest Ecology
Appendix 2(b) Sample sizes
at the starts of period P1
(n86) and sub-periods P2a
and P2b (n96, n99) and
corresponding numbers of
valid trees (nvP1, nvP2a,
nvP2b) for the calculation of
annualized mortality rates
(ma; % year-1), and relative
growth rates (rgr,
mm m-1 year-1), for the 22
most abundant species
within subplots at Danum
165
Species
rgr
ma
n86
n96
n99
nvP1
nvP2a
nvP2b
74
65
62
57
54
57
Ardisia sanguinolenta
166
138
130
125
114
109
Baccaurea tetrandra
Cleistanthus contractus
76
118
66
103
62
97
57
85
52
75
55
86
Aporosa falcifera
Dacryodes rostrata
Dimorphocalyx muricatus
58
54
48
51
45
40
276
250
236
227
209
213
53
41
36
39
32
32
Fordia splendidissima
157
134
122
119
105
101
Litsea caulocarpa
105
72
56
66
52
46
60
49
45
47
42
35
56
Dysoxylum cyrtobotryum
Litsea ochracea
71
66
63
64
60
112
103
98
97
89
86
57
48
45
48
44
44
Mallotus wrayi
713
612
573
569
517
493
Maschalocorymbus corymbosus
120
80
71
65
58
52
26
Lophopetalum beccarianum
Madhuca korthalsii
Mallotus penangensis
58
36
31
34
27
123
108
104
104
97
94
Polyalthia rumphii
53
48
47
46
45
41
Polyalthia sumatrana
Polyalthia xanthopetala
50
59
43
42
36
32
45
41
35
30
31
26
Reinwardtiodendron humile
76
55
48
49
42
40
Shorea fallax
85
64
55
62
51
43
2,720
2,277
2,097
2,097
1,875
1,806
Pentace laxiflora
Polyalthia cauliflora
Totals
Appendix 3(b) Species which had significant differences in
their dynamics variables from random expectation adjusted for
multiple hypothesis testing: 34 species in plots
Variable
Species
codes#
Holm (Sidak)
adjusted P
Family-wise P
ma
bl
0.001508
0.0068
ra
af, dr, px,se, st
0.001605
0.0068
kl
0.001767
0.0116
bt,cc
0.001804
0.0220
rgr
dm
0.001508
0.0068
cmp
dm
0.001508
0.0068
Appendix 3
Tests
Appendix 3(a) Means (±SE) of 34-spp pair-wise correlations
(n = 528) for each of four dynamics variables and % variance
accounted for by first three axes of corresponding principal
components analyses
ma
Coefficient r
1*
2
3
-0.014358 ± 0.000732
3.48
3.41
3.35
ra
-0.014434 ± 0.000738
3.49
3.43
3.41
rgr
-0.014838 ± 0.000778
3.49
3.39
3.35
cmp
-0.014510 ± 0.000777
3.55
3.42
3.37
* % Var. = 100/34 = 2.94 had all axes been equal
# species codes are those of Table 2 in the main text
The Benjamini-Hochberg step-up FDR procedure gave the
same results as the Holm step-down one for ma and cmp; but
for ra three further species were significant: pc, pl and sp
(adjusted P = 0.0132, 0.0147 and 0.0162 resp.), and for rgr
there was one further case: mw (adjusted P = 0.0029). Note
that mw was ranked 4th highest yet was significant (unlike pr
and cs) due to its very much larger population size (maximum
in Appendix 2)
166
A.G. Van der Valk (ed.)
Appendix 3(c) Species which had significant differences in
their rgr from random expectation adjusted for multiple
hypothesis testing: 22 species in subplots
Period
Species
codes#
Holm
(Sidak) P
Family-wise
P
P1-P2a
rh
0.002329
0.0044
P2a-P2b
P1-P2b
rh
0.002329
0.0044
dm
0.002440
0.0042
dm
0.002329
0.0044
mw
0.002440
0.0084
sf
0.002500
0.0400
# Species codes are those of Table 2 in main text
The FDR procedure resulted in the same results, i.e. no
additionally significant species
Appendix 3(d) Bonferroni minimum P-critical values from the
step-down FDR procedure which are used in the Bernoulli formula
1.
2.
P1 - P2: ma, 0.0002; ra, 0.0152; rgr, 0.0024; cmp, 0.0024
Rgr: P1 - P2a, 0.0002; P2a - P2b, 0.0002; P1 - P2b,
0.0020
References
Alder D (1996) Growth modelling for mixed tropical forests.
Forestry Papers No. 30. (Oxford Forestry Institute,
Oxford. Beaman RS, Beaman JH, Marsh CW, Woods PV
(1985) Drought and forest fires in Sabah in 1983. Sabah
Soc J 8:10–30
Bell G (2005) The co-distribution of species in relation to the
neutral theory of community ecology. Ecology 86:1757–
1770. doi:10.1890/04-1028
Bell G, Lechowicz MJ, Waterway MJ (2006) The comparative
evidence relating to functional and neutral interpretations
of biological communities. Ecology 87:1378–1386. doi:
10.1890/0012-9658(2006)87[1378:TCERTF]2.0.CO;2
Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: a practical and powerful approach to multiple
testing. J R Stat Soc B 57:289–300
Botkin DB, Sobel MJ (1975) Stability in time-varying ecosystems. Am Nat 109:625–646. doi:10.1086/283034
Brünig EF (1969) On the seasonality of droughts in the lowlands of Sarawak (Borneo). Erdkunde 2:127–133
Brünig EF (1971) On the ecological significance of drought in
the equatorial wet evergreen (rain) forest of Sarawak
(Borneo). In: Flenley JR (ed) Transactions of the 1st
symposium on Malesian ecology. Geography Department,
University of Hull, UK, pp 66–97
Cane MA (2005) The evolution of El Niño, past and future.
Earth Planet Sci Lett 230:227–240. doi:10.1016/j.epsl.
2004.12.003
Chatfield C (2004) The analysis of time series, 6th edn.
Chapman and Hall, London, UK
Chave J (2004) Neutral theory and community ecology. Ecol
Lett 7:241–253. doi:10.1111/j.1461-0248.2003.00566.x
Chesson PL (1982) The stabilizing effect of a random environment. J Math Biol 15:1–36. doi:10.1007/BF00275786
Chesson P, Huntly N (1997) The roles of harsh and fluctuating
conditions in the dynamics of ecological communities.
Am Nat 150:519–553. doi:10.1086/286080
Christensen JH, Hewitson B, Busuioc A, Chen A, Gao X, Held
I, et al. (2007) Regional climate projections, Chap. 11. In:
Solomon S, Qin D, Manning M, Chen Z, Marquis M,
Averyt K, et al. (eds.), Climate change 2007: the physical
science basis. Contribution of working group I to the
fourth assessment report of the intergovernmental panel
on climate change. Cambridge University Press, Cambridge, UK, pp 847–940
Condit R, Hubbell SP, Foster RB (1995) Mortality rates of 205
neotropical tree and shrub species and the impact of a
severe drought. Ecol Monogr 65:419–439. doi:10.2307/
2963497
Condit R, Hubbell SP, Foster RB (1996) Changes in tree
species abundance in a neotropical forest: impact of climate change. J Trop Ecol 12:231–256
Condit R, Aguilar S, Hernandez A, Perez R, Lao S, Angehr G
et al (2004) Tropical forest dynamics across a rainfall
gradient and the impact of an El Nino dry season. J Trop
Ecol 20:51–72. doi:10.1017/S0266467403001081
DeAngelis DL, Waterhouse JC (1987) Equilibrium and nonequilibrium concepts in ecological models. Ecol Monogr
57:1–21. doi:10.2307/1942636
Engelbrecht BMJ, Comita LS, Condit R, Kursar TA, Tyree
MT, Turner BL et al (2007) Drought sensitivity shapes
species distribution patterns in tropical forests. Nature
447:80–82. doi:10.1038/nature05747
Gibbons JM (1998) Water relations, phenology and drought
adaptation of understorey trees in tropical lowland rain
forest. PhD thesis, University of Stirling, Scotland, UK
Gibbons JM, Newbery DM (2003) Drought avoidance and the
effect of local topography on trees in the understorey of
Bornean lowland rain forest. Plant Ecol 164:1–18. doi:
10.1023/A:1021210532510
Grubb PJ (1977) The maintenance of species-richness in plant
communities: the importance of the regeneration niche.
Biol Rev Camb Philos Soc 52:107–145. doi:10.1111/
j.1469-185X.1977.tb01347.x
Holm S (1979) A simple sequential rejective multiple test
procedure. Scand J Stat 6:65–70
Hubbell SP (2001) The unified neutral theory of biodiversity
and biogeography. Princeton University Press, Princeton,
USA
Hubbell SP (2005) Neutral theory in community ecology and
the hypothesis of functional equivalence. Funct Ecol
19:166–172. doi:10.1111/j.0269-8463.2005.00965.x
Hubbell SP (2006) Neutral theory and the evolution of ecological equivalence. Ecology 87:1387–1398. doi:10.1890/
0012-9658(2006)87[1387:NTATEO]2.0.CO;2
Hulme M, Viner D (1998) A climate change scenario for the
tropics. Clim Change 39:145–176. doi:10.1023/A:100537
6007729
Huston MA (1979) General hypothesis of species diversity.
Am Nat 113:81–101. doi:10.1086/283366
Huston MA (1994) Biological diversity: the coexistence of
species on changing landscapes. Cambridge University
Press, Cambridge, UK
Forest Ecology
Ives AR (1995) Measuring resilience in stochastic systems.
Ecol Monogr 65:217–233. doi:10.2307/2937138
Ives AR, Carpenter SR (2007) Stability and diversity of ecosystems. Science 317:58–62. doi:10.1126/science.1133258
Lingenfelder M (2005) Rain forest dynamics and response to
drought in a Bornean primary lowland dipterocarp forest.
PhD thesis. University of Bern, Switzerland. http://www.
stub.unibe.ch/download/eldiss/05lingenfelder_m.pdf
Malhi Y, Wright J (2004) Spatial patterns and recent trends in
the climate of tropical rainforest regions. Philos T R Soc B
359:311–329. doi:10.1098/rstb.2003.1433
Malhi Y, Pegoraro E, Nobre AD, Pereira MGP, Grace J, Culf
AD et al (2002) Energy and water dynamics of a central
Amazonian rain forest. J Geophys Res-Atmos 107:D20–
D34. doi:10.1029/2001JD000623
Manly B (1997) Randomization, bootstrap and Monte Carlo
methods in biology. Chapman and Hall, London, UK
Marsh CW, Greer AG (1992) Forest land-use in Sabah,
Malaysia: an introduction to the Danum Valley. Philos T
R Soc B 335:331–339. doi:10.1098/rstb.1992.0025
May RM (1974) Stability and complexity in model ecosystems.
Princeton University Press, Princeton, USA
McPhaden MJ (1999) Genesis and evolution of the 1997–98 El
Niño. Science 283:950–954. doi:10.1126/science.283.54
04.950
McPhaden MJ, Zebiak SE, Glantz MH (2006) ENSO as an
integrating concept in earth science. Science 314:1740–
1745. doi:10.1126/science.1132588
Meehl GA, Stocker TF, Collins WD, Friedlingstein P, Gaye
AT, Gregory JM, et al. (2007) Global climate projections,
Chap. 10. In: Solomon S, Qin D, Manning M, Chen Z,
Marquis M, Averyt K, et al. (eds) Climate change 2007:
the physical science basis. Contribution of working group
I to the fourth assessment report of the intergovernmental
panel on climate change. Cambridge University Press,
Cambridge, UK, pp 747–845
Moran MD (2003) Arguments for rejecting the sequential
Bonferroni in ecological studies. Oikos 100:403–405. doi:
10.1034/j.1600-0706.2003.12010.x
Nakagawa M, Tanaka K, Nakashizuka T, Ohkubo T, Kato T,
Maeda T et al (2000) Impact of severe drought associated
with the 1997–1998 El Niño in a tropical forest in Sarawak. J Trop Ecol 16:355–367. doi:10.1017/S02664674
00001450
Nelson W (1982) Applied life data analysis. Wiley, NewYork,
USA
Newbery DM, Lingenfelder M (2004) Resistance of a lowland
rain forest to increasing drought intensity in Sabah, Borneo. J
Trop Ecol 20:613–624. doi:10.1017/S0266467404001750
Newbery DM, Campbell EJF, Lee YF, Ridsdale CE, Still MJ
(1992) Primary lowland dipterocarp forest at Danum
Valley, Sabah, Malaysia: structure, relative abundance
and family composition. Philos T R Soc B 335:341–356.
doi:10.1098/rstb.1992.0026
Newbery DM, Campbell EJF, Proctor J, Still MJ (1996) Primary
lowland dipterocarp forest at Danum Valley, Sabah,
Malaysia. Species composition and patterns in the understorey. Vegetatio 122:193–220. doi:10.1007/BF00044700
Newbery DM, Clutton-Brock TH, Prance GT (1999a) Changes
and disturbance in tropical rainforest in South-East Asia.
Imperial College Press, London, UK
167
Newbery DM, Kennedy DN, Petol GH, Madani L, Ridsdale CE
(1999b) Primary forest dynamics in lowland dipterocarp
forest at Danum Valley, Sabah, Malaysia, and the role of
the understorey. Philos T R Soc B 354:1763–1782. doi:
10.1098/rstb.1999.0519
Payne RW (ed) (2000) GENSTAT reference manual. Lawes
Agricultural Trust, Rothamsted, UK
Phillips OL, Hall P, Gentry AH, Sawyer SA, Vásquez R (1994)
Dynamics and species richness of tropical rain forests. Proc Natl
Acad Sci USA 91:2805–2809. doi:10.1073/pnas.91.7.2805
Potts MD (2003) Drought in a Bornean ever-wet rain forest. J
Ecol 91:467–474. doi:10.1046/j.1365-2745.2003.00779.x
Purves DW, Pacala SW (2005) Ecological drift in nichestructured communities: neutral pattern does not imply
neutral process. In: Burslem DRFP, Pinard M, Hartley S
(eds) Biotic interactions in the tropics. Cambridge University Press, Cambridge, UK, pp 103–138
Sheil D, May RM (1996) Mortality and recruitment rate
evaluations in heterogeneous tropical forests. J Ecol 84:
91–100. doi:10.2307/2261703
Sheil D, Burslem DFRP, Alder D (1995) The interpretation and
misinterpretation of mortality rate measures. J Ecol 83:
331–333. doi:10.2307/2261571
Shugart HH (1998) Terrestrial ecosystems in changing environments. Cambridge University Press, Cambridge, UK
S-Plus 6 (2001) Guide to statistics. Insightful Corporation.
Seattle, Washington, USA
Steele JH (1985) A comparison of terrestrial and marine ecological systems. Nature 313:355–358. doi:10.1038/313355a0
Timmermann A, Oberhuber J, Bacher A, Esch M, Latif M,
Roeckner E (1999) Increased El Niño frequency in a
climate model forced by future greenhouse warming.
Nature 398:694–697. doi:10.1038/19505
Timmermann A, Jin FF, Collins M (2004) Intensification of the
annual cycle in the tropical Pacific due to greenhouse
warming. Geophys Res Lett 31:L12208. doi:10.1029/2004
GL019442
Trenberth KE (1997) The definition of El Niño. Bull Am Meteorol
Soc 78:2771–2777. doi:10.1175/1520-0477(1997)078<
2771:TDOENO>2.0.CO;2
Trenberth KE, Hoar TJ (1997) El Niño and climate change.
Geophys Res Lett 24:3057–3060. doi:10.1029/97GL03092
Vasseur DA, Yodzis P (2004) The color of environmental
noise. Ecology 85:1146–1152. doi:10.1890/02-3122
Walsh RPD (1996a) Drought frequency changes in Sabah and
adjacent parts of northern Borneo since the late nineteenth
century and possible implications for tropical rain forest
dynamics. J Trop Ecol 12:385–407
Walsh RPD (1996b) Climate. In: Richards PW (main author)
The tropical rain forest: an ecological study. Cambridge
University Press, Cambridge, UK, pp 159–236
Walsh RPD, Newbery DM (1999) The ecoclimatology of
Danum, Sabah, in the context of the world’s rainforest
regions, with particular reference to dry periods and their
impact. Philos T R Soc B 354:1869–1883. doi:10.1098/
rstb.1999.0528
Westfall PH, Young SS (1993) Resampling-based multiple
testing. Wiley, New York, USA
Woods P (1989) Effects of logging, drought, and fire on
structure and composition of tropical forests in Sabah,
Malaysia. Biotropica 21:290–298. doi:10.2307/2388278
Red spruce forest regeneration dynamics across a gradient
from Acadian forest to old field in Greenwich, Prince
Edward Island National Park, Canada
N. Cavallin Æ L. Vasseur
Originally published in the journal Plant Ecology, Volume 201, No. 1, 169–180.
DOI: 10.1007/s11258-008-9497-8 Springer Science+Business Media B.V. 2008
Abstract Red spruce forests have declined considerably throughout their range in the past decades. As
agricultural fields are abandoned and land becomes
available for reforestation, the possibility arises for
red spruce forests to expand onto them. This study
addresses the potential for red spruce forests to
expand onto adjacent old fields in Greenwich, Prince
Edward Island National Park, Canada. We examined
red spruce distribution and abundance, plant species
diversity and changes in community composition
along a gradient from the interior of red spruce
forests out towards the centre of adjacent old fields.
Examining the patterns of red spruce distribution and
abundance revealed that, where cultivation and
logging have been abandoned recently in the fields
and forests, regeneration is limited to the forest
stands, but in the sites with older fields and forests,
regeneration extends into and is more vigorous in the
fields. Although species diversity varied from forest
to field only for the tree and shrub layers, important
changes occurred in the ground species composition.
There is no evidence yet that the herbaceous species
N. Cavallin L. Vasseur
Département de Biologie, Université de Moncton,
Moncton, NB, Canada
L. Vasseur (&)
Laurentian University, 935 Ramsey Lake Road, Sudbury,
ON, Canada P3E 2C6
e-mail: lvasseur@laurentian.ca
present in the forest stands will colonise the old
fields. The results suggest that both environmental
differences among sites and length of time since the
fields were abandoned explain red spruce regeneration patterns. In order to more accurately assess the
potential for red spruce regeneration in old fields,
long-term monitoring of the production, dispersal and
viability of red spruce seeds from adjacent forests and
of the constraints to seedling establishment and
survival in old fields will be needed.
Keywords Regeneration Red spruce
Abandoned fields Plant community
Diversity
Introduction
Descriptions of the patterns and processes of succession abound in the ecological literature. However,
since successional trajectories are contingent upon
site-specific conditions, generalisations from one site
cannot necessarily be applied to another (Pickett
et al. 2001). For example, agricultural and logging
disturbances can have similar initial impacts, but the
regeneration dynamics following them can differ
significantly. This is because successional patterns
depend on the nature of the disturbance, on the life
history strategies of each species involved and on the
outcomes of the interactions among the species and
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_13
169
170
their environment (Pickett et al. 2001). This study
focuses on the red spruce (Picea rubens Sarg.)
regeneration in agricultural fields after their
abandonment.
Agricultural practices such as tilling, flattening,
clearing and planting greatly disturb the soil layer,
resulting in a field ground environment that differs
considerably from a forest ground environment. Soil
mycorrhizal colonisation (Barni and Siniscalco
2000), organic matter and nitrogen content (Richter
et al. 2000), nutrient availability and the condition of
the surface horizon (Pickett et al. 2001) change
significantly in agricultural soil. In most cases, the
conversion of forested land to agricultural land leads
to the elimination of the forest herbs (Ramovs and
Roberts 2003). Additional practices that influence
succession include spreading manure (Pickett et al.
2001), fertilisation (Richter et al. 2000) and ploughing (Pickett et al. 2001; Ramovs and Roberts 2003),
as well as post-agricultural management such as
pasturing (Howard and Lee 2002).
Recolonisation, whether in forest ecosystems or in
agricultural ecosystems, depends heavily on: (1) the
reproductive strategies of the species in question,
such as its mode of dispersal, (2) the spatial features
of the available site, such as its size and its distance
from the seed source, (3) the environmental characteristics of the new site and (4) the interactions
between the species (Bazzaz 1996; Li and Wilson
1998; Yao et al. 1999; Meiners et al. 2000, 2002;
Pickett et al. 2001; Howard and Lee 2002).
Red spruce is an eastern North American tree
species that regenerates from wind-dispersed seed.
Although its seedlings and saplings can persist
upwards of 50 years in the shade, its seeds are
seldom viable for longer than 1 year (see, Sullivan
1993 for a review of red spruce ecology). In a forest
environment that is not too disturbed, red spruce
seedlings and saplings can be preserved as advance
regeneration during logging and dominate forest
regeneration later (Greene et al. 1999; Reyes and
Vasseur 2003). Conversely, in an agricultural setting,
seed dispersal from sufficiently stocked, nearby
sources of reproductive trees will be necessary for
natural colonisation.
Red spruce is an important but declining component of eastern North American forests, and it is a
species characteristic of mature Acadian forests
(Loo and Ives 2003). Atmospheric pollution, acid
A.G. Van der Valk (ed.)
deposition (Johnson et al. 1992; McLaughlin et al.
1987) and deforestation (White and Cogbill 1992) are
the principal causes of its decline throughout its
range. Late-successional species, including red
spruce, have declined in abundance and age in Prince
Edward Island as a result of selective forest exploitation and clearing for agriculture. Consequently, the
forest composition has shifted to higher frequencies
of balsam fir (Abies balsamea (L.) P. Mill.), red
maple (Acer rubrum L.), white spruce (Picea glauca
A. Dietr.), white birch (Betula papyrifera Marsh.) and
trembling aspen (Populus tremuloides Michx.) (Loo
and Ives 2003 describe the pre-settlement and current
Acadian forest in detail). For these reasons, it is
important to monitor and protect red spruce in Prince
Edward Island.
This study assesses the species composition along
a gradient from red spruce forests to adjacent,
recently abandoned agricultural fields in order to
better understand the structure of the red spruce
population and its regeneration dynamics. Our study
examines (1) spatial variation of red spruce abundance and density from the forest into the field, (2)
the influence of adjacent forests on red spruce
abundance in the old fields and (3) the influence of
historical disturbance on species composition.
Methodology
Study sites
The study sites were located in Prince Edward Island
National Park (hereafter PEINP). In 1996, the park
acquired new land, including several agricultural
fields (Rennie et al. 1997). The cessation of agricultural disturbances (as of 2001 or earlier) on these
fields raised the possibility for adjacent red spruce
forests to expand onto them, and thus raised the
possibility to study the potential and mechanisms of
such regeneration.
Using aerial photographs from 1997 (scale
1:4,000, provided by PEINP, flight line A31757,
No. 51-52 and 80-83), all the sites in the Greenwich
section of the park where forests and old fields
occurred adjacent to one another, unseparated by
barriers (such as roads or footpaths), were located.
These potential sites were then visited to determine
whether they met the following criteria: (1) red
Forest Ecology
171
spruce is present in the tree layer; (2) the soil
drainage and texture are similar in all sites; (3) the
forest edge of each site is oriented in roughly the
same direction; and (4) the forest stand is large
enough for a full vegetation survey (105 m along the
edge, [50 m deep). In addition, sites with high
presence of white spruce were avoided.
Four sites were selected. They were all situated on
well-drained Orthic Humo-Ferric Podzols of the
Charlottetown map unit (soil classification map
provided by PEINP, Greenwich, 2003). The topography at the sites is gently rolling. The forest edges at
all sites are oriented roughly east-west. The mean
annual temperature of Prince Edward Island is about
5.5C; the summer mean is 15C and the winter mean
is -3.5C (Environment Canada 2005). The frostfree period averages 130 days per year between late
May and early October (Atlantic Climate Centre
2004). The mean annual precipitation ranges from
900 to 1,150 mm (Environment Canada 2005).
The four sites differ in their land use histories and
in some associated characteristics (Table 1). Just
before they were abandoned, the old field portions of
the four sites were used to grow hay, especially
timothy (Phleum pratense L.). Prior to that, they were
also used to grow potatoes (Lajeunesse 2004,
Table 1 Descriptive characteristics and histories of the four study sites in Greenwich, Prince Edward Island National Park
Site 1
Site 2
Site 3
Site 4
Treesa
60
78
17
93
Saplingsb
39
94
8
89
Percentage of red spruce
Aerial photoc
1935
Forest
Logged, hedgerow
Closed canopy
Logged
Logged
Field
Cultivated
Cultivated
Cultivated
Unclear
Forest
Logged
Closed canopy
Logged
Closed canopy
Field
Cultivated
Cultivated
Cultivated
Cultivated
1958
1974
Forest
Partially logged
Closed canopy
Regenerating
Closed canopy
Field
Cultivated
Cultivated
Cultivated
Cultivated
Forest
Closed canopy
Closed canopy
Closed canopy
Closed canopy
Field
Cultivated
Cultivated
Cultivated
Cultivated
Closed canopy
Closed canopy
Closed canopy
Closed canopy
Abandoned
Sapling dotted
Cultivated (hay)
Sapling dotted
Forest
Closed canopy
Closed canopy
Closed canopy
Closed canopy
Field
Abandoned
Sapling dotted
Cultivated (hay)
Sapling dotted
Approximate forest age
13–29 years
over 68 years
13–29 years
±45 years
Approximate number of years since field
abandonment
13
13
\3
13
Forest classificationd
Old-field white
spruce
Old-field white
spruce
Disturbed
hardwood
Old-field white
spruce
1990
1997
Forest
Field
2000
a
DBH C 10 cm
b
DBH \ 10 cm, height[1 m
c
Scales: 1935–1990 (1:\17,500); 1997 (1:4,000); 2000 (1:17,500)
d
Source: Forest classification map provided by PEINP, Greenwich, 2003
172
personal communication). Site 3 is the only one
where red spruce is not the most dominant species in
the tree and sapling layers of the forested component;
in this site, red maple (55%) dominates the tree layer
and a mixture of Amelanchier spp. (35%) and red
maple (31%) dominate the sapling layer.
A.G. Van der Valk (ed.)
positions are identified by the above number in
brackets, which designates the distance that follows
it. Each 5 m 9 5 m quadrat contained two embedded
1 m 9 1 m quadrats, one centered against the west
edge and one centered against the east edge.
Data collection
Species identification
Red spruce was differentiated from white and black
spruce (Picea mariana (P. Mill.) B.S.P.) morphologically, mostly using twig hair characteristics. Since
red spruce was far more abundant than white and
black spruce in the tree and sapling layers of the
study sites and their surroundings, seedlings were
assumed to be red spruce. Since white spruce
saplings sometimes have sparse twig hair (Jablanczy
1964), it is possible that some red spruce saplings in
this study might have been misidentified. It is
important to note that trees identified as red spruce
in our sites were most likely red spruce 9 black
spruce hybrids (Mosseler 2004, personal communication); morphological features are not reliable for
distinguishing pure spruce from hybrid spruce
(Nkongolo et al. 2003). For simplicity, and because
we do not know the degree of hybridisation of the
individual trees or stands, in this study we refer to the
probable red spruce 9 black spruce hybrids simply
as red spruce.
In each 5 m 9 5 m quadrat, we surveyed the mature
trees (DBH C 10 cm), the saplings (DBH \ 10 cm,
height C 1 m) and the shrubs (height C 1 m). In the
1 m 9 1 m quadrats, we surveyed the herbaceous
layer, including tree seedlings and shrubs shorter than
1 m. For the trees and saplings, we counted the
number of stems of each species; for the shrubs and
herbs, we estimated the percent cover for each
species. The survey began in July 2003 and ended
in September 2003. Since the survey began in July,
that year’s cohort of spruce seedlings had already
germinated (the vegetation in the plots was surveyed
only once). This survey design is adapted from the
Ecological Monitoring and Assessment Network’s
Terrestrial Vegetation Biodiversity Monitoring Protocols (Roberts-Pichette and Gillespie 1999) and
from Reyes (2002) and Reyes and Vasseur (2003).
All species were identified using Gleason and Cronquist (1991) and Roland (1998) and species names
were updated according to the Integrated Taxonomic
Information System database (2005, http://www.itis.
usda.gov/).
Survey plot design
Statistical analyses
Each of the four sites was 105 m wide (parallel to the
forest edge) and 110 m long (perpendicular to the
forest edge). We defined the forest edge as the
conspicuous boundary between the forest and the
field, regardless of the age of the trees. The width
included a 20 m buffer zone on each side. The sites
comprised five transects oriented perpendicular to the
forest edge and spaced 15 m apart. From a zero point
at the forest edge, each transect extended 50 m into
the forest and 60 m into the field. Eight 5 m 9 5 m
quadrats, each spaced 10 m apart, lined the west side
of each transect at the following distances from the
forest edge: (1) ?45 to 50 m, (2) ?30 to 35 m, (3)
?15 to 20 m, and (4) 0 to ?5 m in the forest; (5) -10
to 15 m, (6) -25 to 30 m, (7) -40 to 45 m and (8)
-55 to 60 m in the field. These distances are termed
as positions in the remaining of this article; the
Univariate analyses
Three diversity indices (species richness (r), the
Shannon–Weiner index (h0 ) and Simpson’s index (d))
for mature trees, saplings, shrubs and herbs, as well
as red spruce and balsam fir tree and sapling
abundances were calculated at the position level
(the average of the data from the five quadrats at the
same distance from the forest edge within a site).
Species for which there were only one or two
observations might have skewed the data so we
removed them from the data set, except in the shrub
layer where most species occurred only once or
twice. The data were not normally distributed
according to Kolmogorov–Smirnov and Shapiro–
Wilk tests and they could not be transformed to meet
Forest Ecology
173
the conditions for normality; therefore, they were
tested for significant differences among the positions
and among the sites using Kruskal–Wallis tests with
a = 0.05. In order to locate the differences when the
null hypothesis was rejected, a post priori multiple
comparison tests designed to follow Kruskal–
Wallis tests was used (Post Hoc for Kruskal; macro
available on StatSoftInc 2006). All univariate analyses were performed using SPSS 11.0, except the
nonparametric multiple comparisons which were
performed using STATISTICA, Version 6 (StatSoft,
Inc 2001).
We used 2 9 2 contingency tables to analyse the
relationships among red spruce seedling, sapling and
tree occurrences. Chi-square tests were followed with
contingency and uncertainty coefficient calculations.
Contingency coefficients measure the degree of
association between the variables. Uncertainty coefficients measure the proportional reduction of error in
the prediction of one variable when another is known.
Quadrat level correlations between red spruce tree
and sapling abundances were examined using Spearman rank correlation tests.
Multivariate analyses
In order to assess the similarities in forest and field
species composition in our sites, community level
changes in vegetation composition across the forestfield gradient were examined with detrended correspondence analysis (DCA) using CANOCO 4.02.
‘Species’ data were stem counts for the tree and
sapling analysis and percent cover for the herb and
shrub analysis. The ‘downweighting of rare species’
option was selected to prevent infrequently occurring
species from exerting too strong an influence on the
analysis (ter Braak and Šmilauer 1998). CANOCO
gives the option to add a supplementary environmental variable to an ordination by projecting its data
onto the existing ordination axes. This way, that
variable does not influence the calculation of the
axes, but its relation to the other variables in the
ordination can still be interpreted from the results. In
our analyses, we used position relative to the forest
edge as a supplementary environmental variable. The
actual DCA axes represent hypothetical gradients
along which the maximum variability in species
composition is explained (ter Braak and Šmilauer
1998).
Results
In the four sites surveyed, there were a total of seven
species present in the tree layer (DBH C 10 cm),
nine species in the sapling layer (DBH \ 10 cm and
height C 1 m), eight species in the shrub layer
(height C 1 m) and 71 species in the herbaceous
layer. The relative abundances of each tree and
sapling species are presented in Table 2. Only red
spruce and balsam fir saplings were present in the
field portions of the sites, and only two of the sites
Table 2 Relative abundances of the species (trees (DBH C 10 cm) and saplings (DBH \ 10 cm)) in each site
Site 1
Site 2
Site 3
Site 4
Trees
Saplings
Trees
Saplings
Trees
Saplings
Trees
Saplings
Abies balsamea
0.29
0.25
0.21
0.06
0.05
0.02
0.04
0.09
Acer rubrum
0.00
0.03
0.00
0.00
0.55
0.31
0.00
0.00
Amelanchier sp.
0.00
0.00
0.00
0.00
0.07
0.35
0.00
0.00
Betula papyrifera
0.11
0.24
0.01
0.00
0.00
0.01
0.00
0.00
Picea glauca
–
0.01
–
0.00
–
0.00
–
0.00
Picea mariana
0.00
0.00
0.00
0.00
0.07
0.04
0.03
0.02
Picea rubens
0.60
0.39
0.78
0.94
0.17
0.08
0.93
0.89
Populus tremuloides
0.00
0.00
0.00
0.00
0.10
0.19
0.00
0.00
Prunus pennsylvanica
–
0.07
–
0.00
–
0.01
–
0.00
N
70
71
111
400
42
108
101
45
– indicates no stems with DBH C 10 cm
174
A.G. Van der Valk (ed.)
Trees
Saplings
Shrubs
Herbs
25
Richness (r)
20
15
10
5
0
6
Shannon-Weiner (h' )
5
Red spruce distribution
4
3
2
1
0
1.0
0.8
Simpson's (d )
had saplings in their fields. Trees and saplings of all
other species were present only in the forest portions
of the sites.
Five diversity variables varied significantly along
the distance gradient relative to the forest edge
(Kruskal–Wallis, N = 32, df = 7): r trees (X2 =
25.353, P = 0.001), h0 trees (X2 = 25.477, P =
0.001), d trees (X2 = 14.896, P = 0.037), r shrubs
(X2 = 17.220, P = 0.016) and h0 shrubs (X2 =
17.896, P = 0.012) (Fig. 1). The results suggest that
the significant variation lies between forest and field.
As expected, nearly all the trees were located in the
forest. The shrub layer species diversity was highest
at the forest positions near the edge (Fig. 1). Only d
for the sapling layer differed significantly among
sites (X2 = 18.419, df = 3, P = 0.001), with sites 1
and 3, and sites 2 and 4 as two homogenous subsets
(Fig. 1).
0.6
0.4
0.2
+
+
45
-5
0
30
-3
5
+
15
-2
0
+
0-1 5
0-1
-2 5
5-3
-4 0
04
-5 5
560
0.0
Distance from the forest edge (m)
Fig. 1 Mean species richness (r) and mean value for
Shannon–Weiner index (h0 ) and Simpson’s Index (d) ± standard error for each vegetation layer at eight positions relative to
the forest edge; N = 4 (one index per position per site)
Red spruce tree (X2 = 90.561, df = 7, P \ 0.001)
and sapling (X2 = 18.664, df = 7, P = 0.009) abundance differed significantly with the distance from
the forest edge (Fig. 2). Mean red spruce tree stem
density was 1,100 stems/ha in the forest and 0 stems/
ha in the field; maximum density was 1,600 stems/ha
at position 2 (30–35 m into the forest). Mean red
spruce sapling stem density was 1,600 stems/ha in
the forest and 700 stems/ha in the field; maximum
density was 4,700 stems/ha at position 4 (0–5 m into
the forest) and it rapidly declined to 1,200 stems/ha
by position 6 (25–30 m into the field).
Sites differed significantly in red spruce tree
abundance (X2 = 16.466, df = 3, P = 0.001), with
subsets defined as sites 1, 2, and 3 together and sites 1,
2, and 4 together. Testing site differences using only
forest data (the field data were removed from the set
because of the dominance of zeroes), three subsets
were identified. Sites 1 and 2, and sites 2 and 4 were
grouped together, leaving site 3 separate (X2 =
34.476, df = 3, P \ 0.001). The number of red spruce
saplings also varied significantly between sites
(X2 = 39.951, df = 3, P \ 0.001). Sites 1 and 3, sites
1 and 4, and sites 2 and 4 formed three homogenous
subsets. Saplings have not begun to colonise the fields
of sites 1 and 3. Despite their similarities in abundance,
sites 2 and 4 differed in sapling distribution; it was
Forest Ecology
175
b Fig. 2 Mean stem count and standard error of red spruce and
5
RS trees
RS saplings
BF trees
BF saplings
4
Site 1
3
2
1
0
60
50
Site 2
40
30
20
10
0
1.8
1.6
1.4
Site 3
1.2
1.0
0.8
0.6
0.4
0.2
0.0
7
6
4
3
2
1
0
05
-1
015
-2
53
-4 0
04
-5 5
560
-2
15
+
5
30
-3
+
+
45
-5
0
0
+
Site 4
5
Distance from the forest edge (m)
balsam fir trees and saplings in each site at eight positions
relative to the forest edge; positive distances are in the forest
and negative distances are in the field. For each distance at
each site, N = 5 plots (5 m 9 5 m)
more even in the field of site 4 and patchier in the field
of site 2. Red spruce regeneration was most abundant
in site 2 (Fig. 2).
Red spruce trees occurred in 37.5% of the 160
surveyed quadrats and red spruce saplings occurred in
40.6% of the quadrats. There is a relationship
between red spruce tree and sapling occurrence
(Pearson X2 = 17.621, df = 1, P \ 0.001); however,
the association is not very strong (contingency
coefficient = 0.315; symmetric uncertainty coefficient = 0.083, P \ 0.001). The abundances of red
spruce trees and saplings were also significantly
correlated (Spearman rank correlation coefficient (rs):
0.299, P \ 0.001).
Red spruce seedlings (height \ 1 m) occurred in
16.3% of the 160 quadrats, always where red spruce
sapling abundance was the highest. The relationship
between red spruce seedlings and saplings (Pearson
X2 = 29.452, df = 1, P \ 0.001; contingency coefficient = 0.394, symmetric uncertainty coefficient =
0.173, P \ 0.001) was stronger than the one between
saplings and trees. There was no significant relationship between seedling and tree occurrences (Pearson
X2 = 2.755, df = 1, P = 0.097). Separating the
occurrence data into forest and field quadrats gives
the following stocking rates (percentage of quadrats
with at least one occurrence): in the forest, 18.8% for
seedlings, 53.8% for saplings and 72.5% for trees; in
the field, 13.8% for seedlings, 27.5% for saplings and
2.5% for trees.
Balsam fir is the only other tree species to have
begun colonising the old fields. Its abundance and
distribution are presented in Fig. 2. Balsam fir was less
abundant than red spruce in the tree and sapling layers
everywhere except at position 1 (45–50 m into the
forest) in site 1. Mean balsam fir tree stem density was
245 stems/ha in the forest and 0 stems/ha in the field;
maximum density was 360 stems/ha at position 1.
Mean balsam fir sapling stem density was 190 stems/
ha in the forest and 40 stems/ha in the field; maximum
density was 480 stems/ha at position 4. Balsam fir
trees were present in 16% of the quadrats and saplings
were present in 14% of the quadrats.
176
Vegetation gradient
DCA results exposed the strong field/forest distinction
in the species composition of our study sites, as well as
the distinction between the species composition of site
3 and that of the other sites. In the tree and sapling
layer ordination (Fig. 3), the four samples belonging
to the forest portion of site 3 are clustered on the left of
the diagram near the abundance optima for red maple,
trembling aspen and Amelanchier sp. All other forest
samples are positioned between the abundance optima
for balsam fir and red spruce trees, with no particular
order in the positions. All the field samples (that were
not void of saplings) are clustered around the optimum
for red spruce saplings on the right side of the diagram.
The eigenvalue of the first axis was 0.850 and that of
the second axis was 0.259; the cumulative percentage
of variance in the species data explained was 35.4%
for the first axis and 46.2% for the second. The sum of
all eigenvalues was 2.399. The eigenvalue calculated
for position relative to the forest edge as a supplementary environmental variable was 0.344.
Fig. 3 Detrended
correspondence analysis
biplot of the species
surveyed in the tree and
sapling layers at the four
study sites. Circles
represent samples, which
are labelled with their
position numbers. Position
numbers refer to the
distance from the forest
edge, with position 1
beginning in the forest
interior and position 8
ending furthest into the
field. The full species
names corresponding to the
species codes are in
Appendix 1
A.G. Van der Valk (ed.)
In the herbaceous and shrub layer ordination
(Fig. 4), all the forest species and samples are
positioned on the left of the diagram, while all the
field species and samples are positioned on the right
of the diagram. Forest positions and species are
loosely ordered from forest interior to edge from the
left of the diagram to the centre, except for the four
samples belonging to site 3 at the far left. Species
exclusive to or most abundant in site 3 include
Kalmia angustifolia L., Vaccinium angustifolium
Ait., Gaultheria procumbens L., and Nemopanthus
mucronatus (L.) Loes. Among the field species,
grasses are scattered at the top of the diagram.
There is no apparent ordering within the field
positions. The eigenvalue of axis 1 was 0.943 and
that of axis 2 was 0.361. The cumulative percentage
of variance in the species data explained was 24.4%
for the first axis and 33.8% for the second. The sum
of all eigenvalues was 3.858. The eigenvalue
calculated for position relative to the forest edge
as a supplementary environmental variable was
0.702.
Forest Ecology
177
Fig. 4 Detrended
correspondence analysis
biplot of the species
surveyed in the herbaceous
and shrub layers at the four
study sites. Circles
represent samples, which
are labelled with their
position numbers. Position
numbers refer to the
distance from the forest
edge, with position 1
beginning in the forest
interior and position 8
ending furthest into the
field. The full species
names corresponding to the
species codes are in
Appendix 1. Forest edge
Discussion
Red spruce populations appeared to be capable of
slowly expanding into at least some of the abandoned
fields. Saplings, mostly of red spruce, have begun to
colonise the fields of the two sites with the older
forests, but species in the herbaceous layer did not
overlap between forest and field. The fields of the two
sites with the younger forests had no saplings in the
year of our survey. This suggests that the time since
abandonment has been too short for significant
change in the fields. Even after 54 years of succession following clearcutting in an Acadian forest,
Moola and Vasseur (2004) found no evidence that
pre-harvest forest species composition was returning.
Among the species which they found to have the
greatest difficulty in recovering after clearcutting, the
following are also present in our forest sites: Aralia
nudicaulis, Coptis trifolia, Maianthemum canadense,
Monotropa uniflora and Trientalis borealis.
Red spruce regeneration density in our sites was
lower than in studies conducted on harvested forest
stands. For example, Roy et al. (2000) inventoried
red spruce regeneration 3 years after clearcutting
(80 m 9 120 m stripcuts) with protection of advance
growth in southern Quebec and recorded densities
between 900 and 5,000 stems/ha; their stocking rates
were 13–75%. Prévost and Pothier (2003) measured a
mean of 200 stems/ha in unscarified clearcuts and
17,000 stems/ha in scarified sites (the densities in this
study included both red and white spruce). In Nova
Scotia, Reyes (2002) found very high red spruce
regeneration densities with means of 34,000 seedlings/ha as far as 110 m into the cut and
76,000 seedlings/ha between 0 and 20 m from the
forest edge, 3 and 4 years after the cut. The seedlings
in these studies came from both advance regeneration
and from seeds dispersed after the harvest. The
differences in the densities found in our study
compared to those following forest harvesting, as
well as the importance of advance regeneration as a
seedling source in harvested sites, confirm that
recovery of forest stands on agricultural lands cannot
be compared to forest regrowth after logging. In
addition, the red spruce seeds from our sites had a
germination rate of 0.004% (Cavallin and Vasseur, in
press). This low germination rate might have resulted
from the high occurrence of hybridisation between
red and black spruce on Prince Edward Island
(Mosseler 2004, personal communication) and suggests a limited ability for natural regeneration.
The species diversity indices suggest that, other
than for the tree layer, diversity did not vary
significantly from forest to field. Nonetheless, the
pattern we observed in herbaceous species richness
approached the one in Meiners and Pickett (1999),
where species richness increased across the forestfield gradient to a peak 40 m into the field and then
declined. Despite the continuity in species diversity,
species composition greatly differs between habitats.
178
Red spruce was the dominant regenerating tree
species along the forest-to-field gradient in all sites
except site 3. A significant cover of both hardwoods
and Kalmia angustifolia in site 3 could explain the
competitive difficulty for red spruce seedling establishment. Ericaceous species, especially Kalmia
angustifolia, can be inhibitory to spruce establishment (Yamasaki et al. 1998; Mallik 2001).
Hardwoods also can compete considerably with
spruce seedling growth (Lautenschlager 1991; Roy
et al. 2000; Prévost and Pothier 2003). Site-specific
information about past disturbances, other than the
logging and farming visible in the aerial photos, was
not available, but it is possible that the currently
forested portion of site 3 received a different
treatment from the other sites in the past and thus is
undergoing a different successional path. Modifications to the environment such as adding fertilisers to
the soil, tree harvesting and burning all create
conditions favouring Kalmia growth and spread
(Mallik 1994, 1996).
Moreover, the dominance of red spruce trees in the
other three sites provides a greater seed source for
maintaining the species in the forest and for reforesting the fields. In fact, the percentage contribution
of red spruce to the tree composition is much higher
in these three sites (60–93%) than it is in the three
forest types in which it typically occurs in Prince
Edward Island (4.6–17.3%) (Sobey and Glen 2002).
Dibble et al. (1999) regarded the presence of seedbearing red spruce trees as a very important indicator
of red spruce regeneration habitat. We indeed found
the presence and abundance of red spruce saplings to
be significantly correlated with those of mature red
spruce. Balsam fir was the only other tree species
colonising the old fields. Its lower abundance
suggests that red spruce could dominate the future
old field forests as well.
In conclusion, our results suggest that red spruce
forests have the potential to serve as sources for
reforesting adjacent old fields. However, this regeneration will be slow, raising two choices concerning
park management: the fields can be left to reforest
slowly and naturally or they can be artificially
reforested to speed up the process. Further research
into the constraints to red spruce establishment and
persistence in old fields will be needed to develop a
restoration plan favouring natural regeneration. Alternately, red spruce can be grown naturally in fields
A.G. Van der Valk (ed.)
from seedlings (Beaulieu et al. 1989), but its susceptibility to drought and winter desiccation can limit its
success (Beaulieu et al. 1989; Blum 1990). Further
monitoring would be needed to better understand this
option as well.
Acknowledgement This research was funded by the PEINP
and by the K.C. Irving Chair in Sustainable Development. We
thank Denyse Lajeunesse, Lary Brown, Paul Ayles, Geneviève
Duclos, Isabelle Chiasson, Caspian Kilkelly and the PEINP
staff for their valuable assistance. We also thank Marc-André
Villard, Louis Bélanger and two anonymous reviewers for their
helpful comments on the manuscript.
Appendix 1
Species names corresponding to the codes in Figs. 3 and 4
Trees: subscripts t, s and h indicate tree layer, sapling layer and
herbaceous layer, respectively.
BF
Abies balsamea (L.) P. Mill.
RM
Acer rubrum L.
AS
Amelanchier sp. Medik.
WB
Betula papyrifera Marsh.
BS
Picea mariana (P. Mill.) B.S.P.
RS
Picea rubens Sarg.
TA
Populus tremuloides Michx.
PC
Prunus pensylvanica L. f.
Shrub layer
Amel
Ilex
Amelanchier sp. Medik.
Ilex verticilata (L.) Gray
More
Morella pensylvanica (Mirbel) Kartesz, comb. nov.
ined.
Nemo Nemopanthus mucronatus (L.) Loes.
Rosa
Rosa sp. L.
Samb
Sambucus racemosa L.
Spir
Spiraea alba Du Roi
Vibe
Viburnum nudum var. cassinoides (L.) Torr. & Gray
Herbaceous layer
achmil Achillea millefolium L.
elyrep Elymus repens (L.) Gould
agrcap Agrostis capillaris L.
aranud Aralia nudicaulis L.
astsp
Aster sp. L.
cararc Carex arctata Boott ex Hook.
carsp
Carex sp. L.
cirarv
Cirsium arvense (L.) Scop.
clibor
Clintonia borealis (Ait.) Raf.
coptri
Coptis trifolia (L.) Salisb.
corcan Cornus canadensis L.
Forest Ecology
Appendix continued
danspi
Danthonia spicata (L.) Beauv. ex Roemer &
J.A. Schultes
desfle
Deschampsia flexuosa (L.) Trin.
drycar
epirep
Dryopteris carthusiana (Vill.) H.P. Fuchs
Epigaea repens L.
frasp
Fragaria sp. L.
galsp
Galium sp. L.
gaupro
Gaultheria procumbens L.
hieaur
Hieracium aurantiacum L.
hiesp
Hieracium sp. L.
kalaang Kalmia angustifolia L.
linvul
Linaria vulgaris P. Mill.
luzmul
Luzula multiflora var. acadiensis Fern.
mellin
Melampyrum lineare Desr.
miacan Maianthemum canadense Desf.
monuni Monotropa uniflora L.
oxastr
Oxalis stricta L.
phlpra
Phleum pratense L.
poapra
Poa pratensis L.
v_grass Poaceae (vegetative)
pteaqu Pteridium aquilinum (L.) Kuhn
pyrsp
Pyrola sp. L.
ranrep
Ranunculus repens L.
rossp
Rosa sp. L.
rubida
Rubus idaeus L.
rumace Rumex acetosella L.
solcan
Solidago canadensis L.
eutgra
Euthamia graminifolia var. graminifolia (L.) Nutt.
solrug
Solidago rugosa P. Mill.
spialb
Spiraea alba Du Roi
stegra
Stellaria graminea L.
taroff
Taraxacum officinale G.H. Weber ex Wiggers
trapra
Tragopogon pratensis L.
tribor
Trientalis borealis Raf.
tripra
vacang
Trifolium pratense L.
Vaccinium angustifolium Ait.
vibnud
Viburnum nudum var. cassinoides (L.) Torr. & Gray
viccra
Vicia cracca L.
viosp
Viola sp. L.
References
Atlantic Climate Centre (2004) The climate of Prince Edward
http://atlantic-web1.ns.ec.gc.ca/climatecentre/.
Island.
Accessed 28 July 2006
Barni E, Siniscalco C (2000) Vegetation dynamics and arbuscular mycorrhiza in old-field successions of the western
179
Italian Alps. Mycorrhiza 10:63–72. doi:10.1007/s00572
0000059
Bazzaz FA (1996) Recruitment in successional habitats: general trends and specific differences. In: Plants in changing
environments: linking physiological, population, and
community ecology. Cambridge University Press,
Cambridge
Beaulieu J, Corriveau A, Daoust G (1989) Productivité et
stabilité phénotypique de l’épinette rouge au Québec. For
Chron 65:42–48
Blum BM (1990) Red spruce. In: Burns RM, Honkala BH (eds)
Silvics of North America. United States Department of
Agriculture, Forest Service, Washington, DC
Cavallin N, Vasseur L (2008) Potential for red spruce (Picea
rubens Sarg.) establishment from natural seed dispersal in
old fields adjacent to forest stands. Plant Ecol (in press).
doi:10.1007/s11258-008-9409-y
Dibble AC, Brissette JC, Hunter ML Jr (1999) Putting community data to work: some understory plants indicate red
spruce regeneration habitat. For Ecol Manag 114:275–
291. doi:10.1016/S0378-1127(98)00359-4
Environment Canada (2005) Narrative descriptions of terrestrial ecozones and ecoregions of Canada. http://www.ec.
gc.ca/soer-ree/. Accessed 28 July 2006
Gleason HA, Cronquist A (1991) Manual of vascular plants of
northeastern United States and adjacent Canada. New
York Botanical Garden, Bronx, NY, USA
Greene DF, Zasada JC, Sirois L, Kneeshaw D, Morin H,
Charron I et al (1999) A review of the regeneration
dynamics of North American boreal forest tree species.
Can J For Res 29:824–839. doi:10.1139/cjfr-29-6-824
Howard LF, Lee TD (2002) Upland old-field succession in
southeastern New Hampshire. J Torrey Bot Soc 129:60–
76. doi:10.2307/3088683
Hughes JW, Bechtel DA (1997) Effect of distance from forest
edge on regeneration of red spruce and balsam fir in
clearcuts. Can J For Res 27:2088–2096. doi:10.1139/cjfr27-12-2088
Jablanczy A (1964) Identification of black, red and white
spruce seedlings. Canada Department of Forestry Publication No. 1039. Queen’s Printer, Ottawa
Johnson AH, McLaughlin SB, Adams MB, Cook ER, DeHayes
DH, Eagar C et al (1992) Synthesis and conclusions from
epidemiological and mechanistic studies of red spruce
decline. In: Eagar C, Adams MB (eds) Ecology and
decline of red spruce in the eastern United States.
Springer, New York
Lautenschlager RA (1991) Red raspberry ecology and the
effect of raspberry and other forest brush on white spruce
growth. CFRU Information Report, College of Forest
Resources, Maine Agricultural Experiment Station, University of Maine, Orono, ME, USA
Li X, Wilson SD (1998) Facilitation among woody plants
establishing in an old field. Ecology 79:2694–2705
Loo J, Ives N (2003) The Acadian forest: historical condition
and human impacts. For Chron 79:462–474
Mallik AU (1994) Autecological response of Kalmia angustifolia to forest types and disturbance regimes. For Ecol
Manag 65:231–249. doi:10.1016/0378-1127(94)90173-2
Mallik AU (1996) Effect of NPK fertilizations on Kalmia
angustifolia: implications for forest disturbance and
180
conifer regeneration. For Ecol Manag 81:135–141. doi:
10.1016/0378-1127(95)03659-8
Mallik AU (2001) Black spruce growth and understory species
diversity with and without sheep laurel. Agron J 93:92–98
McLaughlin SB, Downing DJ, Blasing TJ, Cook ER, Adams
HS (1987) An analysis of climate and competition as
contributors to decline of red spruce in high elevation
Appalachian forests of the Eastern United States. Oecologia 72:487–501. doi:10.1007/BF00378973
Meiners SJ, Pickett STA (1999) Changes in community
and population responses across a forest-field gradient.
Ecography 22:261–267. doi:10.1111/j.1600-0587.1999.
tb00501.x
Meiners SJ, Handel SN, Pickett STA (2000) Tree seedling
establishment under insect herbivory: edge effects and
inter-annual variation. Plant Ecol 151:161–170. doi:
10.1023/A:1026509529570
Meiners SJ, Pickett STA, Handel SN (2002) Probability of tree
seedling establishment changes across a forest-old field
edge gradient. Am J Bot 89:466–471. doi:10.3732/ajb.89.
3.466
Moola FM, Vasseur L (2004) Recovery of late-seral vascular
plants in a chronosequence of post-clearcut forest stands
in coastal Nova Scotia, Canada. Plant Ecol 172:183–197.
doi:10.1023/B:VEGE.0000026326.09137.06
Nkongolo KK, Deverno L, Michael P (2003) Genetic validation and characterization of RAPD markers differentiating
black and red spruces: molecular certification of spruce
trees and hybrids. Plant Syst Evol 236:151–163. doi:
10.1007/s00606-002-0236-7
Pickett STA, Cadenasso ML, Bartha S (2001) Implications
from the Buell-Small Succession Study for vegetation
restoration. Appl Veg Sci 4:41–52
Prévost M, Pothier D (2003) Partial cuts in a trembling aspenconifer stand: effects on microenvironmental conditions
and regeneration dynamics. Can J For Res 33:1–15. doi:
10.1139/x02-147
Ramovs BV, Roberts MR (2003) Understory vegetation and
environment responses to tillage, forest harvesting, and
conifer plantation development. Ecol Appl 13:1682–1700.
doi:10.1890/02-5237
Rennie F, Curley R, Ristau C, Robichaud H, MacQuarrie K,
MacNeill R et al (1997) Proposed addition of the
Greenwich Dunes area to Prince Edward Island National
Park. Inventory and assessment of resource information
and park development concepts. Submitted to the Greenwich Steering Committee
Reyes GL (2002) Chapter 2: red spruce regeneration. Factors
affecting vegetation re-establishment patterns following
clearcutting of coastal red spruce-balsam fir stands in
southwestern Nova Scotia. Thesis, Dalhousie University,
Halifax
A.G. Van der Valk (ed.)
Reyes G, Vasseur L (2003) Factors influencing deer browsing
damage to red spruce (Picea rubens) seedlings in coastal
red spruce-balsam fir stands of southwestern Nova Scotia.
For Ecol Manag 186:349–357. doi:10.1016/S0378-1127
(03)00303-7
Richter DD, Markewitz D, Heine PR, Jin V, Raikes J, Tian K
et al (2000) Legacies of agriculture and forest regrowth in
the nitrogen of old-field soils. For Ecol Manag 138:233–
248. doi:10.1016/S0378-1127(00)00399-6
Roberts-Pichette P, Gillespie L (1999) Terrestrial vegetation
biodiversity monitoring protocols. EMAN Occasional
Paper Series, Report No. 9. Ecological Monitoring
Coordinating Office, Burlington, Ontario
Roland AE (1998) Roland’s flora of Nova Scotia, 3rd edn,
Revised by Zinck M. Nimbus Publishing and Nova Scotia
Museum, Halifax, NS, Canada
Roy V, Ruel J-C, Plamondon AP (2000) Establishment, growth
and survival of natural regeneration after clearcutting and
drainage on forested wetlands. For Ecol Manag 129:253–
267. doi:10.1016/S0378-1127(99)00170-X
Sobey DG, Glen WM (2002) The forests of Prince Edward
Island: a classification and ordination using multivariate
methods. Can Field Nat 116:585–602
StatSoft, Inc. (2001) STATISTICA (data analysis software
system). Version 6. Tulsa, Oklahoma
StatSoft, Inc. (2006) STATISTICA visual basic programs.
http://www.statsoft.com/downloads/maintenance/
svbprograms.html. Accessed 27 June 2006
Sullivan J (1993) Picea rubens. In: Fire effects information
system. Rocky Mountain Research Station, Fire Sciences
Laboratory (Producer), U.S. Department of Agriculture,
Forest Service. http://www.fs.fed.us/database/feis/. Accessed 15 Feb 2006
ter Braak CJF, Šmilauer P (1998) CANOCO reference manual
and user’s guide to Canoco for windows: software for
canonical community ordination (version 4). Microcomputer Power, Ithaca, NY
White PS, Cogbill CV (1992) Spruce-fir forests of eastern
North America. In: Eagar C, Adams MB (eds) Ecology
and decline of red spruce in the eastern United States.
Springer, New York
Yamasaki SH, Fyles JW, Egger KN, Titus BD (1998) The
effect of Kalmia angustifolia on the growth, nutrition, and
ectomycorrhizal symbiont community of black spruce.
For Ecol Manag 105:197–207. doi:10.1016/S0378-1127
(97)00285-5
Yao J, Holt RD, Rich PM, Marshall WS (1999) Woody plant
colonization in an experimentally fragmented landscape.
Ecography 22:715–728. doi:10.1111/j.1600-0587.1999.
tb00521.x
Distance- and density-dependent seedling mortality caused
by several diseases in eight tree species co-occurring
in a temperate forest
Miki Yamazaki Æ Susumu Iwamoto Æ Kenji Seiwa
Originally published in the journal Plant Ecology, Volume 201, No. 1, 181–196.
DOI: 10.1007/s11258-008-9531-x Springer Science+Business Media B.V. 2008
Abstract To examine whether the Janzen–Connell
mechanism applies to temperate forests, seedling
survival and causes of mortality were investigated at
two distances (beneath, far) from conspecific adults
and at two densities (high, low) at each distance for
seedlings (n = 7935) of eight tree species co-occurring in a hardwood forest. Six of the eight species
showed distance- and/or density-dependent seedling
mortality mainly caused by diseases and rodents. In
four of the five species primarily killed by disease
(i.e. damping-off, blight, rot, powdery mildew), the
infectivity (probability of infection by the disease)
and/or the virulence (proportion of seedlings killed to
those infected by the disease) were higher beneath
than far from conspecific adults. These findings
suggest that host specificity and/or spatially heterogeneous activity of natural enemies play an important
role in the reciprocal replacement of tree species,
maintaining species diversity in temperate forests.
M. Yamazaki (&) K. Seiwa
Laboratory of Forest Ecology, Field Science Center,
Graduate School of Agricultural Science, Tohoku
University, Osaki, Miyagi 989 6711, Japan
e-mail: owl@bios.tohoku.ac.jp
S. Iwamoto
Innovative Drug Research Laboratories, Kyowa Hakko
Kirin Co., Ltd., Machida, Tokyo 194-8533, Japan
Keywords Disease Herbivore Host specificity
Janzen–Connell mechanism Species diversity
Introduction
Attempts to understand the coexistence and replacement of plant species in natural communities have
focussed on biotic interactions (Dinoor and Eshed
1984; Burdon 1991; Coley and Barone 1996; Wright
2002). In forest communities, Janzen (1970) and
Connell (1971) hypothesised that host-specific natural enemies such as pathogens and vertebrate and
invertebrate herbivores can maintain a high diversity
of tree species if they are more likely to damage and
kill juveniles growing at high densities or close to
conspecific adults. This high mortality might, in turn,
liberate areas for colonisation and recruitment by
other tree species and thereby contribute to the
maintenance of high local diversity. To evaluate the
extent to which the Janzen–Connell mechanism leads
to species coexistence, it is important to examine how
many tree species co-occurring within a forest
community exhibit these types of mechanisms (Bell
et al. 2006). If most tree species within a forest show
distance- and/or density-dependent juvenile mortality, the reciprocal replacement of tree species would
be promoted, resulting in a higher probability of tree
species co-existence within the forest. However, most
studies of the Janzen–Connell mechanism have
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_14
181
182
focused on a single tree species in forest communities
of tropical (see references in Wright 2002; Gilbert
2002, 2005; Marquis 2005; Freckleton and Lewis
2006) and temperate regions (Packer and Clay 2000;
Masaki and Nakashizuka 2002; Tomita et al. 2002;
Seiwa et al. 2008). Several community-level studies
have also analysed the distance- and/or densitydependent mortality of juveniles for a substantial
number of tree species co-occurring within a forest
community in tropical (e.g. Wills et al. 1997; Harms
et al. 2000; Peters 2003; Webb et al. 2006) and
temperate forests (Streng et al. 1989; Jones et al.
1994; Hille Ris Lambers et al. 2002), but most of
these studies did not identify the killing agents (but
see Augspurger 1984). In field conditions, competition and adult interference may also affect juvenile
mortality; thus, the killing agents must be identified
to apply this mechanism (Janzen 1970; Clark and
Clark 1984). To elucidate whether the Janzen–
Connell mechanism mediates species diversity in
temperate forests, we conducted a field experiment
investigating both the killing agents that cause
seedling mortality and their manner of attack (i.e.
distance or density dependency or both) for eight tree
species co-occurring in a temperate forest.
In both tropical and temperate forests, tree juveniles are usually attacked by a variety of natural
enemies, such as pathogens and invertebrate and
vertebrate herbivores, and the enemies show a variety
of means of attack. For example, invertebrate and
vertebrate herbivores usually kill juveniles in the
density- and/or distance-dependent manner (e.g.
Maeto and Fukuyama 1997; Wada et al. 2000;
reviewed by Marquis 2005), although relatively
fewer evidences have been reported in vertebrate
herbivores (Akashi 1997; Tomita et al. 2002; Silman
et al. 2003; Wyatt and Silman 2004). On the other
hand, juvenile mortality from pathogens is better
predicted by distance from mature trees than by
juvenile density (Augspurger and Kelly 1984; Packer
and Clay 2000; Masaki and Nakashizuka 2002;
Tomita et al. 2002; Gallery et al. 2007). Although
several exceptions have been observed (e.g. Bell
et al. 2006; reviewed by Freckleton and Lewis 2006),
results of these studies suggest that the manner of
attack differs among types of natural enemies.
However, in forests, juvenile density is often correlated with distance from conspecific adults (Packer
and Clay 2000; Seiwa et al. 2008), leading to
A.G. Van der Valk (ed.)
misunderstandings of the manner of attack of each
agent. Thus, to explicitly distinguish the effects of
density from those of distance, it is necessary to
independently manipulate density and distance in the
field (Clark and Clark 1984). To clarify whether
natural enemies have a specific mode of attack,
comparative studies are required for several tree
species co-occurring within a forest community.
Host-specificity of natural enemies is crucially
important to apply the Janzen–Connell hypothesis
(Janzen 1970; Packer and Clay 2000; Gilbert 2005;
Freckleton and Lewis 2006; Seiwa et al. 2008). In
individual forest communities, seedlings of several
species coexist in a spatially complex understorey
community that is generated by overlapping seed
shadows of the adults. If herbivores or pathogens
attack in a distance- or density-dependent manner
without host specificity, all tree seedlings co-occurring with the focal species (even under heterospecific
adults) would also be affected by generalist pathogens
when total seedling density is high. Several studies
have recently examined the virulence of pathogens
beneath parents that more strongly affect seedlings
through negative feedback mechanisms or local
adaptation, not only in herbaceous and grassland
communities (Bever 1994; Bever et al. 1997; Mills
and Bever 1998; Lively and Dybdahl 2000; Sicard
et al. 2007) but also in forest communities (Packer
and Clay 2000, 2004; Hood et al. 2004; Reinhart et al.
2005; Seiwa et al. 2008). If locally adapted parasites
infect a maximum number of local hosts depending on
the host population density (Kirchner and Roy 2002;
Dybdahl and Storfer 2003), disease infectivity and
consequent seedling mortality will be higher beneath
compared with far from conspecific adults. On the
other hand, several authors have noted that disease
infectivity does not always correspond to disease
lethality (Kirchner and Roy 2002; Dybdahl and
Storfer 2003), probably because of differences in the
extent of local adaptation or intensity of negative
feedback. Therefore, information regarding the virulence as well as infectivity of pathogens may provide
more accurate estimates of host specificity. However,
in field conditions, a variety of disease symptoms are
often observed, even within a single seedling of an
individual tree species, and each symptom may also
involve a variety of fungal species (e.g. GarciaGuzman and Dirzo 2001; Schafer and Kotanen 2004),
pointing to the difficulty in testing host specificity for
Forest Ecology
all diseases by inoculation or sterilisation methods.
Here, for convenience, we classified the diseases
causing seedling mortality into four predominant
symptoms (i.e. damping-off, blight, rot and powdery
mildew), and we evaluated the host specificity of these
diseases for seedlings of five tree species that are
killed primarily by disease by investigating the
difference in pathogenicity beneath and far from
conspecific adults. Specifically, we addressed (i) how
and what natural enemies attack the seedlings of each
of the eight tree species co-occurring within a forest
community; (ii) differences in the manner of attack
among the natural enemies; (iii) host specificity of the
predominant diseases; and (iv) the potential validity
of the Janzen–Connell mechanism mediating species
diversity in a temperate forest.
Materials and methods
Study site and species
This study was conducted in the reserve area (ca.
168 ha) of an experimental forest at the Field Science
Centre of Tohoku University in northeastern Japan
(38480 N, 140440 E, altitude 500–610 m). Mean
monthly temperatures ranged from 1.0C (January) to
22.5C (August) in 2004. Mean annual temperature
and rainfall were 11.0C and 1563 mm, respectively.
Trees in the reserve area have re-established after
clear-cutting 60 years ago, and the area has been
protected from human activity as a forest reserve for
at least last 40 years. The canopy layer was dominated by Fagus crenata and Quercus serrata as
mosaic patches, and the total basal area of the two
species was approximately 20% (Terabaru et al.
2004). In this study, eight deciduous broadleaf tree
species that were common in this forest were selected
as focal plants: Prunus grayana, Cornus controversa,
Magnolia obovata, Fraxinus lanuginosa, Acer mono,
Castanea crenata, F. crenata, and Q. serrata. As
these species belong to different genera, we hereafter
refer to them by their genus names.
Experimental design and demographic censuses
Three adult trees were selected for each study species
in a 30-ha area within the reserve. For each species,
individual adults were isolated from each other and
183
were at least 50 m from the nearest conspecific adult.
In each species, we selected two distance intervals:
0–3 m (beneath) and 25–50 m (far) from each adult
tree, representing non-dispersed fruits and fruits
dispersed far from the adults, respectively (Masaki
et al. 1994; Akashi 1997; Maeto and Fukuyama
1997; Tomita et al. 2002; Seiwa et al. 2008; Yamazaki and Seiwa, unpublished data). For each adult
tree, four and eight quadrats (35 9 45 cm) were
randomly established within the 0–3 m (beneath) and
25–50 m (far) ranges, respectively. Exceptions were
Prunus, Fraxinus and Castanea, for which the far
quadrats were established 50–75 m from conspecific
adults, because there were few flat areas near the
adults. Within the far ranges for each species, each
far quadrat was located under non-conspecific adult
trees. In both Fagus and Quercus, four quadrats were
established at each of the two distance intervals.
Before the seeds were sown, naturally fallen seeds
were removed from the both litter and soil layer of
approximately 3 cm in depth in each quadrat.
In 2003, fresh seeds were collected in the experimental forest from more than three canopy trees for
each of six species (Prunus, Cornus, Magnolia,
Fraxinus, Acer and Castanea) or from the ground
under more than three adult trees for each of Fagus
and Quercus. Fagus seeds were collected from forests
100 km north of the study area, because 2003 was a
non-mast year in the study forest. Seeds were floated
in water to eliminate non-viable seeds. The number of
seeds sown in each quadrat differed among species
according to the density of naturally falling seeds or
newly emerged seedlings (Maeto and Fukuyama
1997; Masaki and Nakashizuka 2002; Tomita et al.
2002; Seiwa et al. 2008). The number of seeds sown
in each high- and low-density quadrat was 200 and 30
for Prunus, Cornus and Fraxinus, 100 and 20 for
Acer, 50 and 10 for Magnolia, Fagus and Quercus,
and 45 and 6 for Castanea, respectively. The total
number of seeds sown was 18,738. To avoid severe
seed predation by mammals and jays during the
winter, the quadrats were covered by 0.3 9 0.3-cm
mesh nets buried to a depth of 10 cm. The height of
the nets was 10 cm. These nets were removed when
seedlings started to emerge during the following
spring.
In each quadrat, all newly emerging seedlings
were tagged and monitored for survivorship, causes
of mortality, and disease symptoms and signs.
184
Measurements started on 7 April 2004, immediately
after snow melt, and were repeated weekly or biweekly until 9 November, and then monthly for the
remainder of the first growing season. The number of
seedlings that emerged in each of the high- and lowdensity treatments in the quadrats is shown in
Appendix 1.
Causes of seedling mortality
All causes of seedling mortality were investigated
during the first growing season for all seedlings
(n = 7935) that emerged in 2004. The causes were
classified as follows: (1) disease: disease symptoms
and signs were determined following Agrios (1997)
and Horst (2001); (2) invertebrate herbivores: seedling death caused by predation of hypocotyls, leaves
(more than 80% of the area) or roots; (3) vertebrate
herbivores: severing of main stems by mammals such
as wood mice, rabbits and Japanese serows. We
distinguished symptoms of vertebrates from those of
pathogens and invertebrates on the strength of the
traces of mammals such as the shape of the teeth, cut
edge, animal droppings and turned soil by mammals;
(4) withering: drying and uprooting; and (5) physical
damage: fallen branches and snow pressure.
Identification of disease types and isolation
of fungi
Although more than one symptom or sign appeared
on individual seedlings, each disease was quantified
separately, and the most crucial symptom that
affected the stem or largest proportion of leaves
was regarded as the disease that caused seedling
death. In this study, diseases that caused seedling
death were classified into four predominant types as
follows: (1) Damping-off: succulent stems became
water-soaked, necrotic, and sunken near the soil line
or at ground level. (2) Blight: primary symptoms
were small round or irregular brown spots on leaves
of Prunus and Fraxinus, dark brown spots on
Magnolia, and black spots on Cornus, Magnolia
and Fraxinus. The spots enlarged and coalesced to
cover most of the individual leaves. Secondary
symptoms were leaf defoliation or enlargement of
lesions to stems, followed by stem die-back from the
tip. (3) Rot: symptoms included softening, discoloration and disintegration of leaves, followed by stem
A.G. Van der Valk (ed.)
die-back. (4) Powdery mildew: patches of white to
greyish and powdery spots on leaf surfaces. Mildew
first grew on young leaves, and then the top of the
stem often died.
For each type of diseases, the disease infectivity,
virulence and seedling mortality were calculated for
all seedlings in each high-density quadrat at each
distance for each species. Disease infectivity was
defined as the proportion of seedlings infected to
those emerged. Virulence was defined as the proportion of seedlings killed by the disease to those
infected by the disease. Seedling mortality was
defined as the proportion of seedlings killed by the
disease to those emerged.
To determine how many and what pathogens
caused the disease symptoms, individual fungi were
identified. We sampled at least two or three diseased
parts of or whole of the dead individuals at each
measurement time. Approximately 10% of dead
individuals were randomly chosen (Appendix 2),
and the fungi were isolated immediately after sampling. Seedlings damaged by disease were dissected
into small pieces of approximately 5 mm in length,
submerged in 70% ethanol (v/v) for 30 s, and then
surface sterilised for 60 s in a solution of 0.1%
sodium hypochlorite (v/v). Samples were then rinsed
in sterile distilled water and placed in corn meal agar
(Nissui, Tokyo, Japan) and incubated at 20C. Five to
a month after inoculation, cultures were purified
using single-spore isolation or by transforming small
portions of the culture medium several times,
including the advancing margin of the mycelia.
Fungal morphotypes were described, and when
possible, morphologically identified to genus.
Micro-environmental conditions
To estimate the environmental light conditions in
which seedlings were growing, hemispherical photographs were taken with a fish-eye camera (Nikon,
F8 mm) at a height of 0.3 m above the ground in each
high- and low-density quadrat both beneath and far
from conspecific adults on 6 August 2004. Canopy
openness in each quadrat was calculated as the ratio
of the open-to-closed portions of the canopy in the
entire hemispherical area. Photographs were converted to computer data, and the photosynthetic
photon flux density (PPFD) was computed using Gap
Light Analyzer (GLA) version 2.0 (Frazer et al.
Forest Ecology
1999). Relative PPFD (rPPFD) was obtained by
dividing PPFD in each canopy openness dataset by
that under an obstructed sky. Soil water potential was
also estimated using soil tensiometers (DIK-8331 pF
meter, Daiki Rika Kogyo Co. Lid.) near each quadrat
on 14 October 2004.
Data analysis
To evaluate differences in micro-environmental conditions (relative PPFD, %, soil moisture, pF values)
between distances (beneath, far), one-way analysis of
variance (ANOVA) was conducted separately for
each species.
To examine whether the percentage of seedling
survival depended on distance from conspecific adults
and/or seedling density, survival time models based on
a Weibull distribution were constructed for each
species. To examine whether the probability of
seedling death depended on distance, seedling density
or the percentage of seedlings killed by each agent
(disease, invertebrate herbivore, vertebrate herbivores,
withering and physical damage), these factors were
analysed using two-way ANOVA for each species.
Prior to ANOVAs, the percentage values were arcsine
transformed. Data for each quadrat at each distance
were pooled across the two different densities.
Each of the three disease indices (disease infectivity, virulence and seedling mortality) was
compared between distances (beneath versus far)
using Wilcoxon tests for each disease symptom. In
this analysis, data from high-density quadrats were
used because of the small sample size in the lowdensity plots. Statistical analyses were performed
using JMP version 4.0.5 (SAS Institute Inc., Cary,
North Carolina).
Results
Micro-environmental conditions
There were few differences in rPPFD (%) and soil
moisture (pF) between the two distances (beneath
versus far) for most of the eight species studied (oneway ANOVA: F \ 0.37, P [ 0.54). As an exception,
rPPFD was greater beneath compared with far from
both Castanea and Quercus (F [ 16.3, P \ 0.01),
whereas the reverse was true in Fagus (F = 36.51,
185
P \ 0.01). The pF value was greater beneath compared with far from Prunus (F = 9.02, P \ 0.05).
Seedling survival
In all the species studied, seedling death occurred
throughout the growing season, particularly during
the rainy period from June to July (Fig. 1). For most
species, seedling survival was usually lower beneath
compared with far from conspecific adults, although
data from Castanea were not significant (Table 1,
Fig. 1). An exception was Quercus, in which the
reverse was true. Seedling survival was significantly
lower at high density relative to low density for
Magnolia, Fraxinus and Cornus (marginal), whereas
the reverse was true for Fagus (Table 1, Fig. 1). For
the other four species, seedling survival did not differ
between densities. An interaction between distance
and density was only observed in Cornus, in which
seedling survival was lower at high density than at
low density beneath conspecific adults but did not
differ between the two densities in the far treatment
(Table 1, Fig. 1).
Causes of seedling mortality
In each species, seedlings were killed by a variety of
agents. In particular, disease accounted for a larger
proportion of seedling deaths compared to other
agents (i.e. vertebrate and invertebrate herbivores,
physical damage, withering) at both distances for
Prunus, Cornus, Magnolia, Fraxinus and Quercus
(Fig. 2). Two-way ANOVAs indicated that the proportion of seedlings dying due to disease was
significantly higher beneath compared with far from
conspecific adults for Prunus, Cornus, Magnolia and
Fraxinus, whereas the opposite was true for Quercus
(Table 2, Fig. 2). The proportion of seedlings dying
due to disease was also higher at high density rather
than low density for Cornus, Fraxinus and Castanea,
whereas minimal differences were observed for the
other five species (Table 2, Fig. 2). An interaction
between distance and density was observed only for
Prunus, in which the probability of death by disease
did not differ between the two densities beneath
adults but was higher at high density compared to low
density in the far treatment (Fig. 2).
In Fraxinus, Castanea and Fagus, the proportion
of seedlings killed by vertebrate herbivores was
186
A.G. Van der Valk (ed.)
Fig. 1 Seedling survivorship from April to November 2004
for two density treatments (high and low) at two distances
(beneath and far) from conspecific adult trees for eight
deciduous broadleaf tree species. High (d) and low (m)
density beneath conspecific adults; high (s) and low (4)
density far from conspecific adults
Table 1 Results of survival time analysis assessing the effects
of distance (beneath or far) from conspecific adult trees, density (high or low) and their interaction on V2 values for the
eight study species
Infectivity, virulence, and seedling mortality
Species
Prunus grayana
Cornus controversa
Magnolia obovata
Distance (Di) Density (De) Di 9 De
d.f. = 1
d.f. = 1
d.f. = 1
93.53***
4.81*
77.59***
Fraxinus lanuginosa 121.27**
1.87
2.25
2.73
20.61***
21.97***
0.68
139.70**
1.84
Acer mono
4.52*
1.03
0.02
Castanea crenata
3.02
0.16
0.00
Fagus crenata
20.18***
5.61*
0.37
Quercus serrata
14.15***
0.20
1.83
P \ 0.1, * P \ 0.05, ** P \ 0.01, *** P \ 0.001
higher beneath compared with far from conspecific
adults (Table 2, Fig. 2), whereas the reverse was true
for Cornus. In Magnolia, the probability of death by
vertebrate herbivores was higher at high density
relative to low density. For each of the other killing
agents (invertebrate herbivores, withering and physical damage), few differences were observed between
distances (F \ 0.007, P [ 0.931) or between densities (F \ 0.005, P [ 0.947) for all the species studied
(Fig. 2).
For Prunus, Cornus, Magnolia, Fraxinus and Quercus, which were mainly killed by disease in a distantdependent manner, seedling mortality was primarily
caused by four predominant diseases (damping-off,
blight, rot, powdery mildew), although several disease symptoms (e.g. sooty mould, rust, leaf spot)
were also observed. Damping-off diseases were
chiefly observed during the early growing season,
particularly from seedling emergence to July, for all
the species. Thereafter, foliar diseases (e.g. blight,
rot, powdery mildew) were observed. The fungal
genera Colletotrichum, Phoma, Fusarium, Cylindrocarpon, Cladosporium and Alternaria were isolated
from samples of damping-off diseases, which were
observed in all the species (Appendix 3). In samples
of blight and rot, the genera Colletotrichum, Phoma
and Cladosporium were primarily detected in dead
seedlings (Appendix 3). In blight of Cornus and
damping-off of Fraxinus, both infectivity and mortality were higher beneath compared with far from
adults, but minimal differences in virulence were
observed between the distances (Fig. 3). In contrast,
in blights of Prunus and Magnolia, both virulence
and mortality were higher beneath compared with far
from adults, although few differences were observed
Forest Ecology
187
Fig. 2 Seedlings killed by each agent at two different densities (high, low) at two different distances (beneath, far)
Table 2 Results of two-way ANOVAs assessing the effects of distance (beneath and far) from conspecific adult trees, seedling
density (high and low) and their interaction on F values of disease and vertebrate herbivores per species
Species
Factor causing death
Distance (Di)
Prunus grayana
Disease
127.45***
1.07
0.68
0.68
0.01
12.81**
0.09
Vertebrate herbivores
Cornus controversa
Disease
Vertebrate herbivores
Magnolia obovata
Fraxinus lanuginosa
Disease
26.42***
Density (De)
Di 9 De
11.78**
8.44**
1.00
1.13
15.56***
1.24
1.04
Vertebrate herbivores
0.41
4.52*
0.06
Disease
4.25*
10.26**
0.00
Vertebrate herbivores
9.06**
0.11
0.01
Disease
0.53
2.19
0.02
Vertebrate herbivores
1.10
2.28
0.90
Castanea crenata
Disease
0.91
5.28*
1.73
Fagus crenata
Vertebrate herbivores
Disease
3.08
3.34
0.56
2.06
0.40
0.05
Acer mono
Vertebrate herbivores
Quercus serrata
21.36***
0.72
0.64
Disease
5.93*
0.01
1.28
Vertebrate herbivores
3.24
0.20
0.31
P \ 0.1, * P \ 0.05, ** P \ 0.01, *** P \ 0.001
Degree of freedom for distance (Di), density (De) and Di 9 De = 1, 32, with exceptions of Fagus and Quercus, where Di, De and
Di 9 De = 1, 20
188
A.G. Van der Valk (ed.)
Fig. 3 Percentages of
infectivity, virulence, and
seedling mortality beneath
(shaded bars) and far (open
bars) from conspecific
adults for four predominant
diseases in five tree species.
Only data for high-density
plots are given. Error bars
represent standard errors.
Asterisks indicate
significant differences
between distances
(Wilcoxon test * P \ 0.05,
** P \ 0.01,
*** P \ 0.001). Numbers
in parentheses are total
numbers of seedlings
infected by each disease
in infectivity (Fig. 3). In rot of Prunus and dampingoff of Magnolia, all the three indices (infectivity,
virulence, mortality) were higher beneath compared
with far from adults (Fig. 3). For all the three indices,
few differences were observed between distances in
damping-off diseases of Prunus, Cornus and Quercus, and in powdery mildew of Cornus (Fig. 3).
Discussion
Distance- and/or density-dependent seedling
mortality by natural enemies
Our results clearly demonstrated that seedlings of
temperate tree species are mainly killed by biotic
agents, particularly diseases and vertebrate herbivores, in a positive distance-dependent and negative
density-dependent manner, rather than by a deficiency of environmental resources such as light and
soil moisture. These patterns were observed for six
of the eight co-occurring study species and resulted
in a greater proportion of seedling death beneath
conspecific adults, particularly at higher seedling
density. These findings strongly suggest that recruitment of heterospecific seedlings to the free space
near conspecific adult trees is largely promoted in
this temperate forest community. If such reciprocal
replacements occur under adults of most common
tree species co-occurring within individual forest
communities, species diversity would be maintained
(Janzen 1970). Although there are increasing evidences in the density- and/or distance-dependent
mortality in juveniles including seed-to-sapling
stages, this short-term study focused on only
seedling stage. Because distance- and densitydependent reductions in performance accumulate as
juveniles grow (Gilbert et al. 2001; Wright 2002;
Packer and Clay 2003; Seiwa et al. 2008), further
studies including seed-to-sapling stages and the
spatial distribution of saplings are needed to confirm
the Janzen–Connell mechanisms. Although the full
extent of recruitment reductions near fruiting conspecifics
may
have
been
systematically
underestimated, the results from this communitylevel study strongly suggest that the Janzen–Connell
mechanism have the potential to affect species
diversity in temperate as well as tropical forests.
Forest Ecology
In this study, Quercus was an exception to the
Janzen–Connell hypothesis. Surprisingly, seedling
mortality of Quercus was lower beneath compared
with far from conspecific adults. Canopy trees of
Quercus usually started to unfold their leaves
approximately 1 month later than conspecific seedlings and other heterospecific canopy trees cooccurring at the study site (Yamazaki and Seiwa
unpublished data). In temperate forests, tree seedlings
with earlier leaf emergence compared to overstorey
trees usually gain more favourable light prior to
canopy closure, thereby producing large and sturdy
seedlings that usually show high resistance to herbivores and disease compared with late emerging
seedlings (Seiwa 1998). For seedlings of Quercus,
such phenological advantages in acquiring spring
light and escaping from natural enemies would
enhance survival under conspecific adults compared
with heterospecific trees. These traits suggest that
phenological events may be one of the most important factors required to evaluate the validity of the
Janzen–Connell hypothesis, especially in temperate
forests.
Primary killing agents and the manner of attack
In this study, the most important killing agent clearly
differed among species, according to seed size.
Vertebrate herbivores (mainly rodents) were the
major cause of mortality for the large-seeded species
(Castanea and Fagus), whereas disease was most
important for the small-seeded species (Prunus,
Cornus, Magnolia and Fraxinus). This difference is
probably due to the preference of rodents for large
seeds (Jensen 1985; Seiwa and Kikuzawa 1996;
Ostfeld et al. 1997). Predation usually occurred
during the early growing season, when seed reserves
remain in the cotyledons, prior to disease attack.
Furthermore, large seeds may also produce sturdy
seedlings that are more disease resistant, resulting in
little damage from pathogens.
The manner of attack (distance or density dependency) did not differ between the two primary killing
agents (disease and rodents), both of which usually
exhibited distance dependence or a combined effect
of both distant and density dependence in disease
attack for two species (Cornus and Fraxinus). These
189
results are in accordance with previous studies on
disease attacks (Packer and Clay 2000; reviewed by
Gilbert 2005; Bell et al. 2006; Seiwa et al. 2008), but
not rodent predation (Tomita et al. 2002). For
rodents, particularly wood mice (Apodemus spp.),
the close proximity (1 m) of the quadrats between
high- and low-density treatments may have promoted
seed consumption even at low density, thus reducing
the effects of density. Because the experiment was
conducted in a non-mast year for the large-seeded
study species, seed predation in low density quadrats
may also have been promoted by the lack of naturally
regenerating seedlings. To clarify the manner of
attack by rodents, temporal fluctuations in seed crops
and rodents should be considered. In contrast,
distance-dependent disease attacks are primarily
caused by crowding of disseminated seeds and
seedlings beneath conspecific adults, which enhances
cultures of soil microbial communities (Bever 1994;
Packer and Clay 2000; Hood et al. 2004). In forest
communities, a greater proportion of the seed crop is
deposited below canopies of bird- or wind-dispersed
species (Houle 1992; Clark et al. 2005). Approximately 80% of seeds were disseminated beneath a
conspecific canopy without bird dispersal in C.
controversa (Masaki et al. 1994).
Adult trees may also serve as leaf disease ‘incubators’, because diseases in the canopy are often
shared by juveniles beneath conspecific adults (Gilbert 1995; Hood et al. 2004; Gallery et al. 2007). In
this study, the foliar disease ‘‘blight’’ observed in
seedlings was also observed in leaves of conspecific
adults in Cornus (e.g. zonate leaf blight, grey mould),
Prunus (e.g. Monilinia blight, angular leaf spot) and
Fraxinus (brown leaf spot; Yamazaki and Seiwa,
unpublished data). Because the majority of leaves fall
in the vicinity of adults, infected leaves and spores
accumulate beneath them. Foliar diseases strongly
reduce the photosynthetic area of leaves and frequently destroy constructive tissues such as stems
(Gilbert 1995). Such negative effects of foliar
diseases, together with soil-borne pathogens, such
as damping-off diseases, resulted in higher seedling
mortality beneath conspecific adults. Subsequent
studies of foliar disease in both seedlings and
conspecific adult trees warrant further work following
individual diseases.
190
Host specificity of predominant diseases
In our experiment, four predominant symptoms
(damping-off, blight, rot, powdery mildew) were
observed in the dead seedlings of the five tree species
that exhibited distant-dependent mortality due to the
disease. For each disease symptom, the higher
infectivity and/or virulence beneath compared with
far from conspecific adults strongly indicated the host
specificity of the disease attack. In blight of Cornus
and damping-off of Fraxinus, higher infectivity was
observed beneath relative to far from conspecific
adults, whereas virulence did not differ between
distances. These patterns possibly suggest that higher
seedling mortality due to these diseases beneath
conspecific adults is likely due to greater abundance
and/or infective activity of the pathogens beneath
adult trees. In blights of both Prunus and Magnolia,
infectivity did not differ between distances, but
virulence was significantly higher beneath compared
with far from adults. These results potentially indicate that pathogens causing blight were ubiquitous,
but virulence was stronger beneath conspecific adults,
resulting in higher seedling mortality beneath them.
In rot of Prunus and damping-off of Magnolia, both
infectivity and virulence were higher beneath compared with far from conspecific adults, suggesting
that combined effects of pathogenicity caused higher
seedling mortality beneath the adults. In Prunus,
Magnolia and Fraxinus, more than two disease
symptoms exhibited significant differences in pathogenicity between distances. Even though each disease
attacks host seedlings independently, their combination may synergistically affect seedling mortality,
resulting in distance-dependent seedling mortality for
these study species.
Furthermore, fungal species of Colletotrichum,
Phoma, Fusarium, Cylindrocarpon, Cladosporium
and Alternaria, which were isolated from dead
seedlings infected with damping-off, blight and rot,
were observed in most of the eight species studied
(Appendix 3). These fungal species are considered
facultative pathogens, which originally exhibited
wide host ranges (Agrios 1997; Horst 2001).
Although little is known of the causal relationships
between infectivity and host mortality in most of the
individual fungal species, particularly for host tree
species (but see Sahashi et al. 1995; Packer and
A.G. Van der Valk (ed.)
Clay 2000; Augspurger and Wilkinson 2007; Seiwa
et al. 2008), the observed distant-dependent attacks
by the predominant diseases (including several
fungi) may suggest that generalist pathogens with
broad host ranges may cause seedling mortality in a
host-specific manner in this forest community.
Recently, Sicard et al. (2007) found that both
infectivity and the degree of leaf damage by the
pathogen Colletotrichum lindemuthianum differed
among individual host populations and among host
plants, because the pathogens were adapted to the
local genotypes of the host plant. Our results,
together with evidence of local adaptation (Sicard
et al. 2007) and negative feedback (Packer and Clay
2004; Kotanen 2007), may suggest that some soilborne pathogens are ubiquitous, but infectiousness
and virulence of the pathogens are frequently higher
beneath conspecific adults, because generalist pathogens sometimes adapt specifically to their local host
populations (Bever 1994; Bever et al. 1997; Mills
and Bever 1998; Lively and Dybdahl 2000). However, we compared pathogenicity between distances
for predominant disease symptoms (including several fungi; Appendix 3), instead of individual fungi.
To clarify the host specificity of individual pathogens, further experimental studies that include
inoculation trials, fungicide experiments and molecular identification must be conducted (Gilbert 2005).
In conclusion, our study clearly revealed that
biotic natural enemies (diseases and rodents) strongly
influence seedling mortality in a distance-dependent
manner for six of eight tree species co-occurring in a
temperate forest. Comparisons of pathogenicity of
diseases between distances from conspecific adults
(i.e. four predominant disease symptoms) indicated
that both soil-borne and foliar diseases may affect
seedlings in a host-specific manner. These traits
strongly suggest that the Janzen–Connell mechanism
is important for maintaining local plant diversity in
temperate as well as tropical forests.
Acknowledgements We are very grateful to Arnold Gerard
van der Valk, Owen Lewis and Rachel Gallery for their
valuable comments on the manuscript. We thank T. Miyamoto
for providing seeds of F. crenata. We thank many members of
the Laboratory of Forest Ecology, Tohoku University, for help
with the fieldwork. This research was funded by the Ministry of
Education, Culture, Sports, Science and Technology of Japan
(No. 17380087: KS).
Forest Ecology
191
Appendices
Appendix 1 Number of seedlings emerged at two densities (high and low) and at two distances (beneath and far) from conspecific
adult trees for the eight tree species studied
Seedling density (m-2, ± SE)
Species
Beneath
Far
High
Low
High
Low
Prunus grayana
403.2 ± 28.5
48.7 ± 20.8
507.9 ± 67.0
57.6 ± 9.2
Cornus controversa
295.2 ± 49.6
59.3 ± 30.8
312.7 ± 63.8
68.3 ± 46.4
Magnolia obovata
143.9 ± 61.2
161.5 ± 42.0
15.9 ± 8.5
Fraxinus lanuginosa
520.6 ± 36.1
117.5 ± 4.8
511.6 ± 84.7
118.0 ± 8.4
21.2 ± 14.8
Acer mono
203.2 ± 51.4
22.2 ± 4.8
222.8 ± 46.7
46.9 ± 6.4
Castanea crenata
Fagus crenata
80.4 ± 16.4
265.6 ± 29.6
13.8 ± 2.8
37.0 ± 17.3
79.4 ± 8.7
255.0 ± 8.5
19.0 ± 4.3
50.8 ± 6.6
Quercus serrata
127.0 ± 14.9
25.4 ± 2.8
124.9 ± 5.7
17.0 ± 1.9
Appendix 2 The number of seedlings died and the number of seedlings for fungal isolations at two distances (beneath and far) from
conspecific adults for the eight tree species
Host species
Beneath
Far
Number of dead Number of seedlings
seedlings
for fungal isolations
Percentage
Number of dead Number of seedlings
seedlings
for fungal isolations
Percentage
11.0
Prunus grayana
246
20
8.1
209
23
Cornus controversa
257
25
9.7
300
22
7.3
Magnolia obovata
88
9
10.2
4
1
25.0
Fraxinus lanuginosa 345
12
3.5
482
2
0.4
Acer mono
41
3
7.3
47
5
10.6
Castanea crenata
Fagus crenata
11
11
3
7
27.3
63.6
15
24
2
2
13.3
8.3
Quercus serrata
20
2
10.0
44
5
11.4
Percentages are defined as the proportion of seedlings for fungal isolations
192
Appendix 3 Description of disease symptoms and the isolated fungal genera at two distances (beneath and far) from conspecific adult trees for each tree species studied
Symptom
Genus of fungi
Damping-off Colletotrichum
Acer mono
Fraxinus
lanuginosa
Magnolia
obovata
Cornus
controversa
Prunus
grayana
Castanea
crenata
Qurcaus
serrata
Fagus
crenata
Beneath
(5)
Far
(15)
Beneath
(7)
Far
(5)
Beneath
(3)
Far
(1)
Beneath
(3)
Far
(1)
Beneath Far Beneath
(5)
(5) (6)
Far
(2)
Beneath Far Beneath
(7)
(5) (2)
Far
(4)
?
?
?
?
-
?
?
?
?
?
-
?
-
?
?
?
?
?
?
?
-
-
?
?
?
?
?
?
-
-
?
Fusarium
?
?
-
?
?
-
-
-
?
?
-
-
?
-
-
-
Cylindrocarpon
?
?
?
?
?
-
?
-
-
?
-
-
?
-
-
-
Cladosporium
-
?
-
?
?
-
-
?
?
?
?
?
?
?
-
-
Alternaria
-
?
-
?
-
-
-
-
?
?
?
-
?
-
?
-
Cylindrocladium
-
-
-
-
-
-
-
-
?
-
?
-
-
-
?
?
Trichoderma
?
?
?
?
-
-
-
-
-
-
-
-
-
-
?
-
Mucor
?
?
?
?
-
-
?
-
-
-
-
?
-
-
-
-
Idriella
-
-
-
-
-
-
-
-
-
-
?
-
-
-
?
?
Arthrinium
-
-
-
-
-
-
-
-
-
?
-
?
-
-
-
?
Rhizoctonia
-
?
-
-
?
-
-
-
-
-
?
-
-
-
-
-
Epicoccum
Pythium
?
-
?
-
?
-
-
-
-
-
?
-
-
-
-
-
-
-
-
Penicillium
-
-
-
-
-
-
-
-
-
-
?
-
-
-
-
-
Pestalotiopsis
-
-
-
-
-
-
-
-
-
?
-
-
-
-
-
-
Gliocephalotrichum -
?
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Scytalidium
-
?
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Mortierella
-
?
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Monochaetia
-
-
-
-
-
-
-
-
-
-
?
-
-
-
-
-
Polyscytalum
-
-
-
-
-
-
-
-
-
-
?
-
-
-
-
-
Dictyochaeta
-
-
-
-
-
-
-
-
-
-
-
-
-
-
?
-
Unknown
?
?
-
-
?
-
-
-
?
?
?
-
?
?
?
?
A.G. Van der Valk (ed.)
?
Phoma
Forest Ecology
Appendix 3 continued
Symptom
Blight and die-back
Genus of fungi
Prunus grayana
Cornus controversa
Magnolia obovata
Fraxinus lanuginosa
Beneath
Far
Beneath
Far
Beneath
Far
Beneath
Far
(13)
(5)
(11)
(10)
(5)
(3)
(3)
(3)
Colletotrichum
Phoma
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
Cladosporium
?
?
?
?
-
-
?
-
Alternaria
?
?
?
?
-
-
?
-
Cylindrocarpon
?
?
-
?
?
-
?
Cylindrocladium
?
?
?
?
?
?
-
-
Fusarium
?
?
?
?
-
-
-
-
Trichoderma
?
?
-
?
-
-
-
-
Idriella
-
-
?
?
-
-
-
-
Epicoccum
-
?
-
?
-
-
-
-
Clonostachys
-
?
-
?
-
-
-
-
Mucor
?
-
-
-
-
-
-
-
Botrytis
-
-
?
-
-
-
-
-
Pythium
-
-
?
-
-
-
-
-
Aureobasidium
-
?
?
-
-
-
-
-
Penicillium
?
?
-
-
-
-
-
-
Curvularia
Gliocladium
?
?
-
-
-
-
-
-
-
Macrophoma
?
-
-
-
-
-
-
-
Cladosporium
-
-
?
-
-
-
-
-
Polyscytalum
-
-
-
?
-
-
-
-
Unknown
?
-
-
-
-
-
-
?
193
194
A.G. Van der Valk (ed.)
?, the presence of fungi isolated from diseased seedling; -, no isolation
Numerals in parenthesis are number of seedlings for fungal isolations
?
Unknown
?
Epicoccum
?
Aureobasidium
?
Alternaria
?
Botrytis
?
Idriella
?
?
Cladosporium
?
?
Phoma
?
?
Colletotrichum
Rot
Appendix 3 continued
(2)
(3)
References
Agrios GN (1997) Plant pathology, 4th edn. Academic Press,
London
Akashi N (1997) Dispersion pattern and mortality of seeds and
seedlings of Fagus crenata Blume in a cool temperate
forest in western Japan. Ecol Res 12:159–165. doi:
10.1007/BF02523781
Augspurger CK (1984) Seedling survival of tropical tree species: interactions of dispersal distance, light-gaps, and
pathogens. Ecology 65:1705–1712. doi:10.2307/1937766
Augspurger CK, Kelly CK (1984) Pathogen mortality of
tropical tree seedlings: experimental studies of the effects
of dispersal distance, seedling density, and light conditions. Oecologia 61:211–217. doi:10.1007/BF00396763
Augspurger CK, Wilkinson HT (2007) Host specificity of
Pathogenic Pythium species: implications for tree species
diversity. Biotropica 39:702–708. doi:10.1111/j.17447429.2007.00326.x
Bell T, Freckleton RP, Lewis OT (2006) Plant pathogens drive
density-dependent seedling mortality in a tropical tree. Ecol
Lett 9:569–574. doi:10.1111/j.1461-0248.2006.00905.x
Bever JD (1994) Feedback between plants and their soil
communities in an old field community. Ecology
75:1965–1977. doi:10.2307/1941601
Bever JD, Westover KM, Antonovics J (1997) Incorporating
the soil community into plant population dynamics: the
utility of the feedback approach. J Ecol 85:561–573. doi:
10.2307/2960528
Burdon JJ (1991) Fungal pathogens as selective forces in plant
populations and communities. Aust J Ecol 16:423–432.
doi:10.1111/j.1442-9993.1991.tb01072.x
Clark DA, Clark DB (1984) Spacing dynamics of a tropical
rainforest tree, evaluation of the Janzen–Connell model.
Am Nat 124:769–788. doi:10.1086/284316
Clark CJ, Poulsen JR, Bolker BM, Connor EF, Parker VT (2005)
Comparative seed shadows of bird-, monkey-, and winddispersed trees. Ecology 86:2684–2694. doi:10.1890/04-1325
Coley PD, Barone JA (1996) Herbivory and plant defenses in
tropical forests. Annu Rev Ecol Syst 27:305–335. doi:
10.1146/annurev.ecolsys.27.1.305
Connell JH (1971) On the role of natural enemies in preventing
competitive exclusion in some marine animals and in rain
forest trees. In: van der Boer PJ, Gradwell GR (eds)
Dynamics of populations. Proceeding of the advanced
study institute, Osterbeek, 1970. Centre for Agricultural
Publication and Documentation, Wageningen, pp 298–312
Dinoor A, Eshed N (1984) The role and importance of pathogens
in natural plant communities. Annu Rev Phytopathol
22:443–466. doi:10.1146/annurev.py.22.090184.002303
Dybdahl MF, Storfer A (2003) Parasite local adaptation: Red
Queen versus Suicide King. Trends Ecol Evol
18:523–530. doi:10.1016/S0169-5347(03)00223-4
Frazer GW, Canham CD, Lertzman KP (1999) Gap light
analyzer (GLA), Version 2.0. Copyright 1999: Simon
Fraser University, Burnaby, British Columbia, and the
Institute of Ecosystem Studies, Millbrook, New York
Freckleton RP, Lewis OT (2006) Pathogens, density dependence and the coexistence of tropical trees. Proc R Soc
Lond 273:2909–2916. doi:10.1098/rspb.2006.3660
Forest Ecology
Gallery RE, Dalling JW, Wolfe BT, Arnold AE (2007) The
influence of seed source, habitat and fungi on Cecropia
seed survival in two Neotropical forests. In: Dennis A,
Green R, Schupp E, Westcott D (eds) Theory and its seed
dispersal, application in a changing world. CABI Press,
Wallingford, UK, pp 479–498
Garcıa-Guzman G, Dirzo R (2001) Patterns of leaf-pathogen
infection in the understory of a Mexican rain forest:
incidence, spatiotemporal variation, and mechanisms of
infection. Am J Bot 88:634–645. doi:10.2307/2657063
Gilbert GS (1995) Rain forest plant diseases: the canopy—
understory connection. Selbyana 16:75–77
Gilbert GS (2002) Evolutionary ecology of plant diseases in
natural ecosystems. Annu Rev Phytopathol 40:13–43. doi:
10.1146/annurev.phyto.40.021202.110417
Gilbert GS (2005) Dimensions of plant disease in tropical
forests. In: Burslem DFRP, Pinard MA, Hartley SE (eds)
Biotic interactions in the tropics. Cambridge University
Press, Cambridge, pp 141–164
Gilbert GS, Harms KE, Hamill DN, Hubbell SP (2001) Effects
of seedling size, El Niño drought, seedling density, and
distance to nearest conspecific adult on 6-year survival of
Ocotea whitei seedlings in Panamá. Oecologia 127:509–
516. doi:10.1007/s004420000616
Harms KE, Wright SJ, Calderon O, Hernandez A, Herre EA
(2000) Pervasive density-dependent recruitment enhances
seedling diversity in a tropical forest. Nature 404:493–
495. doi:10.1038/35006630
Hille Ris Lambers J, Clark JS, Beckage B (2002) Densitydependent mortality and the latitudinal gradient in species
diversity. Nature 417:732–735. doi:10.1038/nature00809
Hood LA, Swaine MD, Mason PA (2004) The influence of
spatial patterns of damping-off disease and arbuscular
mycorrhizal colonization on tree seedling establishment in
Ghanaian tropical forest soil. J Ecol 92:816–823. doi:
10.1111/j.0022-0477.2004.00917.x
Horst RK (2001) Wescott’s plant disease handbook, 6th edn.
Kluwer, Norwell
Houle G (1992) Spatial relationship between seed and seedling
abundance and mortality in a deciduous forest of northeastern North America. J Ecol 80:99–108. doi:10.2307/
2261066
Janzen DH (1970) Herbivores and the number of tree species in
tropical forests. Am Nat 104:501–528. doi:10.1086/
282687
Jensen TS (1985) Seed–predator interactions of European
beech (Fagus silvatica L.) and forest rodents, Clethrionomys glareolus and Apodemus flavicollis. Oikos 44:149–
156. doi:10.2307/3544056
Jones RH, Sharitz RR, Dixon PM, Segal DS, Schneider RL
(1994) Woody plant regeneration in four floodplain forests. Ecol Monogr 64:345–367. doi:10.2307/2937166
Kirchner JW, Roy BA (2002) Evolutionary implications of
host-pathogen specificity: fitness consequences of pathogen virulence traits. Evol Ecol Res 4:27–48
Kotanen PM (2007) Effects of fungal seed pathogens under
conspecific and heterospecific trees in a temperate forest.
Can J Bot Rev Can Bot 85:918–925. doi:10.1139/B07-088
Lively CM, Dybdahl MF (2000) Parasite adaptation to locally
common host genotypes. Nature 405:679–681. doi:
10.1038/35015069
195
Maeto K, Fukuyama K (1997) Mature tree effect of Acer mono
on seedling mortality due to insect herbivory. Ecol Res
12:337–343. doi:10.1007/BF02529463
Marquis RJ (2005) Impacts of herbivores on tropical plant
diversity. In: Burslem DFRP, Pinard MA, Hartley SE
(eds) Biotic interactions in the tropics. Cambridge University Press, Cambridge, pp 328–346
Masaki T, Nakashizuka T (2002) Seedling demography of
Swida controversa: effect of light and distance to conspecifics. Ecology 83:3497–3507
Masaki T, Kominami Y, Nakashizuka T (1994) Spatial and
seasonal patterns of seed dissemination of Cornus controversa in a temperate forest. Ecology 75:1903–1910.
doi:10.2307/1941595
Mills KE, Bever JD (1998) Maintenance of diversity within
plant communities: soil pathogens as agents of negative
feedback. Ecology 79:1595–1601
Ostfeld RS, Manson RH, Canham CD (1997) Effects of rodents
on survival of tree seeds and seedlings invading old fields.
Ecology 78:1531–1542
Packer A, Clay K (2000) Soil pathogens and spatial patterns of
seedling mortality in a temperate tree. Nature 404:278–
281. doi:10.1038/35005072
Packer A, Clay K (2003) Soil pathogens and Prunus serotina
seedling and sapling growth near conspecific trees. Ecology 84:108–119. doi:10.1890/0012-9658(2003)084[0108:
SPAPSS]2.0.CO;2
Packer A, Clay K (2004) Development of negative feedback
during successive growth cycles of black cherry. Proc R
Soc Lond 271:317–324. doi:10.1098/rspb.2003.2583
Peters HA (2003) Neighbour-regulated mortality: the influence
of positive and negative density dependence on tree
populations in species-rich tropical forests. Ecol Lett
6:757–765. doi:10.1046/j.1461-0248.2003.00492.x
Reinhart KO, Royo AA, Van der Putten WH, Clay K (2005)
Soil feedback and pathogen activity in Prunus serotina
throughout its native range. J Ecol 93:890–898. doi:
10.1111/j.1365-2745.2005.01028.x
Sahashi N, Kubono T, Shoji T (1995) Pathogenicity of Colletotrichum dematium isolated from current-year beech
seedling exhibiting damping-off. Eur J Forest Pathol
25:145–151. doi:10.1111/j.1439-0329.1995.tb00329.x
Schafer M, Kotanen P (2004) Impacts of naturally-occurring
soil fungi on seeds of meadow plants. Plant Ecol 175:19–
35. doi:10.1023/B:VEGE.0000048096.00772.23
Seiwa K (1998) Advantages of early germination for growth and
survival of seedlings of Acer mono under different overstorey phenologies in deciduous broad-leaved forests. J
Ecol 86:219–228. doi:10.1046/j.1365-2745.1998.00245.x
Seiwa K, Kikuzawa K (1996) Importance of seed size for the
establishment of seedlings of five deciduous broad-leaved
tree species. Vegetation 123:51–64. doi:10.1007/BF000
44887
Seiwa K, Miwa Y, Sahashi N, Kanno H, Tomita M, Ueno N,
Yamazaki M (2008) Pathogen attack and spatial patterns
of juvenile mortality and growth in a temperate tree,
Prunus grayana. Can J Res 38:2445–2454. doi:10.1139/
X08-084
Sicard D, Pennings PS, Grandclement C, Acosta J, Kaltz O,
Shykoff JA (2007) Specialization and local adaptation of
a fungal parasite on two host plant species as revealed by
196
two fitness traits. Evol Int J Org Evol 61:27–41. doi:
10.1111/j.1558-5646.2007.00003.x
Silman MR, Terborgh J, Kiltie R (2003) Population regulation
of a dominant rain forest tree by a major seed predator.
Ecology 84:431–438. doi:10.1890/0012-9658(2003)084
[0431:PROADR]2.0.CO;2
Streng DR, Glitzenstein JS, Harcombe PA (1989) Woody
seedling dynamics in an east Texas floodplain in forest.
Ecol Monogr 59:177–204. doi:10.2307/2937285
Terabaru M, Yamazaki M, Kanou K, Suyama Y, Seiwa K
(2004) Influence of topographic positions on tree distribution patterns in a temperate broad-leaved deciduous
forest. Bull Integr Field Sci Cent 20:21–26 (In Japanese)
Tomita M, Hirubuki Y, Seiwa K (2002) Post-dispersal changes
in the spatial distribution of Fagus crenata seeds. Ecology
83:1560–1565
Wada N, Murakami M, Yoshida K (2000) Effects of herbivorebearing adult trees of the oak Quercus crispula on the
A.G. Van der Valk (ed.)
survival of their seedlings. Ecol Res 15:219–227. doi:
10.1046/j.1440-1703.2000.00343.x
Webb CO, Gilbert GS, Donoghue MJ (2006) Phylodiversitydependent seedling mortality, size structure, and disease
in a Bornean rain forest. Ecology 87:S123–S131. doi:
10.1890/0012-9658(2006)87[123:PSMSSA]2.0.CO;2
Wills C, Condit R, Foster RB, Hubbell SP (1997) Strong
density- and diversity-related effects help to maintain tree
species diversity in a neotropical forest. Proc Natl Acad
Sci USA 94:1252–1257. doi:10.1073/pnas.94.4.1252
Wright SJ (2002) Plant diversity in tropical forests: a review of
mechanisms of species coexistence. Oecologia 130:1–14
Wyatt JL, Silman MR (2004) Distance-dependence in two
Amazonian palms: effects of spatial and temporal variation in seed predator communities. Oecologia 140:26–35.
doi:10.1007/s00442-004-1554-y
Response of native Hawaiian woody species to lava-ignited
wildfires in tropical forests and shrublands
Alison Ainsworth Æ J. Boone Kauffman
Originally published in the journal Plant Ecology, Volume 201, No. 1, 197–209.
DOI: 10.1007/s11258-008-9538-3 Springer Science+Business Media B.V. 2008
Abstract Wildfires are rare in the disturbance
history of Hawaiian forests but may increase in
prevalence due to invasive species and global climate
change. We documented survival rates and adaptations facilitating persistence of native woody species
following 2002–2003 wildfires in Hawaii Volcanoes
National Park, Hawaii. Fires occurred during an El
Niño drought and were ignited by lava flows. They
burned across an environmental gradient occupied by
two drier shrub-dominated communities and three
mesic/wet Metrosideros forest communities. All the
19 native tree, shrub, and tree fern species demonstrated some capacity of postfire persistence. While
greater than 95% of the dominant Metrosideros trees
were top-killed, more than half survived fires via
basal sprouting. Metrosideros trees with diameters
[20 cm sprouted in lower percentages than smaller
trees. At least 17 of 29 native woody species
colonized the postfire environment via seedling
A. Ainsworth (&)
Department of Fisheries and Wildlife, Oregon State
University, Corvallis, OR 97331, USA
e-mail: aliainsworth@hotmail.com
Present Address:
A. Ainsworth
Division of Forestry and Wildlife, State of Hawaii,
19 E. Kawili St., Hilo, Hawaii 96720, USA
J. Boone Kauffman
Institute of Pacific Islands Forestry, USDA Forest Service,
60 Nowelo St., PO Box 4370, Hilo, Hawaii 96720, USA
establishment. Although the native biota possess
adaptations facilitating persistence following wildfire,
the presence of highly competitive invasive plants and
ungulates will likely alter postfire succession.
Keywords Disturbance Dodonaea viscosa
Fire adaptations Hawaii Metrosideros
polymorpha Sprouting
Introduction
Wildfires have a dramatic effect on Hawaiian landscapes (D’Antonio et al. 2000). Yet, little is known on
the fire history of the Hawaiian Islands and its role in
the evolution and development of Hawaiian ecosystems (Vogl 1969; Mueller-Dombois 1981, 2001; Smith
and Tunison 1992). Studies of sediment cores collected
in bogs and radiocarbon data from charcoal studies
indicated that wildfires have occurred in Hawaii prior
to European settlement (Mueller-Dombois 1981;
Smith and Tunison 1992; Burney et al. 1995). The
occurrence of natural ignition sources including lightning and volcanism (Vogl 1969; Tunison and Leialoha
1988) and continuous vegetation cover in many
ecosystems (Wagner et al. 1999) further suggests that
fire did occur historically and did influence the
disturbance history of Hawaiian ecosystems.
Although there is a poor fire record, the response
of native woody species to wildland fire provides
insights into historical fire patterns because
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_15
197
198
adaptations evolve within the context of each
ecosystem’s natural disturbance regime (Kauffman
1990). Species adaptations in ecosystems are linked
to their capacity to survive, establish, and reproduce
in the disturbance regime of their habitats (White and
Pickett 1985). Examples of traits that promote
survival of individuals following fire include: thick
bark, protected buds from dense leaf bases, and
sprouting from either epicormic or subterranean
tissues. Adaptations that facilitate establishment of
species or populations, but not the individual following fire include: fire-stimulated germination or
flowering, seed storage on plants (e.g., serotinous
cones), and wind-borne seeds (Kauffman 1990).
Many Hawaiian wet forest species possess characteristics frequently associated with long fire-return
intervals (e.g., thin bark, buried seeds requiring heat
or other disturbance to germinate, enhanced seedling
establishment on downed wood, sprouting). Basal
and epicormic sprouting following fire have been
observed for Metrosideros polymorpha, a dominant
Hawaiian tree species (Parman and Wampler 1977;
Hughes et al. 1991; Tunison et al. 1995; D’Antonio
et al. 2000). Another dominant native tree, Acacia
koa has the capacity to sprout following disturbance
(Tunison et al. 2001) and produces refractory seeds
capable of surviving in the soils for decades until
disturbance stimulates germination. The tree ferns
Cibotium glaucum and Sadleria cyathoides survive
and rapidly produce new fronds, presumably because
the meristematic tissues are protected by frond scales
(Smith and Tunison 1992).
Wind dispersal and capacity to establish on bare
substrate is a common adaptation that facilitates
invasion and establishment following disturbance
such as fire (Kauffman 1990). Metrosideros has
long-ranging and abundant wind-dispersed seeds
(Drake 1992; Hatfield et al. 1996). Seedling recruitment has been observed following wildfire in
Metrosideros-dominated wet forests (Tunison et al.
2001). In addition, the seeds of a dominant native
shrub species in Hawaiian ecosystems, Dodonaea
viscosa, were found to break dormancy following
exposure to heat (Hodgkinson and Oxley 1990) and
have also been found to germinate readily after fire
(Hughes et al. 1991; Shaw et al. 1997; D’Antonio
et al. 2000).
Although postfire response of many native Hawaiian species suggest that they may be adapted to
A.G. Van der Valk (ed.)
disturbance, the majority of studies have been limited
to the seasonally dry Metrosideros woodlands of
Hawaii Volcanoes National Park (Hughes et al. 1991;
Hughes and Vitousek 1993; Freifelder et al. 1998;
Ley and D’Antonio 1998; D’Antonio et al. 2000;
D’Antonio et al. 2001; Mack et al. 2001). Nonnative
grass invasions during the past century have led to a
dramatic increase in fire frequency and size in these
dry woodlands (Smith and Tunison 1992). Rapid
grass recovery or fine fuel re-accumulation following
fire (Hughes et al. 1991) coupled with drier, windier
microclimatic conditions (Freifelder et al. 1998) has
led to additional fires creating a grass/fire cycle
(D’Antonio and Vitousek 1992). Consequently, many
previously native-dominated woodlands have been
type converted to nonnative-dominated grasslands.
In contrast to the drier Hawaiian woodlands, few
recorded fires have occurred and no studies have been
conducted in the wetter Metrosideros forests with
understories dominated by herbaceous species and
tree ferns. The effects of fire are expected to differ
from those in the dry woodlands because of differences in fuels and microclimatic conditions despite
some similarity in species (e.g. Metrosideros).
Although fires may have been infrequent historically,
climate change, nonnative species invasions, and
increasing human ignition sources are likely to result
in more frequent larger fires in wet Hawaiian forests.
Naturally ignited wildfires during particularly strong
El Niño (ENSO), mediated droughts in 2002 and
2003 created an opportunity to examine fire effects in
relatively intact wet forests as well as adjacent
perturbed shrublands.
We hypothesized that native Hawaiian species
would persist following fire through individual survival or establish from propagules in the postfire
environment because these species evolved in a
landscape subjected to a wide array of infrequent
disturbance events (fires, volcanism, tropical storms,
etc). We measured the response of native Hawaiian
woody species and tree ferns for the first two years
following the 2003 Luhi and Panauiki lava-ignited
wildfires in five community types across an elevation/
moisture gradient in Hawaii Volcanoes National
Park. The specific objectives of this study were to:
(1) examine the postfire survival rates and describe
the mechanisms of persistence of native Hawaiian
trees, tree ferns, and shrubs partitioned by species and
size class; and (2) quantify native woody seedling
Forest Ecology
establishment across this elevation/moisture gradient
for the first two years following fire. Information
from this study should provide insights regarding
historic fire regimes in this area and native species’
response to fire, and will assist managers in evaluating the potential threat of fire to native forest
recovery in these unique communities.
Methods
Study site
This study was conducted at Hawaii Volcanoes
National Park on the Island of Hawaii (19200 1100 N
and 15570 2900 W). Elevation ranged from 350 m in
the relatively dry shrub-dominated communities to
825 m in wet forest communities; all communities
occurred within 5 km of each other. The study area
was located over a very steep precipitation gradient
from dry shrublands to wet forest and encompassed
four distinct Holdridge life zones: subtropical basal
moist forest, subtropical basal wet forest, subtropical
lower mountain moist forest, and subtropical lower
mountain wet forest (Tosi et al. 2001). Substrate
across the gradient consisted of young (400 to
750 yr-old) pahoehoe lava flows with minimal topographic relief (Trusdell et al. 2005). Two basic soil
types are present: the Kalapana series and the
Makaopuhi series. Both series are very shallow to
shallow soils formed in ash deposited over pahoehoe
lava with 2–10% slopes, and are classified as Medial,
ferrihydritic, isothermic, Lithic Udivitrands (well
drained), and Hapludands (poorly drained). The
shrub-dominated communities are on the Kalapana
dry phase soils, the mesic forest communities are on
Kalapana medial course sandy loam, and the wet
forest community is on Makaopuhi very paragravelly
muck (Jasper 2007).
Metrosideros polymorpha is the dominant forest
tree across the elevation gradient, but ranges in
percent canopy cover from \1% in the shrublands to
[60% in the mesic forests. The study area contained
five major plant communities (Ainsworth 2007). The
Dodonaea viscosa/Andropogon virginicus community (350–450 m) was dominated by native
Dodonaea in the shrub layer (*9,000 individuals/
ha) with the nonnative perennial bunch grass Andropogon dominating the understory. A few trees
199
(Metrosideros) were scattered across the landscape,
but were primarily restricted to lava uplifts where
past fires did not kill them. This community is located
within the mapped boundaries of past wildfires that
occurred in 1972 and 1992 and will be referred to
hereafter as the ‘‘Andropogon shrubland.’’
The Dodonaea/Nephrolepis multiflora shrub-dominated community (450–550 m) is also dominated by
Dodonaea in the shrub tier (*8,500 individuals/ha)
with the nonnative fern Nephrolepis multiflora dominating the understory. Similar to the Andropogon
shrubland, remnant Metrosideros trees are scattered
throughout this community. This community will be
referred to as the ‘‘Nephrolepis shrubland.’’ While the
tree component of these two communities is now
sparse due to the recent fires, historic photos indicate
that the area was characterized as relatively open
Metrosideros woodlands with scattered shrubs and a
mixed understory prior to the 1972 wildfire (Hawaii
Department of Land and Natural Resources 1966).
We sampled the Metrosideros/Nephrolepis multiflora forest community (550–640 m) which is
dominated by Metrosideros in the overstory (*700
individuals/ha) and the nonnative fern Nephrolepis
multiflora in the understory. This community will be
referred to as the ‘‘Nephrolepis forest.’’ The Metrosideros/Dicranopteris linearis forest community
(640–750 m) contains Metrosideros in the overstory
(*850 individuals/ha) and the native, mat forming
fern Dicranopteris in the understory. This community
will be referred to as the ‘‘Dicranopteris forest.’’ The
wettest and highest elevation community sampled
was the Metrosideros/ Cibotium glaucum forest
community (700–850 m). This community has an
open canopy overstory of Metrosideros (*500
individuals/ha) with a native tree fern Cibotium
glaucum midstory (*2,800/ha) and the native fern
Dicranopteris and nonnative grasses in the understory. This community will be referred to as the
‘‘Cibotium forest.’’
Fire history
Lava has been an ignition source in this area of the
Park at least from 1916 to present (Gassaway et al.
2002). Multiple fires have occurred in the coastal
lowlands in the last 30 years including a 1992 fire
which burned the Andropogon and Nephrolepis
shrublands. The Panauiki Fire (January, 2003)
200
A.G. Van der Valk (ed.)
reburned over half (860 ha) of Andropogon and
Nephrolepis shrublands between 60 and 670 m. In
May 2003, the Luhi fire burned over 75% (2,000 ha)
of the forested study area (National Park Service
2003). We established replicate plots (n = 5) in each
of the five vegetation communities in the areas
burned in the 2003 wildfires and unburned controls to
determine tree and tree fern responses and the postfire
seedling establishment.
recorded. For tree ferns (with fronds [50 cm long),
basal diameter, trunk length, and crown mortality
were recorded. From these data, percent crown
mortality and individual plant death were calculated
for all trees, tree ferns, and shrubs by species, and by
diameter size class (\10, 10–20, [20 cm) for the
dominant canopy (Metrosideros) and subcanopy
(Cibotium) species. Tree fern survival was also
analyzed by length class (\1, 1–2, [2 m).
Field methods
Analysis
In the burned areas for each of the sampled communities, we established five randomly located 20 9 50 m
permanent plots and measured the vegetation response
one (2004) and two (2005) years following fire. Sample
locations were selected based on composition and
structure, elevation, fire history, and proximity to
unburned sites. Unburned plots were sampled once—
two years (2005) following fire except the Nephrolepis
forest community which was sampled one year (2004)
following fire. We selected unburned plots in each
community type based on comparable elevation, and
vegetation composition and structure. Flowering plant
nomenclature followed that of Wagner et al. (1999), and
tree fern nomenclature followed that of Palmer (2003).
We sampled trees, tree ferns, shrubs, and woody
seedlings using a nested plot design. Individuals in the
burned plots were recorded as sprouts if the live
portion was attached to an older burned stem or root.
Tree seedlings, defined as individuals less than 1.3 m
tall, tree fern juveniles (those with fronds \50 cm
long), and shrubs were measured in six subplots
(1 9 5 m). Trees \10 cm diameter at breast height
(dbh; 1.3 m in height) and tree ferns \10 cm in
diameter at the point below past years frond shed were
measured in six 2 9 10 m subplots. Trees[10 cm dbh
and tree ferns with trunk diameters [10 cm were
measured in the entire 20 9 50 m plot. Species with
individuals that reached reproductive maturity within
the first two years following fire were recorded.
Quantitative measures recorded for all trees, tree
ferns, and shrubs included: plant mortality and mode
of sprouting (basal if it originated from subterranean
plant organs at the base of trees \50 cm above
ground, and epicormic if it originated from dormant
meristematic tissue in the bole or mainstems)
(Kauffman 1990). For trees ([1.3 m tall) diameter
at breast height (dbh) and crown mortality were
Native Hawaiian woody species and tree ferns were
grouped according to Rowe’s (1981) plant response
classification system which incorporates life history
traits of species and characteristics of fire regimes.
The five categories include: invaders (high dispersal
ability), evaders (long lived propagules stored in the
soil), avoiders (shade tolerant and slow invaders
following fire), resisters (thick bark or an anomalous
arrangement of mertistematic tissues that facilitates
fire survival), and endurers (capacity to sprout from
dormant surviving meristematic tissues) (Rowe
1981). Species often have multiple or changing
adaptations and therefore can fit into more than one
category. This universal life-form classification is a
useful way to examine species response to fire on a
per site basis because categories incorporate the
influence of environmental factors (Agee 1993).
The sampling unit used in analysis for all the
parameters was the 20 9 50 m plot. Average values
were calculated per plot and used in analysis for
vegetation parameters that were sampled in subplots
(e.g., seedlings, small trees, and tree ferns). Metrosideros percentage survival and population structure were
analyzed as two factor ANOVA’s with tree diameter
size class, community, and size class 9 community as
fixed effects. Differences among plant communities
were compared using Tukey’s multiple comparison
tests. Tree count data used to examine population
structure were log base 10 transformed (log ? 1) to
equalize variance. ANOVA and t-test analyses were
performed at an a = 0.10 in order to increase the
power (1 - b).
Differences in Cibotium survival among size
classes were compared using nonparametric tests
(Kruskal–Wallis Rank Test and Wilcoxon Rank Test
for pair-wise comparisons). Nonparametric tests were
also used to detect differences in native species
Forest Ecology
seedling density between treatments and years
(unburned vs. two year postfire) for each community.
Results
Sprouting response
Wildland fire resulted in greater than 95% crown
mortality of the dominant Metrosideros trees. There
were remarkably few unburned islands within the fire
perimeters. Despite the near complete crown mortality, many individuals of the native Hawaiian species
survived fire across the elevation gradient via vegetative sprouting. Nineteen tree, shrub, and tree fern
species were observed to have survived fire primarily
through basal spouting (Table 1). In addition to
sprouting from the base or root crown, scattered
individuals of three woody species, Dodonaea, Metrosideros, and Santalum paniculatum, were also
observed to have sprouted from epicormic tissues.
Postfire reproduction of surviving individuals can be
rapid as we observed fruiting or spore production of
individuals of all tree fern and shrub species within
the first two years following fire (Table 1). In
addition, two tree species, Hedyotis terminalis and
Santalum were also observed to be fruiting during the
second postfire year.
Despite high crown mortality, more than half
(57%) of the 911 individual Metrosideros trees
sampled in the burned communities survived fire
through basal sprouting. Survival significantly differed among diameter classes, where trees with larger
diameters ([20 cm) were less likely to sprout
following fire than those with smaller diameters
(P = 0.05; Fig. 1). The influence of plant size on
survival was most pronounced in the Dicranopteris
forest community, where [70% of the smaller trees
(\10 cm and 10–20 cm dbh) and only 38% of the
larger trees ([20 cm dbh) survived fire (P = 0.02).
The postfire survival of Metrosideros (all sizes
combined) differed among communities where survival was 71% in the Dicranopteris community, 48%
in the Cibotium community and 52% in the Nephrolepis forest community (P = 0.07). However, we found
no difference in survival among communities when
controlling for differences among size classes by using
a two factor ANOVA with size class and community
(P = 0.36; Fig. 1). Differences in survival among
201
communities were related to differences in Metrosideros population structure among communities
(P \ 0.01; Fig. 2). The population structure of the
Dicranopteris and Nephrolepis forest communities
was composed of smaller individuals with greater than
75% of Metrosideros trees in the smallest size class
(\10 cm dbh). In contrast, in the Cibotium forest
\30% of the trees were in the smallest size class and
over 50% in the largest size class ([20 cm dbh).
Tree ferns survived the fires in very high percentages ([86%; N = 1,195 Cibotium tree ferns
sampled). While existing foliage of tree ferns were
killed by fire, the individuals were observed to
rapidly refoliate from the apical meristems that were
apparently protected from lethal temperatures by the
bark, and leaf bases. Tree fern size affected rates of
survival where smaller sized individuals (\10 cm
diameter) had lower (42%) survival than the larger
classes (10–20 cm and [20 cm diameter; Fig. 3). In
the largest size class [90% of the individual’s
possessed live fronds one year postfire (P \ 0.01).
Although there was a difference (P \ 0.10) in
survival between the two larger diameter classes,
this difference is probably not ecologically meaningful considering that survival was extremely high
([90%) in both classes. For Cibotium individuals
[10 cm in diameter, no difference in survival was
detected among trunk length classes (\1 m, 1–2 m,
[2 m; P = 0.38).
Seedling response
There were a total of 29 native woody species and
tree ferns that were found on the entire study area,
and seedlings or juveniles of 17 were found to occur
in the postfire plots (Table 1). Seedlings of 10 species
were found only in burned areas while seedlings of
three species were found only in unburned sites and
seven species were found in both burned and
unburned sites. The majority of species found in the
burn following fire were present both as seedlings and
as sprouts including four tree, six shrub, and two tree
fern species. Of the five species present, only as
seedlings, two were tree species and three were shrub
species. For three shrub species, Clermontia hawaiiensis, Lythrum maritimum, and Sida fallax no
individuals (living or dead) were found in the study
area suggesting that these species either dispersed
into the area from outside or had been present only as
202
Table 1 Native woody
species and tree ferns that
survived fire and/or
established from seed in the
postfire environment
Mode of survival was
recorded as apical for tree
ferns and basal or epicormic
sprouting for tree and shrub
species. Asterisks denote
species with individuals that
fruited or flowered within
two years following fire
A.G. Van der Valk (ed.)
Species
Life form
Individual survival
Apical
Broussaisia arguta
Shrub
Cheirodendron trigynum
Tree
Cibotium glaucum
Cibotium menziesii
Tree fern
Tree fern
Basal
Postfire
Epicormic
Seedlings
X
X
X*
X*
X
Clermontia hawaiiensis
Shrub
Coprosma menziesii
Shrub
X*
Dodonaea viscosa
Shrub
X*
Hedyotis terminalis
Tree
X*
X
Ilex anomala
Tree
X
X
Leptecophylla tameiameiae
Shrub
X*
X
X
X
Lythrum maritimum
Shrub
Melicope clusiifolia
Tree
X
Melicope radiata
Tree
Tree
X
Myrsine lessertiana
Tree
X
Myrsine sandwicensis
Tree
X
Osteomeles anthyllidifolia
Shrub
X*
Pipturus albidus
Shrub
Psychotria hawaiiensis
Sadleria cyatheoides
Tree
Tree fern
Santalum paniculatum
Tree
X*
Scaevola chamissoniana
Shrub
X*
X
X
X
X*
X
X*
X
Sida fallax
Shrub
Vaccinium calycinum
Shrub
X*
Vaccinium reticulatum
Shrub
X*
propagules in the soil seed bank. Growth and
maturation from seed was rapid for Dodonaea,
Pipturus albidus, and Sida fallax. Individuals of
these three species were observed to have flowered
within the first two postfire (and post germination)
years.
There were no differences in shrub species
seedling densities when comparing between
unburned and burned sites for any community
(Table 2) except for the common shrub Dodonaea.
Dodonaea seedlings had dramatically higher densities in burned, compared to unburned sites. For
example, in the Andropogon shrubland, the second
postfire year Dodonaea density was 3,333/ha in the
unburned and 12,333/ha in burned sites (P = 0.16).
Similarly in the Nephrolepis shrubland, Dodonaea
seedlings densities were 5,733/ha in the unburned
sites, but densities were almost 8-fold greater in
X*
X
Metrosideros polymorpha
Total
X
X*
X*
X*
3
16
X
X
3
17
burned sites of this community (45,267/ha) two years
postfire (P = 0.01). In the forest communities,
Dodonaea was not encountered in the unburned sites
(Table 2), but did establish from seed in low densities
in the burned sites of the Nephrolepis (200/ha;
P = 0.07), Dicranopteris (67/ha; P = 0.43), and
Cibotium (267/ha; P = 0.18) forest communities.
The relatively low seedling densities of rare tree
and tree fern species did not differ between unburned
and burned sites within each community type
(Table 2). However, for the canopy dominant species,
Metrosideros, seedling density did differ between
burned and unburned sites within the three forest
communities. Only one seedling was found in the
unburned plots, but two years following fire many
more seedlings were found in the burned plots of the
Nephrolepis (667/ha; P = 0.06) and Dicranopteris
(267/ha; P = 0.07) forests (Fig. 4). Alternatively, in
Forest Ecology
<10 cm dbh
100
100
10-20 cm dbh
>20 cm dbh
80
Cibotium Survival (%)
Metrosideros survival (%)
203
60
40
20
c
80
60
a
40
20
0
0
NF
DF
3000
<10 cm dbh
10-20 cm dbh
2500
>20 cm dbh
2000
1500
1000
500
0
NF
DF
<10 cm
CF
Fig. 1 Postfire survival (%) via sprouting of Metrosideros
polymorpha individuals in the three sampled forest communities 12 months following fire at Hawaii Volcanoes National
Park (NF = Nephrolepis forest; DF = Dicranopteris forest;
and CF = Cibotium forest). Survival differed by diameter size
class (dbh) across the three forest communities (P = 0.04),
with the greatest mortality in the largest size class. Survival did
not differ among communities when controlling for size class
(P = 0.36) and no interaction was detected (Size 9 Community: P = 0.59). Data are means ?1 SE
Metrosideros trees (/ha)
b
CF
Fig. 2 The population structure of Metrosideros polymorpha
in the three different forest communities at Hawaii Volcanoes
National Park. Structure differed among the three forest
communities (P \ 0.01; NF = Nephrolepis forest, DF = Dicranopteris forest, CF = Cibotium forest). Data are means ±1
SE. A total of 911 trees were measured
the Cibotium forest seedling density was high (8,267/
ha) in the unburned sites, whereas in the burned sites
two years postfire seedling density was significantly
lower (733/ha; P = 0.09; Fig. 4). Juveniles of the
subcanopy dominant tree fern species, Cibotium
glaucum, were more abundant in the burned Cibotium
forest two years following fire (3,200/ha) than the
unburned forest (400/ha), but this difference was not
significant (P = 0.16) due to high variation among
plots.
10-20 cm
>20 cm
Fig. 3 Cibotium glaucum survival by diameter class in the
Cibotium forest community (P \ 0.01). The greatest mortality
was found to occur in the smallest size class. Data are
means ?1 SE. Letters indicate significant differences. A total
of 1195 tree ferns were measured
Discussion
Previous observations of postfire sprouting have been
made for many of the species present in this study
(Warshauer 1974; Parman and Wampler 1977; Tunison et al. 1994, 1995; D’Antonio et al. 2000). Yet
this study is the first study to quantify survival and
mortality rates and how survival differed by species,
size class, and plant communities across an elevation
gradient. These findings are particularly relevant
because context (e.g. community type, elevation, fire
characteristics, invasive species, post fire competition, etc) may greatly influence species response to
fire (Kauffman 1990; Kauffman and Martin 1990,
Sampaio and Kauffman 1993). Given the characteristics of the relatively wet climate and the relatively
low incidence of lightning, naturally occurring fires
were likely a rare occurrence in native Hawaiian wet
forests. However, we observed that native Hawaiian
woody plants in this study possessed several adaptations that facilitated fire survival.
The majority of species were characteristic of
‘‘endurers’’ (i.e. they were top-killed, but sprouted
after fire; Table 3). Nearly all tree and shrub sprouts
originated from the base or root crown where bark
tends to be the thickest and where soils provide a
great deal of insulation (Agee 1993). Tree ferns were
characteristic of ‘‘resistors’’ in that above ground
tissues and plant structures survived fire and refoliated within a few months following fire (Table 3).
Cibotium and Sadleria fern species have been
observed to recover following wildfires (1969–
1973) in the region (Warshauer 1974), but the high
204
A.G. Van der Valk (ed.)
Table 2 Native shrub, tree, and tree fern seedling densities in unburned (U) and burned (B) sites two years following fire for the
three forest communities (NF = Nephrolepis forest, DF = Dicranopteris forest, CF = Cibotium forest)
Species
NF
DF
U
CF
U
B
B
U
B
Coprosma menziesii
0
0
67 (67)
0
133 (133)
67 (67)
Dodonaea viscosa
0
* 200 (82)
0
67 (67)
0
267 (163)
Labordia hedyosmifolia
0
0
0
0
67 (67)
0
Leptecophylla tameiameiae
67 (67)
0
0
67 (67)
0
0
Pipturus albidus
0
0
0
67 (67)
0
67 (67)
Vaccinium calycinum
0
0
0
0
133 (82)
67 (67)
Vaccinium reticulatum
0
0
0
67 (67)
0
0
Cheirodendron trigynum
0
0
0
0
133 (82)
0
Hedyotis terminalis
Ilex anomala
0
0
0
0
0
0
67 (67)
67 (67)
0
0
0
133 (82)
Melicope clusiifolia
0
0
0
0
400 (245)
1000 (350)
Melicope radiate
0
0
0
0
0
67 (67)
Metrosideros polymorpha
67 (67)
* 667 (236)
0
* 267 (125)
8267 (3165)
* 733 (386)
Myrsine lessertiana
0
0
0
0
267 (125)
*0
Myrsine sandwicensis
0
0
133 (82)
0
0
0
Cibotium glaucum
0
0
0
333 (333)
400 (323)
3200 (1948)
Sadleria cyatheoides
0
0
0
67 (67)
0
0
Shrub species
Tree species
Tree fern species
Mean densities per hectare are reported with standard errors in parentheses. Asterisks denote significant differences (P \ 0.10)
between unburned and burned sites for each community
Metrosideros seedlings (/ha)
14000
12000
Unburned
Burned
10000
8000
6000
4000
*
2000
*
*
0
NF
DF
CF
Fig. 4 Metrosideros polymorpha seedling density in unburned
and burned sites two years following fire in the forest
communities (NF = Nephrolepis forest, DF = Dicranopteris
forest and CF = Cibotium forest). Means ?1 SE are reported
and asterisks denote significant differences between treatments
for each community
degree of survival ([90%) of these tree ferns has
not been previously quantified. In Australia, tree
ferns (Cyanea spp.) have also been reported to
recover rapidly following disturbance including fire
(Ough and Murphy 2004). We attribute the high
survival of tree ferns to their unique morphology in
which the meristematic tissues were protected from
lethal temperatures by the thick fibrous bark, live
tissues embedded within the main stem, and cover
by green frond bases. Survival of species with this
suite of morphological traits would be increased
with increasing trunk diameter as was observed in
this study.
In rare cases, some woody species including
Metrosideros, Santalum paniculatum, and Dodonaea
were also observed to regenerate from epicormic buds
along the stem and in the crown. Vegetative sprouting
from trunks of Metrosideros individuals has also been
observed to occur following tree-fall in unburned wet
forests (Drake and Mueller-Dombois 1993). Sprouting
from aboveground tissues provides a competitive
advantage in terms of rapid recovery of leaf area over
individuals sprouting from the base or establishing
from seed (Agee 1993). However, in this study,
Forest Ecology
Table 3 Native woody
species and tree fern plant
adaptations that facilitate
survival following fire in
tropical wet forests and
shrublands of Hawaii
Volcanoes National Park,
Hawaii
205
Adaptation Trait
Species
Resistors
Protected meristems Tree ferns (3):
Endurers
Sprouters
Cibotium glaucum, Cibotium menziesii, Sadleria cyathoides
Trees (8):
Hedyotis terminalis, Ilex anomala, Melicope clusiifolia,
Metrosideros polymorpha, Myrsine lessertiana, Myrsine
sandwicense, Psychotria hawaiiensis, Santalum paniculatum
Shrubs (8):
Broussaisia arguta, Coprosma menziesii, Dodonaea viscosa,
Leptecophylla tameiameiae, Osteomeles anthyllidifolia,
Scaevola chamissoniana, Vaccinium calycinum, Vaccinium
reticulatum
Invaders
Wind-borne seeds
Tree ferns (2):
Cibotium glaucum and Sadleria cyathoides
Trees (5):
Cheirodendron trigynum, Hedyotis terminalis, Ilex anomala,
Melicope clusiifolia, Metrosideros polymorpha
a
Shrubs (10):
Hughes and Vitousek
1993
Adaptations follow that of
Rowe (1981). Numbers in
parentheses are the total
number of species where
the specific trait was
observed
Clermontia hawaiiensis, Coprosma menziesii, Dodonaea
viscosa, Leptecophylla tameiameiae, Lythrum maritimum,
Melicope radiate, Pipturus albidus, Sida fallax, Vaccinium
calycinum, Vaccinium reticulatum
Evaders
Soil seed bank
epicormic sprouting was very rare (\1% of individuals), presumably because temperature extremes and
durations during the fire reached lethal levels to kill
epicormic buds present beneath the thin scaly bark of
the Metrosideros.
Rapid maturation and reproductive effort following fire is also an adaptation facilitating persistence
and establishment following fire (Kauffman 1990).
All the eight shrub and the three tree fern species that
survived fire vegetatively had individuals that
reached sexual maturity (i.e., were observed to be
fruiting or produced spores) within two years following fire (Table 1). In addition, two tree species
Hedyotis terminalis and Santalum also had sprouts
that reached sexual maturity during the second
postfire year. The majority of reproducing Santalum
originated from previously burned basal or root
sprouts, suggesting that flowering for this species
was fire enhanced.
Specific characteristics promoting survival in an
individual plant will often change with age (Kauffman
1990). Size class distribution of the dominant tree and
tree fern species influenced rates of survival. We
found that postfire survival rates of Metrosideros
Shrubs (2):
Dodonaea viscosaa and Osteomeles anthyllidifoliaa
decreased with increasing size. Tunison et al. (1995)
observed that Metrosideros survival following fire
appeared to be inversely correlated with tree size.
Conversely, in the dry Metrosideros woodlands
D’Antonio et al. (2000) found mortality to be
independent of size class. Although increased mortality with age and size has been documented in other
tree species such as Quercus spp. (Griffin 1980), the
loss of sprouting capacity usually signifies a shift in
the mode of survival to a general thickening of the
bark tissue to protect cambial and meristematic tissues
(Kauffman and Martin 1990). In this study, the stem
and crown of Metrosideros trees were extremely
sensitive because of the thin-barked nature of all
individuals in all size classes. This heat sensitivity of
aboveground tissues of Metrosideros was exemplified
by the near complete crown mortality and paucity of
epicormic sprouting following fire. Based on these
findings, predicted increases in fire frequency associated with climate change responses (e.g., warmer
temperatures, greater frequency, and severity of El
Niño events; IPCC 2007) may limit structural complexity and increase dominance of nonnative species
as has already occurred at lower elevations.
206
In addition to the individual species adaptations to
fire, context-specific factors such as pre-fire population, vegetation structure, weather conditions, fire
behavior, and fuel consumption may greatly influence
individual survival (Agee 1993; D’Antonio et al.
2000). Survival of Metrosideros was the greatest in
the Dicranopteris forest community (71%). Differences in Metrosideros population structure between
the Dicranopteris and Cibotium communities partially explain the lower survival in the Cibotium
community (48%). The Cibotium community appears
to be a later successional forest with a greater
proportion of trees in the largest size class (Fig. 2)
which had a lower number of surviving/sprouting
individuals. Many explanations as to why the trees
are larger in the Cibotium community as compared to
the Dicranopteris community are possible (e.g., older
substrate age; less historical disturbance from
humans, ungulates, and plants; moister microclimatic
conditions; and/or differences in soil ash content).
Total survival of Metrosideros in the Nephrolepis
forest community (52%) was also significantly lower
than in the Dicranopteris (71%) community
(P = 0.02), but population structures of these two
young forests were similar (Fig. 2). These data
suggest that the difference in survival may be related
to differences in fire severity (i.e., fuel consumption
or fire intensity). In the burned Dicranopteris forest
community, the quantity of unconsumed residual
surface fuels was greater than that of the other two
communities suggesting lower fuel consumption.
Lower fuel consumption would result in lower heat
flux around the base of the trees and could explain the
higher survival observed in this community. In the
coastal lowlands and seasonal submontane zones of
Hawaii Volcanoes National Park, D’Antonio et al.
(2000) observed that individual mortality was greater
for native Hawaiian woody species in sites where fuel
consumption was the highest.
In the burned landscape, about the same number of
native Hawaiian plant species were encountered as
seedlings (17) as surviving individuals from sprouts
(19). Two tree (Cheirodondron trigynum and Melicope radiata) and four shrub species (Clermontia
hawaiiensis, Lythrum maritimum, Pipturus albidus,
and Sida fallax) were observed to be obligate seeders
(i.e., no vegetative survival by sprouting). It is not
surprising that many native species can be classified
as ‘‘invaders’’ (i.e. those that disperse onto the site
A.G. Van der Valk (ed.)
following fire) according to Rowe’s (1981) classification (Table 3). Mueller-Dombois (1987) also
characterized Metrosideros as an early successional
tree species that colonized new volcanically derived
substrates or gaps created by tree falls while remaining dominant in mature wet forest communities.
Species with seeds that can survive fire in the soil
seed back or on individual plants are classified as
‘‘evaders’’ (Table 3). Some of the seedlings detected
following fire in this study may also have originated
from the soil seed bank, but the relative importance
of the seed bank is unknown. Hughes and Vitousek
(1993) found that some native Hawaiian shrub
species retain the capacity to germinate following
exposure to high-temperature treatments (120C for
5 min) including Osteomeles and Dodonaea.
Seedling establishment following fire differed
among species and among communities across the
elevation gradient of this study (Table 2). Fewer
Metrosideros seedlings were found in the burned
Cibotium forest as compared to those in the
unburned Cibotium forest (Table 2). Lower Metrosideros seedling density in the burned plots is not
likely because of limited seed source considering the
proximity to unburned forests and the dispersal
capabilities of these small wind-blown seeds. The
majority of Metrosideros seedlings found in the
unburned forest had established on moss covered
tree fern nurse logs. Although tree fern trunks were
abundant in the burned sites, mosses were burned
off during the fire, and conditions on the trunks
were presumably drier and less favorable for
Metrosideros establishment.
The short-term changes in Metrosideros seedling
density following fire in this study do not necessarily
indicate differences in the future forest because this
site is not likely seed limited as Metrosideros is well
suited for establishing in canopy gaps (Drake 1992;
Hatfield et al. 1996). Drastically lower seedling
density in the burned Cibotium forest (733/ha) may
be inconsequential because the density of 2-year old
seedlings was still much greater than the canopy
density. In addition, higher light conditions in the
burned forest presumably will result in a higher
likelihood of reaching the canopy than those seedlings in the unburned forest (Burton and MuellerDombois 1984).
In contrast to the Cibotium forest, Metrosideros
seedling recruitment appeared to be enhanced by fire
Forest Ecology
in the Nephrolepis and Dicranopteris forests where
seedling densities were greater in burned sites than
unburned sites (Table 2). Sampling other disturbed
sites, Restrepo and Vitousek (2001) found that
Metrosideros seedling establishment was greater on
recent (4 to 17-year old) landslides than undisturbed
mesic forests on the Island of Hawaii. Because the
Nephrolepis and Dicanopteris forest communities are
younger and lack woody and tree fern nurse logs, the
dense herbaceous fern understory in the unburned
sites may limit light and space for seedling establishment. Fire temporarily removed understory
barriers and allowed for seedling establishment. The
fact that, seedling densities remain lower in these
communities than the Cibotium forest even following
fire, supports the idea that seedling establishment
across the study area is facilitated by tree fern nurse
logs.
The native shrub Dodonaea viscosa was the only
woody species present in all the five communities
following fire. Dodonaea colonized the postfire
environment through dispersal (i.e. seeds are
enclosed in wind-borne bracts) and from the soil
seed bank where temperatures during fire scarify the
seeds (Hodgkinson and Oxley 1990). Therefore,
Dodonaea possesses traits characteristic of evaders,
invaders, and endurers (Table 3).
In the three forest communities of this study, no
Dodonaea shrubs were found in the unburned sites
and those in the burned sites had established from
seed presumably from offsite colonization (i.e.
neighboring shrubland communities). In the shrubland communities, however, Dodonaea individuals
both sprouted and established from seed. This species
had higher seedling densities in burned areas than in
unburned areas. It is also one of three native woody
species to reach sexual maturity from seed within two
postfire years. Other fire studies at Hawaii Volcanoes
National Park found that Dodonaea seedlings were
abundant in burned mesic forest and shrublands
(Warshauer 1974), submontane sites (Hughes et al.
1991; Hughes and Vitousek 1993; D’Antonio et al.
2000), and coastal lowlands (D’Antonio et al. 2000).
Many of the native Hawaiian woody species and
tree ferns in this study area possessed traits that
facilitated or ensured the persistence of individuals
and/or species following fire. As would be expected
the effects of fire differed among species; populations, and vegetation communities. The majority of
207
woody species demonstrated the capacity to sprout,
thus conferring these plants with advantages over
individuals that rely solely on seeds for establishment
in the postfire environment. It is unclear whether
these are evolutionary adaptations to fire or causal
adaptations of traits derived in response to other
disturbances common in the region (volcanism,
landslides, hurricanes, etc.). These adaptations may
not be sufficient to insure dominance of native
species in the future as the presence of invasive
plant and nonnative ungulate species coupled with
changes in climate may dramatically alter postfire
succession and dominance in these ecosystems.
Acknowledgments This study was supported by a grant from
the Joint Fire Science Program. Tim Tunison, Rhonda Loh, and
Flint Hughes provided guidance and advice throughout the
study. We thank the personnel of Hawaii Volcanoes National
Park for access and logistical support. This research would not
have been possible without the dedication of a number of
terrific field assistants including Mychal Tetteh, Lyndsay
Frady, Cristel Weitl, Liz Band, Jon Boehner, Wataru, Sally
Madden, and Tina Hartell. In addition, two anonymous referees
and Creighton Litton gave valuable inputs to this article.
References
Agee J (1993) Fire ecology of pacific northwest forests. Island
Press, Washington, DC
Ainsworth A (2007) Interactive influences of wildfire and
nonnative species on plant community succession in
Hawaii Volcanoes National Park. Oregon State University, Corvallis, OR
Burney D, DeCandido R, Burney L et al (1995) A holocene
record of climate change, fire ecology and human activity
from montane Flat Top Bog, Maui. J Paleolimnol 13:209–
217. doi:10.1007/BF00682765
Burton P, Mueller-Dombois D (1984) Response of Metrosideros polymorpha seedlings to experimental canopy
opening. Ecology 65:779–791. doi:10.2307/1938050
D’Antonio C, Vitousek P (1992) Biological invasions by exotic
grasses, the grass/fire cycle, and global change. Annu Rev
Ecol Syst 23:63–87
D’Antonio C, Hughes F, Vitousek P (2001) Factors influencing
dynamics of two invasive C4 grasses in seasonally dry
Hawaiian woodland. Ecology 82:89–104
D’Antonio C, Tunison J, Loh R (2000) Variation in the impact
of exotic grasses on native plant composition in relation to
fire across an elevation gradient in Hawaii. Austral Ecol
25:507–522
Drake D (1992) Seed dispersal of Metrosideros polymorpha
(Myrtaceae): a pioneer tree of Hawaiian lava flows. Am J
Bot 79:1224–1228. doi:10.2307/2445048
Drake D, Mueller-Dombois D (1993) Population development
of rain forest trees on a chronosequence of Hawaiian lava
flows. Ecology 74:1012–1019. doi:10.2307/1940471
208
Freifelder R, Vitousek P, D’Antonio C (1998) Microclimate
change and effect on fire following forest-grass conversion in seasonally dry tropical woodland. Biotropica
30:286–297. doi:10.1111/j.1744-7429.1998.tb00062.x
Gassaway L, Tunison J, Loh R (2002) Kupukupu Fire Hawaii
Volcanoes National Park: burned area emergency rehabilitation plan. Hawaii Volcanoes National Park, Hawaii
Griffin J (1980) Sprouting in fire-damaged valley oaks, Chews
Ridge, California. In: Ecology, management, and utilization of California Oaks. General Technical Report PSW-44.
Pacific Southwest Forest and Range Experiment Station,
Forest Service, U.S. Department of Agriculture:216–219
Hatfield J, Link W, Dawson D et al (1996) Coexistence and
community structure of tropical trees in a Hawaiian
montane rain forest. Biotropica 28:746–758. doi:
10.2307/2389061
Hawaii Department of Land and Natural Resources (1966)
Aerial photo series of Hawaii Island. Air Survey Hawaii,
Honolulu
Hodgkinson K, Oxley R (1990) Influences of fire and edaphic
factors on germination of the arid zone shrubs. Aust J Bot
38:269–279. doi:10.1071/BT9900269
Hughes F, Vitousek P (1993) Barriers to shrub reestablishment
following fire in the seasonal submontane zone of Hawaii.
Oecologia 93:557–563. doi:10.1007/BF00328965
Hughes F, Vitousek P, Tunison J (1991) Alien grass invasion
and fire in the seasonal submontane zone of Hawaii.
Ecology 72:743–746. doi:10.2307/2937215
IPCC (2007) Summary for policymakers. In: Parry M, Canziani J, Palutikof P, van der Linden P, Hanson C (eds)
Climate change 2007: impacts, adaptation and vulnerability. Contribution of working group II to the fourth
assessment report of the intergovernmental panel on
climate change. Cambridge University Press, Cambridge,
7–22
Jasper C (2007) Soil survey of Island of Hawaii. USDA-NRCS.
U.A. Gov. Print. Office, Washington, DC (in press)
Kauffman B (1990) Ecological relationships of vegetation and
fire in Pacific Northwest Forests. In: Walstad J, Radosevich S, Sandberg D (eds) Natural and prescribed fire in the
Pacific Northwest Forests. Oregon State University Press,
Corvallis
Kauffman JB, Martin RE (1990) Sprouting shrub response to
varying seasons and fuel consumption levels of prescribed
fire in Sierra Nevada mixed conifer ecosystems. For Sci
36:748–764
Ley R, D’Antonio C (1998) Exotic grass invasion alters
potential rates of N fixation in Hawaiian woodlands.
Oecologia 113:179–187. doi:10.1007/s004420050366
Mack M, D’Antonio C, Ley R (2001) Alteration of ecosystem
nitrogen dynamics by exotic plants: a case study of C4
grasses in Hawaii. Ecol Appl 11:1323–1335
Mueller-Dombois D (1981) Fire in tropical ecosystems. In:
Mooney H, Bonnicksen T, Christiansen N Jr et al. (eds)
Proceedings of the conference in fire regimes and ecosystem properties. USDA Forest Service General
Technical Report WO-26, Washington DC, National Park
Service
Mueller-Dombois D (1987) Forest dynamics in Hawaii. Trends
Ecol Evol 2:216–220
A.G. Van der Valk (ed.)
Mueller-Dombois D (2001) Biological invasion and fire in
tropical biomes. In: Galley K, Wilson T (eds) Proceedings
of the invasive species workshop: the role of fire in the
control and spread of invasive species. Fire conference
2000: the First National Congress on fire ecology, prevention, and management. Miscellaneous publication no.
11, Tall Timbers Research Station, Tallahassee, pp 112–
121
National Park Service (2003) Fire history map for Hawaii
Volcanoes National Park. Pacific West Region, Honolulu
Ough K, Murphy A (2004) Decline in tree-fern abundance after
clearfell harvesting. For Ecol Manage 199:153–163
Palmer D (2003) Hawaiı̀’s ferns and fern allies. University of
Hawaiı̀ Press, Honolulu
Parman T, Wampler K (1977) The Hilina Pali fire: a controlled
burn exercise. Technical Report 18, Cooperative National
Park Resources Studies Unit, University of Hawaii,
Honolulu
Restrepo C, Vitousek P (2001) Landslides, alien species, and
the diversity of a Hawaiian montane mesic ecosystem.
Biotropica 33:409–420
Rowe J (1981) Concepts of fire effects on plant individuals and
species. In: Wein R, Maclean D (eds) The role of fire in
northern circumpolar ecosystems. John Wiley and Sons,
New York
Sampaio E, Kauffman J (1993) Effects of different fire severities on coppicing of Caatinga vegetation in Serra
Talhada, PE, Brazil. Biotropica 25:452–460. doi:10.2307/
2388868
Shaw R, Castillo J, Laven R (1997) Impacts of wildfire on
vegetation and rare plants with the Kipuka Kalawamauna
endangered plant habitat area, Pohohakuloa Training
Area, Hawaii. In: Greenlee J (ed) Proceedings: first conference on fire effects on rare and endangered species and
habitats conference. International Association of Wildland
Fire, Coeur d’Alene, ID, pp 253–264
Smith C, Tunison J (1992) Fire and alien plants in Hawaii:
research and management implications for native ecosystems. In: Stone C, Smith S, Tunison J (eds) Alien plant
invasions in native ecosystems of Hawaii: management
and research. Cooperative National Park Resources
Studies Unit, Honolulu, pp 394–408
Tosi J, Watson V, Bolanos R (2001) Life zone map of Hawaii.
Based on the world life zone system of L.R. Holdridge.
UTM grid zone designation 40. Tropical Science Center,
San Jose, Costa Rica
Trusdell F, Wolfe E, Morris J (2005) Digital database of the
geologic map of the island of Hawaii. DS 144. U.S.
Geological Survey, Reston, VA
Tunison J, D’Antonio C, Loh R (2001) Fire and invasive plants
in Hawaii Volcanoes National Park. In: Galley K, Wilson
T (eds) Proceedings of the invasive species workshop: the
role of fire in the control and spread of invasive species.
Fire conference 2000: the First National Congress on fire
ecology, prevention, and management. Miscellaneous
publication no. 11, Tall Timbers Research Station,
Tallahassee, pp 122–131
Tunison J, Leialoha J (1988) The spread of fire in alien grasses
after lightning strikes in Hawaii Volcanoes National Park.
Hawaii Bot Soc 27:102–109
Forest Ecology
Tunison J, Leialoha J, Loh R et al (1994) Fire effects in the
coastal lowlands Hawaii Volcanoes National Park.
Cooperative National Park Resources Study Unit, Technical Report no. 88, Cooperative Agreement CA 8007-29004. University of Hawaii Press, Honolulu
Tunison J, Loh R, Leialoha J (1995) Fire effects in the submontane seasonal zone Hawaii Volcanoes National Park.
Cooperative National Park Resources Study Unit, Technical Report no. 97, Cooperative Agreement CA 8007-29004. University of Hawaii Press, Honolulu
Vogl R (1969) The role of fire in the evolution of the Hawaiian
flora and vegetation. In: Tall Timbers Fire Ecology Conference, vol 9, pp 5–60
209
Wagner W, Herbst D, Sohmer S (1999) Manual of the Flowering Plants. Revised edition. Bishop Museum, Honolulu
Warshauer F (1974) Biological survey of Kealakomo and
vicinity affected by 1969–1973 lava generated wildfires,
Hawaii Volcanoes National Park. Hawaii Volcanoes
National Park, Hawaii
White P, Pickett S (1985) Natural disturbance and patch
dynamics: an introduction. In: Pickett S, White P (eds)
The ecology of natural disturbance and patch dynamics.
Academic Press, New York
Evaluating different harvest intensities over understory
plant diversity and pine seedlings, in a Pinus pinaster Ait.
natural stand of Spain
J. González-Alday Æ C. Martı́nez-Ruiz Æ
F. Bravo
Originally published in the journal Plant Ecology, Volume 201, No. 1, 211–220.
DOI: 10.1007/s11258-008-9490-2 Springer Science+Business Media B.V. 2008
Abstract Although modern forestry takes into
consideration the analysis of the effects of forest
management on plant structure, diversity and seedlings, little is known about how those parameters
respond to harvest techniques in the Mediterranean
region. We investigated the effect of three different
harvest intensities, respect to uncut controls, on
understory plant species functional groups, richness,
diversity and pine seedlings in a natural Maritime
pine stand in Spain, three years after harvesting. The
harvest treatments produced a reduction of the
number of Pinus pinaster seedlings and woody
species cover, and an increase of species richness
(total and of annual species) and plant cover of
annual species respect to control plots (CO). The
Shannon diversity values showed no differences
between treatments. These results emphasize that
the tree harvest treatments analyzed are not suitable
J. González-Alday (&) C. Martı́nez-Ruiz
Área de Ecologı́a, E.T.S. de Ingenierı́as Agrarias de
Palencia, Universidad de Valladolid, Campus La Yutera,
Avda. de Madrid 44, 34071 Palencia, Spain
e-mail: josucham@agro.uva.es
F. Bravo
Área de Producción Vegetal, E.T.S. de Ingenierı́as
Agrarias de Palencia, Universidad de Valladolid, Campus
La Yutera, Avda. de Madrid 44, 34071 Palencia, Spain
F. Bravo
Joint Research Unit INIA-UVa Sustainable Forest
Management, Madrid, Spain
for the management of this P. pinaster stand.
Otherwise, the reduction of pine seedling density by
harvest treatments and the changes in richness and
cover of functional groups would not induce the
natural regeneration of this stand maintaining the
understory plant layer.
Keywords Anthropogenic disturbance
Herbaceous layer Mediterranean ecosystem
Silviculture Woody species
Introduction
One of the major challenges for modern forestry is to
combine conservation of biodiversity and ecosystem
functioning with wood production and other values
(Hummel 2003; Decocq et al. 2004; Nagai and
Yoshida 2006; Newmaster et al. 2007). These general
principles will obviously need to be achieved using
adequate management practices (Kimmins 2004). It is
generally assumed that management practices, and
especially harvesting, modulate simultaneously the
availability of different types of resources (e.g. light,
water and soil nutrients; Decocq et al. 2004). As a
result, understory species diversity and flora, which
play a fundamental role in the structure and function
of ecosystems (Roberts and Gilliam 1995; Newmaster
et al. 2007), become quite affected (Hughes and
Fahey 1991; Zenner et al. 2006). Therefore, the
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_16
211
212
knowledge of the effects of different harvest disturbances on understory plant layer is an essential
element to implement sustainable management of
forest landscapes (Halpern and Spies 1995; Roberts
and Gilliam 1995; Fahey and Puettmann 2007).
Maritime pine (Pinus pinaster Ait.) is a natural
forest species characteristic of the western Mediterranean basin, mainly distributed over the Iberian
Peninsula, France and Italy (Alı́a et al. 1996).
Traditionally, P. pinaster in central Spain has been
used for resin production and soil protection against
mobile continental dunes (Bravo-Oviedo et al.
2007), with wood production as secondary objective. An important step towards ecologically sound
wood production procedures is to test different
management alternatives (i.e. harvest intensities) to
induce the natural revegetation of these stands. At
the same time, these alternatives should always
contribute to maintain the landscape and ecological
protection functions, mushrooms production and
biodiversity of the stands (Oria de Rueda 2003),
while sustainable wood and resin production is
obtained.
The effect of forest management on plant diversity
and flora is complex and more difficult to generalize
than it was originally thought (Tárrega et al. 2006),
underlining the importance of studying plant and
diversity responses for different forest types and
harvest techniques (Gilliam 2002). Moreover, most
published studies concern managed forests in North
America, whose history and tradition radically differ
from Europe (Decocq et al. 2004), and particularly
from the Mediterranean region (Scarascia-Mugnozza
et al. 2000). The aim of this study, therefore, is to
investigate the effects of three harvest intensities,
relative to untreated control areas, on understory
species richness, diversity, functional groups (life
forms) and P. pinaster seedlings, and their relation to
the remaining basal area and canopy cover of a
natural maritime pine stand in semi-arid Mediterranean conditions in Spain, three years after harvest.
We hypothesized that: (1) the number of P. pinaster
seedlings in such semi-arid Mediterranean conditions
would be reduced by harvest intensity, (2) the
functional groups (annual and perennial herbs and
woody species) cover and richness would be affected
by harvest treatments, and (3) the understory plant
richness and diversity would be markedly affected by
harvest intensity.
A.G. Van der Valk (ed.)
Methods
Study area
This study was conducted in a flat natural Maritime
pine forest located in the Segovia province (Cuellar,
757 m a.s.l.; 41220 N, 4290 W; Central Spain). The
original stand density was 140 stems/ha, tree age
ranges from 80 to 100 years and silvicultural practice
is based on natural regeneration following a shelterwood system adapted to resin production. The
climate is semi-arid Mediterranean, with a mean
annual temperature of 11.2C, a mean annual rainfall
of 461 mm and dry period from the middle of June to
the middle of September (M.A.P.A 1987). The soils
are sandy siliceous of Quaternary age (Junta de
Castilla y León 1988), and the vegetation of the area
is dominated by Pinus pinaster with some isolated
trees of Stone pine (Pinus pinea L.) and crop fields.
Treatments
About 16 continuous hectares of natural Maritime
pine were selected in a ca. 15,000 ha of forest to
delimit twelve 70 9 70 m permanent plots. To record
the variation caused by silviculture treatments rather
than to site variability, the selected hectares shared
the same abiotic conditions, forest structure and
vegetation composition. After plots were established
two variables were recorded for all trees with
diameter at breast height greater than 7.5 cm found
inside the plots: diameter at breath height (DBH; cm),
and crown diameters (m). The DBH and crown
diameters were measured in order to obtain the basal
area (BA) and the canopy cover (%) as informative
parameters of the light conditions for ground vegetation (Härdtle et al. 2003). Three levels of harvest
intensity with three replicates for each one were
applied over nine of the permanent plots: (1) 25% of
basal area removed (close plots, H25), (2) 50% of
basal area removed (open plots, H50), and (3) 100%
of basal area removed (clear-cut plots; CC). All
treatments were randomly allocated on these nine
plots, whereas the remaining three permanent plots
were used as control units (CO) not receiving any
treatment during the study. Harvesting was carried
out manually with handsaw once all trees selected for
cutting were marked according to the basal area
removal criteria. The trees were harvested using a
Forest Ecology
silvicultural criterion to facilitate the natural regeneration, i.e. trees showing disease or physical damage
were removed first, followed by the smaller trees and
finally by others with larger diameters, to increase the
amounts of low- and mid-story shade. Moreover,
harvesting was designed to distribute residual overstory canopies as uniform as possible inside every
particular harvest plot (H25, H50). The sampling of
DBH and crown diameters was carried out in summer
2003, whereas the harvest operations were made in
autumn 2003.
Understory vegetation sampling
To sample understory vegetation in each of 12
permanent plots, 20 quadrats of 1 9 1 m were placed
using simple random sampling design (Krebs 1999).
However, in order to evade edge effect the first 10 m
from the plot edge were avoided. In each quadrat, the
cover (%) of all vascular plant species present and the
number of P. pinaster seedlings (criteria = maximum 3-years old) were estimated visually by the
same observer in May 2006.
Data analyses
Diversity of understory plant communities was
assessed using the Shannon index (H0 ) (Shannon
and Weaver 1949) with logs to base 2, and its two
components, richness (S) and evenness (J0 ) (Pielou
1969). Shannon diversity and richness were calculated on two scales, similar to Tárrega et al. (2006):
(1) on small scale (per quadrat or m2), alpha diversity
or microcosmic diversity (Magurran 2004); and (2)
on a community scale for each plot (4,900 m2), plot
gamma diversity or macrocosmic diversity (from the
joint consideration of the 20 samples carried out for
each studied plot). Evenness, however, was calculated only on a community scale. By using the
comparison of both types of diversity, beta diversity
or spatial heterogeneity was calculated: Sb by the
Whittaker (in Magurran 2004) formula, Sb =
(S/Sa) - 1, and Hb0 as the difference between H0
and the average of Ha0 (Margalef 1972). The number
of P. pinaster seedlings is referred to the total
number of seedlings in the 20 quadrats of each plot.
To evaluate the significance of different harvest
treatments, relative to controls, on the number of
P. pinaster seedlings, functional groups cover and
213
richness (annual herbs, perennial herbs and woody
species), species richness (S), evenness (J0 ) and
diversity values (Ha0 , Hb0 , H0 , Sa and Sb), one-way
analyses of variance (ANOVA) were applied followed by Tukey’s HSD tests to enable pairwise
comparisons of means (P \ 0.05). In all cases, the
inspection of residuals was carried out to check for
normality and homoscedasticity. Nevertheless, when
variables do not meet normality and variance
assumptions, data were transformed using arcsine
squared-root transformation for binomially distributed variables (i.e. plant cover) and squared-root
transformation for count data (i.e. richness) (Zar
1996).
In order to determine possible relationships among
the 13 variables analyzed, a Pearson’s correlation
matrix was constructed considering: canopy cover,
basal area, number of P. pinaster seedlings, number
of woody species, number of perennial and annual
herbs and J0 , Ha0 , Hb0 , H0 , Sa, Sb and S. A Principal
components analysis (PCA) was used to summarize
the relationships among treatments and the variables
as a whole. Data for the 13 variables used in PCA
were standardized prior to analysis, to correct for
different measuring units.
Results were expressed as mean ± standard error
and all statistical computations were implemented in
the R software environment (version 2.7.0; R Development Core Team 2008).
Results
P. pinaster seedlings
The density of P. pinaster seedlings found in the
plots was lower than 3.3 seedlings/m2, however,
significant differences among harvest intensities were
found (F[3,8] = 23.4, P \ 0.001; Fig. 1). Untreated
control plots (CO) showed the greatest number of
seedlings per plot (66 ± 13.5), clear cut plots (CC)
the lowest (1 ± 0.58), and H25 and H50 treated plots
an intermediate number (16 ± 8.5 and 8 ± 1.8,
respectively).
Functional groups (life forms)
Annual species dominated, in number and cover, the
understory plant communities in the four treatments
214
A.G. Van der Valk (ed.)
only differed between the CC and open plots (H50)
(F[3,8] = 5.36, P = 0.026).
Annual species number significantly varied with
harvest intensity. Clear cut plots (CC), with maximum values (41 ± 1.78), followed by open plots
(H50; 31 ± 0.33) showed significantly greater values
than control plots (CO) (F[3,8] = 23.21, P \ 0.001;
Fig. 2b). Perennial species number was also significantly greater in the clear cut plots (CC) than in the
rest (F[3,8] = 7.3, P = 0.011; Fig. 2b), whereas the
number of woody species did not differ with harvest
intensity (F[3,8] = 0.58, P = 0.647).
100
Pinus pinaster seedlings
60
40
b
20
Number of seedlings
80
a
bc
c
0
Richness and diversity
H25
The small scale richness (Sa) varied between 12 and
17 species/m2 in control and clear cut plots, respectively, but not differed significantly among treatments
(F[3,8] = 2.54, P = 0.130; Fig. 3a). In contrast, species richness on a community scale (S), which ranged
between 37 and 62 species per treatment, showed
significantly greater values in CC plots than in the
remainder (F[3,8] = 16.86, P \ 0.001; Fig. 3a). In
spite of that, Shannon diversity index, that showed
high values in the four treatments (H0 always above
4.2; Fig. 3b), did not differ significantly with harvest
intensity (F[3,8] = 0.28, P = 0.835), due to a reduction of evenness; thus, no statistically significant
differences between treated plots (H25, H50 and CC)
and controls (CO; Fig. 3d) were found. There were
also no significant differences in spatial heterogeneity
among harvest intensities (Fig. 3c).
CC
H50
TREATMENTS
Fig. 1 Comparison of the number of P. pinaster seedlings per
plot (total number of seedlings in the 20 quadrats of each plot)
among treatments (mean ± SE). CO: control plots; H25: 25%
of basal area removed (close plots); H50: 50% of basal area
removed (open plots); CC: 100% of basal area removed (clear
cut). Different letters above the bars indicate significant
differences (P \ 0.05) by Tukey’s test
(Fig. 2). Annual cover was similar in the tree treated
areas (H25, H50 and CC), ranging between 37 and
41%, and significantly greater than in the untreated
CO (F[3,8] = 16.59, P \ 0.001; Fig. 2a). On the
contrary, the cover of woody species was significantly greater in the CO than in the CC and open
plots (H50) (F[3,8] = 7.01, P = 0.013), where it
hardly reached a 2%. The cover of perennial herbs
Species richness
50
b)
ab
a
a
40
30
b
ab
a
b
a
a
a
H25
H50
b
b
0
ab
c
20
Species richness (S)
40
30
20
b
10
Plant cover (%)
a
ab
Annual spp.
Peren. herb spp.
Woody spp.
b
b
b
50
Plant cover
a)
0
Fig. 2 Comparison of
annual, perennial
herbaceous and woody
species cover and richness
among treatments
(mean ± SE). See Methods
or Fig. 1 caption for
treatment description.
Different letters above the
bars indicate significant
differences among
treatments (P \ 0.05) by
Tukey’s test
10
CO
CO
H25
H50
TREATMENTS
CC
CO
TREATMENTS
CC
H50
3
H25
Diversity (H')
a
1
0
CO
CC
CO
d)
CC
0.8
0.2
2
0.6
3
0.4
H' beta
S beta
Evenness (J')
Hetrogeneity
H50
TREATMENTS
5
4
H25
1.0
TREATMENTS
c)
H' alpha
4
H' gamma
2
60
70
b
a
a
30
40
50
S gamma
S alpha
20
Species richness (S)
b)
10
a)
0
Fig. 3 Comparison of
different richness (a),
Shannon diversity (b),
heterogeneity (c) and
evenness (d) values among
treatments (mean ± SE).
See Methods or Fig. 1
caption for treatment
description. Different letters
above the bars indicate
significant differences
among treatments
(P \ 0.05) by Tukey’s test
215
5
Forest Ecology
1
0.0
0
CO
H25
H50
CC
TREATMENTS
Relationship between variables
The correlation analysis carried out to determine the
relationship among the 13 variables analyzed
(Table 1) showed that basal area, canopy cover and
P. pinaster seedlings were negatively correlated with
different richness values (S, Sa and Sb), and to the
number of perennial and annual herbs. However, in
general those parameters were not correlated with
diversity values (H0 , Ha0 and Hb0 ). The Shannon
diversity (H0 ) showed a significant positive correlation with Ha0 , Hb0 and J0 , the number of perennial
herbs and woody species. The number of P. pinaster
seedlings was negatively correlated with annual
species number and positively with basal area.
The PCA performed for the joint comparison of all
the variables produced an ordination of plots with the
first two axes accounting for 78% of the total
variance. The first component explained 55% of
variance and was strongly positively correlated with
S, Sa, Ha0 , and number of perennial and annual herbs,
on the contrary it was strongly negatively correlated
with basal area, canopy cover and number of
CO
H25
H50
CC
TREATMENTS
P. pinaster seedlings (Table 2). The second component explained an additional 23% and showed only
positive correlation with diversity (H0 and Hb0 ),
evenness (J0 ) and woody species number (Table 2).
In the ordination diagram, the first axis ordered the
sites according to their treatment, increasing harvest
intensity from the left to the right (Fig. 4). Controls
(CO) were located on the left associated with greater
basal area and lower richness. Close plots (H25) were
located near controls and open plots (H50) in
intermediate position. However, clear cuts (CC)
appeared on the right without basal area and greater
species number. The second axis was related to
diversity gradient, increasing diversity, evenness and
woody species number to the positive end, and
produced a separation between plots within the same
treatment.
Discussion
The results illustrate that three harvest treatments
applied over a natural stand of Maritime pine in Spain
216
A.G. Van der Valk (ed.)
Table 1 Pearson correlation matrix between richness, diversity and functional groups richness
Sc
Sa
Sb
H c0
Ha0
Hb0
J0
As
Ps
Ws
Pp
Cc
Sc
1
Sa
0.85
1
Sb
Hc0
0.41
0.53
-0.13
0.71
1
-0.20
Ha0
0.60
0.83
-0.30
0.66
1
Hb0
0.06
0.04
0.05
0.60
-0.21
-0.32
0.00
-0.57
0.63
0.15
0.66
As
0.94
0.72
0.50
0.26
0.52
-0.21
-0.57
1
Ps
0.92
0.79
0.37
0.59
0.59
0.14
-0.19
0.79
Ws
0.46
0.54
-0.07
0.72
0.33
0.58
0.42
0.28
0.36
Pp
-0.67
-0.73
0.01
-0.26
-0.65
0.35
0.38
-0.72
-0.55
0.03
Cc
-0.83
-0.52
-0.69
-0.19
-0.29
0.06
0.55
-0.86
-0.79
-0.12
0.50
1
Ba
-0.86
-0.59
-0.61
-0.19
-0.40
0.18
0.58
-0.90
-0.81
-0.10
0.60
0.98
J0
Ba
1
1
1
1
1
1
1
As: annual species number; Ps: perennial species number; Ws: woody species number, Pp: number of P. pinaster seedlings, Cc:
canopy cover and Ba: basal area. In bold type are significant correlations at P \ 0.05
Table 2 Correlation coefficients of plot scores along axes 1
and 2 and the 13 variables used in the principal components
analysis (PCA)
Axis 1
Axis 2
Sc
0.98
0.12
Sa
0.82
0.46
Sb
0.43
-0.55
Hc0
0.45
0.87
Ha0
0.61
0.46
Hb0
-0.06
0.65
J0
-0.41
0.87
Annual species number
0.96
-0.16
Perennial herbs species number
0.90
0.18
Woody species number
0.32
0.72
Number of Pinus pinaster seedlings
-0.74
0.04
Canopy cover
-0.88
0.28
Basal area
Eigenvalues
-0.93
8.17
0.29
3.47
Explained variance
55%
23%
In bold type are significant correlations at P \ 0.01
influenced on species richness, annual herbs and
woody species cover, and reduced the number of
P. pinaster seedlings. These results were in agreement with previous studies that have documented
how overstory alterations conditioned the post-disturbance response of understory vegetation (Ramovs
and Roberts 2003).
P. pinaster seedlings
An important result was that the three harvest
intensities reduced the number of P. pinaster seedlings in comparison with control plots; thus the first
hypothesis is accepted. The reduction in the number
of established seedlings from control to clear cut
plots was correlated positively with basal area and
negatively with annual species number. Therefore,
this reduction may be caused by a combination of
factors: (1) a decrease of seed inputs caused by the
elimination of trees in treated plots in comparison
with controls; (2) a reduction of canopy cover, which
undoubtedly changed understory microclimate
(Aussenac 2000), increasing the radiation intensity
during summer and reducing the water availability
for seedlings and their viability (Castro et al. 2004;
Gómez-Aparicio et al. 2005; Calvo et al. 2008) and
(3) competition for water and nutrients between
coniferous seedlings and annual species (Peltzer et al.
2000), since annuals were able to dry up the upper
soil layer leading to seedling mortality, especially
during the early period of seedling development
(Sternberg et al. 2001). Indeed, those factors could be
highly emphasized by the intense summer droughts
detected in the study area at 2004–2006 periods.
Especially over treated plots, because temperature
and moisture stress are lower in the presence of an
overstory cover (Pérez and Moreno 1998; Aussenac
Forest Ecology
217
Fig. 4 First two axes of the
PCA ordination of different
harvest intensity plots. See
Methods or Fig. 1 caption
for treatment description.
The number after the
treatment abbreviation
indicates the number of
replicate
3
CO-3
H25-2
CC-2
2
Control plots
1
Close plots
CO-2
Axis 2 (23 %)
0
Open plots
H25-3
CO-1
CC-3
H50-2
H25-1
H50-1
-1
H50-3
Clear cut plots
-2
-3
-4
CC-1
-5
-6
-4
-2
0
2
4
6
8
Axis 1 (55 %)
2000). The relative importance of each of these
possible explanations required further investigation.
In any case, the density of seedlings found in this
stand three years after harvest is very low, even in
controls (3.3 seedlings/m2), compared with 8 seedlings/m2 recommended to ensure natural regeneration
(Luis-Calabuig et al. 2002). Therefore, artificial
reintroduction of seeds or seedlings may be a suitable
option to increase the seedling density (Pausas et al.
2004), with the objective of facing up to the survival
loss caused by inter-specific competition (Eshel et al.
2000), and water availability by summer droughts
(Gómez-Aparicio et al. 2005), which would become
normal in this area in near future as a consequence of
climate change (Intergovernmental Panel on Climate
Change (IPCC) 2007).
Functional groups (life forms)
The harvest treatments, in comparison with controls,
influenced the richness of annual and perennial herbs
and plant cover of annual herbs and woody species,
thus the second hypothesis is partially accepted.
Different studies have reported that harvesting
increases potential growing space in the understory
(Newmaster et al. 2007), and the relative availability
of resources (Fredericksen et al. 1999), especially
light (Zenner et al. 2006), improving the conditions
for establishment of early colonizer species
(Newmaster et al. 2007). Not surprisingly, our results
provided similar patterns, with an increase of annual
species richness and cover along the harvest intensity
gradient (from controls to clear-cuts).
In these semi-arid Mediterranean forests, with
three months of summer drought, harvesting generates habitats with a strong seasonal stress and with
understory vegetation dominated by annuals. Under
these conditions, perennial species establish themselves with difficulty compared to annuals whose life
cycle is adapted to this seasonal stress (Madon and
Médail 1997). At the same time, and as we said
previously, the pine seedling establishment may be
reduced by the great cover development of annual
species through inter-specific competition (Eshel
et al. 2000).
On more disturbed plots (clear-cut), with greater
solar radiation intensity during summer, species
richness of perennial herbs showed greater values
than on control plots. This may be caused because the
new established species were characteristic of
218
Mediterranean open sites (e.g. Cynodon dactylon or
Armeria arenaria), in accordance with previous
research findings in recent clear-cut stands (Roberts
and Gilliam 1995; North et al. 1996).
Woody species showed an opposite pattern, maintaining their species richness and decreasing in cover
along the harvest intensity gradient (from controls to
clear-cuts). Woody species were more abundant in
sites with higher tree cover, as in control and close
(H25) plots, than in clear-cut and open (H50) plots. It
is possible that the partial shade provided by trees
may alleviate the harsh environmental factors prevailing under full-sun environments (Alrababah et al.
2007), enhancing the woody species growth. However, under the most severe treatments, although
woody species richness was similar to control plots,
the physical destruction of existing woody species by
the harvest operations (Newmaster et al. 2007),
linked to the marked seasonal stress may cause their
cover reduction.
The different responses of annual and perennial
herbs, and woody species richness along the harvest
treatments supports the hypothesis of Peet’s (1978),
who found different response patterns of plant species
richness for different structural groups (woody and
herbs).
Richness and diversity
The influence of harvest is clear only in the case of
richness, therefore the third hypothesis is partially
accepted. Three years after harvesting, understory
plant richness was higher in treated plots than in
controls, although differences were significant only
for the most severe disturbance treatment (clearcutting). At the same time, plant richness had
negatively significant relationship with basal area,
suggesting an increase in richness as harvest intensity
increases, as observed in similar studies in temperate
forest (Fredericksen et al. 1999; Götmark et al. 2005;
Zenner et al. 2006). Harvesting increased species
richness because of the colonization of annuals and
some perennial herbs (Swindel et al. 1983; Götmark
et al. 2005), which were favoured by the modification
of the stand habitat-conditions (Jobidon 1990).
Despite the positive influence of harvesting on
species richness showed in this study, no differences
with control plots on the understory Shannon diversity values were found, as in other studies in
A.G. Van der Valk (ed.)
temperate forest (Gilliam et al. 1995; Gilliam 2002;
Krzic et al. 2003). The relative high Shannon diversity values reached under all treatments indicated that
plant communities after harvesting were not dominated by just a few species (Krzic et al. 2003). On the
contrary, these results did not suggest that an increase
in harvest intensity did not influence the understory
species layer. Peltzer et al. (2000) found that plant
diversity did not change when increasing the intensity
of silvicultural disturbances, but a higher number of
herb species appeared. These results are consistent
with our findings of increasing annual and perennial
herbs richness with harvest intensity.
The Shannon index (H0 ) is affected by species
richness and evenness (Westman 1990). As previously explained, richness increased as harvest
intensity increases, whereas evenness decreased,
resulting in no changes in the Shannon diversity
index (H0 ). This suggest that in control plots the
relative abundance of species is more similar than in
treated plots (H25, H50 and CC), in which some of
new species tend to be relatively uncommon or rare
(Small and McCarthy 2002).
The separation between different harvest treatments was clearly connected with basal area, canopy
cover and P. pinaster seedlings reduction, and with
the increase of richness (S and Sa) and herbs richness
(annual and perennial). This indicates that the
elimination of tree cover favoured the establishment
of new herbs species, which increased their cover by
the addition of more species, rather than by the
growth increase of a few of them (Gilliam 2002). In
contrast, diversity (H0 and Hb0 ), evenness and woody
species number were related with differences
between plots of the same treatment, rather than
with differences between harvest treatments. These
results emphasize the difficulty in making general
conclusions of the effects of harvest treatments
(disturbances) on diversity, supporting the conclusions of Gilliam (2002) and Tárrega et al. (2006).
Finally, our results emphasize that the tree harvest
treatments assessed are not suitable for the management of this P. pinaster stand. Otherwise, the
reduction of pine seedling density and the changes
in richness and cover of functional groups by harvest
treatments would not induce the natural regeneration
of this stand, maintaining the understory plant layer.
Managers must realize that even controls would have
problems to ensure natural regeneration, because the
Forest Ecology
pine seedling density reached in three years is not
enough to guarantee it. Therefore, further investigations are needed to assess seedling establishment
limiting factors, the effectiveness of reintroduction of
pine seeds or seedlings and other silvicultural alternatives (i.e. single tree selection or nurse plant
strategies) to achieve adequate management practices, including wood production, with respect to
ecosystem functioning.
Acknowledgements We thank Sonia Garcı́a-Muñoz, Cristobal
Ordóñez and Ana I. de Lucas for fieldwork assistance, and Pilar
Zaldı́var for species nomenclature assistance. This study was
supported by a grant from the Basque-Country Government to
J. González-Alday (BFI06.114), and Research Projects from the
Spanish Science National Program (codes AGL2001-1780 and
AGL2004-07094-C02-02/FOR) to Felipe Bravo.
References
Alı́a R, Martı́n R, de Miguel J, Galera RM, Agúndez D, Gordo J
et al (1996) Regiones de procedencia Pinus pinaster Ait.
DGCN, Madrid
Alrababah MA, Alhamad MA, Suwaileh A, Al-Gharaibeh M
(2007) Biodiversity of semi-arid Mediterranean grasslands: impact of grazing and afforestation. Appl Veg Sci
10:257–264. doi:10.1658/1402-2001(2007)10[257:BOSM
GI]2.0.CO;2
Aussenac G (2000) Interactions between forest stands and
microclimate: ecophysiological aspects and consequences
of silviculture. Ann Sci 57:287–301. doi:10.1051/forest:
2000119
Bravo-Oviedo A, del Rı́o M, Montero G (2007) Geographic
variation and parameter assessment in generalized algebraic difference site index modeling. For Ecol Manag
247:107–119
Calvo L, Santalla S, Valbuena L, Marcos E, Tárrega R, LuisCalabuig E (2008) Post-fire natural regeneration of Pinus
pinaster forest in NW Spain. Plant Ecol. doi:10.1007/
s11258-007-9362-1
Castro J, Zamora R, Hódar JA, Gómez JM (2004) Seedling
establishment of boreal tree species (Pinus sylvestris) at
its southern most distribution limit: consequences of being
in a marginal Mediterranean habitat. J Ecol 92:266–277.
doi:10.1111/j.0022-0477.2004.00870.x
Decocq G, Aubert M, Dupont F, Alard D, Saguez R, WattezFranger A et al (2004) Plant diversity in a managed
temperate deciduous forest: understory response to two
silvicultural systems. J Appl Ecol 41:1065–1079. doi:
10.1111/j.0021-8901.2004.00960.x
Eshel A, Henig-Sever N, Ne’eman G (2000) Spatial variation
of seedling distribution in an east Mediterranean pine
woodland at the beginning of post-fire succession. Plant
Ecol 148:175–182. doi:10.1023/A:1009880416760
Fahey RT, Puettmann KJ (2007) Ground-layer disturbance and
initial conditions influence gap partitioning of understory
219
vegetation. J Ecol 95:1098–1109. doi:10.1111/j.13652745.2007.01283.x
Fredericksen TS, Ross BD, Hoffman W, Morrison ML, Beyea
J, Johnson B, Lester MB, Ross E (1999) Short-term
understory plant community responses to timber-harvesting intensity on non-industrial private forestlands in
Pennsylvania. For Ecol Manag 116:129–139
Gilliam FS (2002) Effects of harvesting on herbaceous layer
diversity of a central Appalachian hardwood forest in
West Virginia, USA. For Ecol Manag 155:33–43
Gilliam FS, Turril NL, Adams MB (1995) Herbaceous-layer
and overstory species in clear-cut and mature Appalachian
hardwood forests. Ecol Appl 5:947–955. doi:10.2307/226
9345
Gómez-Aparicio L, Gómez JM, Zamora R, Boettinger JL
(2005) Canopy vs soil effects of shrubs facilitating tree
seedlings in Mediterranean montane ecosystems. J Veg
Sci 16:191–198. doi:10.1658/1100-9233(2005)016[01
91:CVSEOS]2.0.CO;2
Götmark F, Paltto H, Norden B, Götmark E (2005) Evaluating
partial cutting in broadleaved temperate forest under
strong experimental control: short-term effects on herbaceous plants. For Ecol Manag 214:124–141
Halpern CA, Spies TA (1995) Plant species diversity in natural
and managed forests of the Pacific Northwest. Ecol Appl
5:913–934. doi:10.2307/2269343
Härdtle W, von Oheimb G, Wesphal C (2003) The effects of
light and soil conditions on the species richness of the
ground vegetation of deciduous forest in northern Germany (Schleswig-Holstein). For Ecol Manag 182:327–338
Hughes JW, Fahey TJ (1991) Colonization dynamics of herbs
and shrubs in a disturbed northern hardwood forest. J Ecol
79:605–616. doi:10.2307/2260656
Hummel SS (2003) Managing structural and compositional
diversity with silviculture. In: Monserud RA, Haynes RW,
Johnson AC (eds) Compatible forest management. Kluwer
Academic Publishers, the Netherlands, pp 85–120
Intergovernmental Panel on Climate Change (IPCC) (2007)
The physical science basis. Cambridge University Press,
New York
Jobidon R (1990) Short-term effect of three mechanical site
preparation methods on species diversity. Tree Plant
Notes 41:39–42
Junta de Castilla y León (1988) Análisis del medio fı́sico de
Segovia. EPYPSA, Valladolid
Kimmins JP (2004) Forest ecology: a foundation for sustainable forest management and environmental ethics in
forestry. Prentice Hall, NJ
Krebs CJ (1999) Ecological methodology. Addison-Welsey
Educational Publishers, Menlo Park, CA
Krzic M, Newman RF, Broersma K (2003) Plant species
diversity and soil quality in harvested and grazed boreal
aspen stands of northeastern British Columbia. For Ecol
Manag 182:315–325
Luis-Calabuig E, Torres O, Valbuena L, Calvo L, Marcos E
(2002) Impact of large fires on a community of Pinus
pinaster. In: Trabaud L, Prodon R (eds) Fire and biological processes. Backhuys Publishers, Leiden, pp 1–12
Madon O, Médail F (1997) The ecological significance of
annuals on a Mediterranean grassland. Plant Ecol
129:189–199. doi:10.1023/A:1009759730000
220
Magurran AE (2004) Measuring biological diversity. Blackwell Publishing, Oxford
M.A.P.A (1987) Caracterización agroclimática de la provincia
de Segovia. Ministerio de Agricultura Pesca y Alimentación, Madrid
Margalef R (1972) Homage to Evelyn Hutchinson, or why is
there an upper limit to diversity. Trans Conn Acad Arts
Sci 44:211–235
Nagai M, Yoshida T (2006) Variation in understory structure
and plant species diversity influenced by silvicultural
treatments among 21- to 26-year old Picea glehnii plantations. J For Res 11:1–10. doi:10.1007/s10310-0050176-5
Newmaster SG, Parker WC, Bell FW, Paterson JM (2007)
Effects of forest floor disturbances by mechanical site
preparation on floristic diversity in a central Ontario
clearcut. For Ecol Manag 246:196–207
North M, Chen J, Smith G, Krakowiak L, Franklin J (1996)
Initial response of understory plant diversity and overstory
tree diameter growth to a green tree retention harvest.
Northwest Sci 70:24–35
Oria de Rueda JA (2003) Guı́a de los árboles y arbustos de
Castilla y León. Calamo, Palencia
Pausas JG, Bladé C, Valdecantos A, Seva JP, Fuentes D, Alloza JA et al (2004) Pines and oaks in the restoration of
Mediterranean landscapes of Spain: new perspectives for
an old practice-a review. Plant Ecol 171:209–220. doi:
10.1023/B:VEGE.0000029381.63336.20
Peet RK (1978) Forest vegetation of the Colorado Front Range:
pattern of species diversity. Vegetatio 37:65–78. doi:
10.1007/BF00126830
Peltzer DA, Bast ML, Wilson SD, Gerry AK (2000) Plant
diversity and tree responses following contrasting disturbances in boreal forest. For Ecol Manag 127:191–203
Pérez B, Moreno JM (1998) Fire-type and forestry management
effects on the early postfire vegetation dynamics of a Pinus
pinaster woodland. Plant Ecol 134:27–41. doi:10.1023/
A:1009733818670
Pielou EC (1969) An introduction to mathematical ecology.
Wiley, New York
R Development Core Team (2008) R: a language and environment for statistical computing. R Foundation for
A.G. Van der Valk (ed.)
Statistical Computing, Vienna, Austria. ISBN 3-90005107-0, URL: http://www.R-project.org
Ramovs BV, Roberts MR (2003) Understory vegetation and
environment responses to tillage, forest harvesting, and
conifer plantation development. Ecol Appl 13:1682–1700.
doi:10.1890/02-5237
Roberts MR, Gilliam FS (1995) Patterns and mechanisms of
plant diversity in forested ecosystems: implications for
forest management. Ecol Appl 5:969–977. doi:
10.2307/2269348
Scarascia-Mugnozza G, Oswald H, Piussi P, Radoglou K
(2000) Forests of the Mediterranean region: gaps in
knowledge and research needs. For Ecol Manag 132:97–
109
Shannon CE, Weaver W (1949) The mathematical theory of
communication. University of Illinois Press, Urbana
Small JC, McCarthy BC (2002) Spatial and temporal variability of herbaceous vegetation in an eastern deciduous
forest. Plant Ecol 164:37–48. doi:10.1023/A:1021
209528643
Sternberg M, Danin A, Noy-Meir I (2001) Effects of clearing
and herbicide treatments on coniferous seedling establishment and growth in newly planted Mediterranean
forests. For Ecol Manag 148:179–184
Swindel BF, Conde LF, Smith JE (1983) Plant cover and
biomass response to clear-cutting, site preparation, and
planting in Pinus elliottii flatwoods. Science 219:1421–
1422. doi:10.1126/science.219.4591.1421
Tárrega R, Calvo L, Marcos E, Taboada A (2006) Forest
structure and understory diversity in Quercus pyrenaica
communities with different human uses and disturbances.
For Ecol Manag 227:50–58
Westman EW (1990) Managing for biodiversity: unresolved
science and policy questions. Bioscience 40:26–33. doi:
10.2307/1311236
Zar JH (1996) Biostatistical analysis. Prentice Hall, London
Zenner EK, Kabrick JM, Jensen RG, Peck JE, Grabner JK
(2006) Responses of ground flora to a gradient of harvest
intensity in the Missouri Ozarks. For Ecol Manag
222:326–334
Land-use history affects understorey plant species
distributions in a large temperate-forest complex, Denmark
Jens-Christian Svenning Æ Karen H. Baktoft Æ
Henrik Balslev
Originally published in the journal Plant Ecology, Volume 201, No. 1, 221–234.
DOI: 10.1007/s11258-008-9557-0 Springer Science+Business Media B.V. 2008
Abstract In Europe, forests have been strongly
influenced by human land-use for millennia. Here, we
studied the importance of anthropogenic historical
factors as determinants of understorey species distributions in a 967 ha Danish forest complex using 156
randomly placed 100-m2 plots, 15 environmental, 9
spatial, and 5 historical variables, and principal
components analysis (PCA), redundancy analysis
(RDA) as well as indicator species analysis. The
historical variables were status as ancient (1805 AD)
high forest, reclaimed bogs, B100 m from Bronze
Age burial mounds, or former conifer plantation, and
stand age. The PCA results showed that the main
gradients in species composition were strongly
related to the explanatory variables. Forward variable
selection and variation partitioning using RDA
showed that although modern environment was the
dominant driver of species composition, anthropogenic historical factors were also important. The pure
historical variation fraction constituted 13% of the
variation explained. The RDA results showed that
ancient-forest status and, secondarily, reclaimed bog
status were the only significant historical variables.
Many typical forest interior species, with poor
dispersal and a strong literature record as ancient-
J.-C. Svenning (&) K. H. Baktoft H. Balslev
Department of Biological Sciences, Aarhus University,
Ny Munkegade, Build 1540, Aarhus C DK-8000,
Denmark
e-mail: svenning@biology.au.dk
forest species, were still concentrated in areas that
were high forest in 1805. Among the younger forests,
there were clear floristic differences between those on
reclaimed bogs and those not. Apparently remnant
populations of wet-soil plants were still present in the
reclaimed bog areas. Our results emphasize the
importance of historical factors for understanding
modern vegetation patterns in forested landscapes.
Keywords Ancient-forest indicator species
Ancient woodland Dispersal limitation
Forest history Forest management
Historical factors
Introduction
An important goal in ecology is to establish the
determinants of species distributions and community
composition. Plant species distributions in temperate
forest ecosystems are well-known to partly reflect
natural heterogeneity in the environment, notably
edaphic gradients and forest canopy structure (Ellenberg 1988; Motzkin et al. 1999; Svenning and Skov
2002; Dupré and Ehrlén 2002). However, natural and
anthropogenic historical factors will often also be
important. Historical factors may act not only via
persistent environmental changes, but also via purely
population dynamic legacies, notably dispersal limitation and the occurrence of remnant populations
(Gilliam 2007; Eriksson 1996). Natural historical
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_17
221
222
factors include the legacies of past natural disturbances such as hurricanes or fires (Motzkin et al.
1999) as well as postglacial migration (Popiela 2004;
Van der Veken et al. 2007; Svenning et al. 2008) and
other dispersal processes (Svenning and Skov 2002;
Miller et al. 2002). Human land-use have pervasively
affected temperate forest ecosystems for several
thousands years (Ellenberg 1988; Peterken 1996;
Björse and Bradshaw 1998; Bradshaw and Holmqvist
1999). A significant proportion of forests in Europe
and North America are secondary forests that have
developed on previously cleared lands (Peterken 1996;
Gilliam 2007). Plant diversity in ancient forests often
exceeds that of secondary forests and may continue to
do so for hundreds of years (Peterken and Game 1984;
Matlack 1994). The persistent differences in large part
reflect the limited dispersal capabilities of many forest
herbs (e.g. Matlack 1994; Hermy et al. 1999; Honnay
et al. 2002; Hermy and Verheyen 2007). However,
past deforestation and agricultural land-use can cause
persistent soil changes (Kristiansen 2001; Dupouey
et al. 2002; Dambrine et al. 2007), which may also
contribute to the diversity and floristic differences
between ancient and secondary forests (Honnay et al.
1999; Dupouey et al. 2002; Dambrine et al. 2007; but
cf. Graae et al. 2004). Agricultural after-effects,
legacy effects sensu Gilliam (2007), on plant species
distributions may last [1,700 years (Dupouey et al.
2002; Dambrine et al. 2007). Although less studied,
other aspects of past land-use than deforestation,
notably past grazing practices, establishment of conifer plantations within broadleaved deciduous forests,
artificial drainage and clear-felling of large stands,
may also be important controls of understorey plant
species distributions. For example, past grazing may
have persistent effects due to slow recolonization by
grazing-sensitive species (Brunet 1992).
The understorey is the most species-rich vegetation
layer and an important ecological component of forest
ecosystems (Gilliam 2007). The present study investigates the factors determining understorey species
distributions, using a large Danish forest complex as a
case study. In Denmark, forests have been strongly
influenced by human land-use during the last
5,500 years (Nørrevang and Lundø 1980; Fritzbøger
1994; Bradshaw and Holmqvist 1999). In 1805 AD,
unsustainable use had caused forest cover in Denmark
to dwindle to 4%, but this has since increased to c.
11% due to protection and afforestation (Nørrevang
A.G. Van der Valk (ed.)
and Lundø 1980; Fritzbøger 1994). During the last
200 years, most Danish forests have been subject to
intense management by clear-felling, drainage and
frequent conversion of broadleaved forest to plantations of introduced conifers (Nørrevang and Lundø
1980). Here, we use constrained ordinations, variation
partitioning and indicator species analyses to evaluate
the importance of anthropogenic historical factors as
determinants of understorey species distributions in
the broadleaved stands of a 967 ha forest complex in
Denmark. We investigated five historical factors: (1)
prehistoric forest clearance, as indicated by the
occurrence of Bronze Age burial mounds with ancient
forest; (2) historical forest continuity, as indicated by
the contrast between ancient (1805 AD) high forest
and younger forest; forest management, as represented by (3) the contrast between drained and
afforested bogs and other forest; (4) the contrast
between broadleaf-converted conifer plantations and
other forest and (5) stand age. We asked the following
specific questions: (1) what is the relative importance
of the anthropogenic historical factors, modern environment and broadscale spatial factors, potentially
reflecting unmeasured environmental variation or
unknown historical effects, e.g. due to natural dispersal processes? (2) which of the anthropogenic
historical factors affect modern distributions of
understorey species in the forest complex?
Methods
Study area
The study area is the 967 ha Engestofte–Søholt forest
complex (UTM coordinates: 663,100–667,800 E,
6,066,600–6,070,800 N) in south-eastern Denmark
(Fig. 1). Annual precipitation is 600 mm and average
January and July temperatures are -0.1 and 16.5C,
respectively (Laursen et al. 1999). The landscape is
rather flat (maximum 30 m above sea-level), and
soils are a mosaic of late-glacial moraine clay and
moraine sand and 19th century alluvial sediments
(Rasmussen 1966; Høy and Dahl 2000). The forest
complex consists of nine more or less contiguous
forests and is a mosaic of managed Fagus sylvaticaor, less commonly, Quercus robur-dominated forest,
Alnus glutinosa-dominated swamps, mixed forests,
part of which are left relatively unmanaged, conifer
Forest Ecology
223
Fig. 1 Map of Denmark
showing the study area in the
south-eastern parts of the
country. The larger map
shows the forest complex of
Engestofte–Søholt. The
locations of the 156 study
plots are shown, with
symbols indicating whether
they were located in ancient
forest, on reclaimed bog, or
elsewhere. The black
triangle indicates one plot
categorized both as
reclaimed bog and ancient
forest, probably reflecting
the coarser resolution of the
older maps used for the latter
classification (see Methods)
plantations and open bogs. Large areas of the forest
complex have been drained, and bogs and meadows
have been afforested, partly with conifer plantations,
during the last 200 years (cf. the historical maps
discussed under Methods: Explanatory variables).
The central-northern part contains 62 Bronze Age
burial mounds dated to 1,500–800 BC (Henrik
Schilling, Lolland-Falsters Stiftsmuseum, pers.
comm., 2002). About half of the currently forested
area was given status as protected high forest by the
Forest Conservation Act in 1805 (Fritzbøger 1994).
Much of the remaining area also had some degree of
tree cover, but was not protected by law in 1805
mainly due to the degraded, patchy and scrubby
nature of any forest vegetation present (Nørrevang
and Lundø 1980; Fritzbøger 1994).
stumps, large stones or large trees by moving the
mesoplot to the northeast, if necessary. We used an
Økland frame (Økland 1990), dividing the mesoplot
into 16 microplots and recording the frequency of each
species as the number of microplots in which it was
rooted. In addition, understorey species present in the
macroplot, but only outside the mesoplot, were given
the frequency 0.1. Changing the value to 0.01 produced
similar ordinations (results not shown). Plant nomenclature follows Hansen (1984). The species pairs,
Viola reichenbachiana and Viola riviniana; and
Rubus fruticosus agg. and Rubus corylifolius agg,
were lumped as Viola sp. and Rubus fruticosus,
respectively.
Understorey vegetation sampling
Five descriptors of past anthropogenic disturbance
were derived for each macroplot based on historical
maps. (1) Ancient forest: it was scored whether (1) or
not (0), a macroplot was located in forest that was
protected by the 1805 Forest Conservation Act,
which protected contemporary high forest (Nørrevang and Lundø 1980; Fritzbøger 1994), according to
the 1802–1815 Royal Danish Academy of Sciences
and Letters maps (resolution slightly coarser than
1:20,000) available at the National Survey and
Cadastre, Copenhagen, Denmark. The small Staverholm area was not covered by these maps, but since it
is known as a traditional grazing area it was not
scored as ancient forest. We note that the large
During July–October 2002, we inventoried vegetation
and environmental descriptors in 156 sq. 10 9 10 m2
macroplots placed throughout the study area using
random coordinates, located using a Garmin eTrex
Venture GPS (Garmin International, Kansas, USA).
Sites on trails, below conifers, or in open areas (canopy
openness [30%, see below) were not considered.
During July–August 2002, all understorey species, i.e.
species with potential height below 4 m, were recorded
for each macroplot. Species frequencies were determined based on a 1 9 1 m2 mesoplot placed at the
centre of each macroplot; avoiding fallen trunks,
Explanatory variables
224
majority of the younger forest areas were already
afforested by the mid-1800s according to the 1842–
1899 1:20,000 resolution General Staff’s Topographic Department ordnance maps available at the
National Survey and Cadastre. (2) Burial mounds: it
was scored whether (1) or not (0), a macroplot was
B100 m from a burial mound, as shown on the
ordnance maps. Danish Bronze age burial mounds
were located near settlements (Laursen 1994); hence,
the surrounding area was most likely deforested at
that time (1800–500 BC). (3) Reclaimed bog: it was
scored whether (1) or not (0), a given macroplot was
located in areas consisting largely of open bog
according to the 1842–1899 ordnance maps, but
now drained and forested. (4) Converted conifer
plantation: we also assessed whether (1) or not (0), a
macroplot was located in a conifer plantation according to the 1842–1899 ordnance maps, albeit it now
has a broadleaved overstorey. (5) Stand age, computed from the year of establishment of the present
tree stand according to stand-scale forestry maps
made by Hedeselskabet, Viborg, Denmark in 1996.
Fifteen descriptors of the modern environment
were also recorded for each macroplot. (1) Slope: the
slope inclination measured by clinometer along the
steepest diagonal of each macroplot. (2) Heat index:
calculated as tan(a1) 9 cos(a2), where a1 is the slope
inclination and a2 is the compass deviation from
south-west (202.5), the most favourable aspect for
plant growth due to high incoming radiation at times
of day with high temperatures (Lawesson et al.
2000). Based on one sample per mesoplot of the
topmost 20 cm of soil taken using a 3-cm diameter
auger, the soil type was described using indicator
variables (McCune and Grace 2002) as (3) sandy, not
possible to press the soil into a roll; (4) sandy–clay,
the soil can be pressed together into a roll, but the roll
disintegrates when rolled between ones palms; (5)
clay, the soil can be made into a coherent roll or (6)
organic, i.e. soils, which appear to be purely
composed of organic material. The sandy–clay variable was chosen as the omitted category (McCune
and Grace 2002). (8) Litter cover: scored as present
(1) or absent (0) in each mesoplot in July–August. (7)
Soil moisture: volumetric water content (%) was
measured in each mesoplot at five random plots using
a Theta probe type HH1 instrument (Thermo VG
Scientific, East Grinstead, England) during September 27–30 after 1 month largely without rain.
A.G. Van der Valk (ed.)
Values [ 60% were truncated to 60% due to an
apparent non-linearity in the measurements. The
mean soil moisture per mesoplot was used in analyses
after log transformation to reduce skewness. (9) Soil
pH: measured in distilled water following Schierup
and Jensen (1979) for one soil sample per mesoplot.
Samples were collected September 27–30. (10)
Canopy openness: estimated using a canopy-scope
(Brown et al. 2000) 1.30 m above the centre of each
mesoplot. Canopy openness was square-root transformed to reduce skewness. (11) Tree shade: based on
the proportional tree species dominance in the canopy
above each macroplot, a weighted shade score was
computed based on Ellenberg’s (1988) species-specific ‘ability to produce shade’ categories (very
low = 1, low = 2, medium = 3, high = 4, and very
high = 5). (12) Basal area: computed based on the
diameter at 1.30 m above-ground (dbh) of all woody
stems with dbh C 1.0 cm rooted in the macroplot.
(13) Stem density: the number of woody stems with
dbh C 1.0 cm rooted in the macroplot. Stem density
was square-root transformed to reduce skewness. (14)
Tree diversity: the number of tree species with stems
with dbh C 1.0 cm rooted in the macroplot. (15)
Forest heterogeneity: estimated using the score system developed by Aude and Lawesson (1998): ?1 for
uprooted trees or left broken trees; ?1 for branches on
the forest floor; ?1 for each number of woody stem
diameter classes represented (0–5, 6–10, 11–20, 21–
40, 41–60 and[60 cm); ?2 for single trees[75 cm in
diameter; ?1 for big stones; -1 for trees in rows and
-1 for being adjacent to conifer stands (our addition
to emphasize natural structural heterogeneity). Understorey species were not considered.
To account for broad-scale spatial trends not
accounted for by the historical and environmental
descriptors, the nine terms of a cubic trend surface
polynomial expansion (X, Y, X2, Y2, XY, X3, Y3, X2Y
and XY2), where X and Y are the centred UTM
coordinates, were used to derive as a set of spatial
explanatory variables (Borcard et al. 1992).
Data analyses
We used unconstrained ordination to describe the main
gradients in understory species composition across the
156 macroplots. The method employed was principal
components analysis (PCA) on the Hellinger distancetransformed species data (Legendre and Gallagher
Forest Ecology
2001). This transformation allows species distribution
data with many zeroes and non-linear species response
curves to be analyzed by Euclidean-based ordination
methods like PCA, which thereby offer an often
preferable alternative to the chi-square distance-based
correspondence analysis and its derivatives (Legendre
and Gallagher 2001); notably, the Hellinger distance
does not give rare species differential weighting. We
used Wilcoxon rank sum chi-square approximation
and Spearman rank correlation tests to test for
relationships between the PCA axes and the explanatory variables. To directly assess the importance of
the historical, environmental and spatial variables as
controls of understorey species composition, linear
redundancy analysis (RDA) was used, which can be
viewed as the canonical extension of PCA with the
ordination vectors being constrained by multiple
regression to be linear combinations of the original
explanatory variables (Legendre and Legendre 1998).
By using the Hellinger distance transformation, RDA
has the same advantages as described for PCA above.
Furthermore, in contrast to the more frequently used
canonical correspondence analysis RDA does not give
sites with many species and individuals’ higher
weighting (Legendre and Gallagher 2001). Significance of the canonical models was tested using 9,999
unrestricted reduced-model permutations (ter Braak
and Smilauer 2002). To provide an estimate of the best
set of non-redundant variables for predicting species
composition and to provide a ranking of the relative
importance of the individual explanatory variable, we
used forward selection of the explanatory variables in
order of additional variance explained and tested by a
permutation test (using 999 permutations) using P-to
enter \0.05. We used RDA and partial RDA to
partition the variation in the understory species
distribution data into independent variation components (Borcard et al. 1992; Økland and Eilertsen 1994;
Økland 1999) using all 28 explanatory variables.
Following Økland (1999), the explained variation was
expressed as fractions of the total variation explained
(TVE) by the complete set of descriptors. The TVE
was divided into seven non-overlapping fractions, viz.
pure environmental, pure historical, pure spatial,
mixed environmental–historical, mixed environmental–spatial, mixed historical–spatial and mixed
environmental–historical–spatial. The pure fractions
were computed using partial RDA, e.g. the pure
historical fraction was computed from a partial RDA
225
with the historical variables as explanatory variables
and the environmental and spatial variables as covariables. The mixed fractions were computed by
subtraction and addition of the variation fractions
produced by RDA and partial RDA of particular sets of
explanatory and co-variables following Cushman and
Wallin (2002). The unique importance of each single
explanatory variable was assessed using partial RDA
with all the other 27 variables set as co-variables. In
addition, to further elucidate the historical influence a
partial RDA with the historical factors as explanatory
variables and the environmental variables as covariables was also computed using the forward
selection procedure described above. Since historical
effects mediated by dispersal limitation are expected
to be present in large part as broad- rather than finescale relationships with the causal historical factors
(e.g. presence of ancient forest), we did not use the
spatial variables as co-variables in this analysis, since
this would remove the broad-scale component (DinizFilho et al. 2003). The PCA and RDAs were computed
after centring, but not standardizing the species data
table, i.e. on the covariance matrix (Legendre and
Legendre 1998; McCune and Grace 2002). Indicator
species analysis (Dufrêne and Legendre 1997) was
used to pinpoint characteristic species for the categories of the historical factors found to have significant
effects on understorey species composition according
to the RDA. In addition to the original categories, we
also derived new sets of categories for the historical
factors based on the sample scores from final partial
RDA model, with the historical factors as explanatory
variables and the environmental variables as covariables, to find characteristic species for the nonenvironmental historical legacies. CANOCO 4.5 was
used for computing RDA and PCA (ter Braak and
Smilauer 2002), while indicator species analysis was
computed using PC-ORD, version 4 (McCune and
Mefford 1999). All other analyses were done in JMP
4.0.4 (SAS Institute Inc., Cary, North Carolina, USA,
2001).
Results
In the 156 macroplots in the Engestofte–Søholt forest
complex, we found 166 species of understorey plants,
whereof 52 species (31%) were listed as ancientforest species in Hermy et al. (1999). Understorey
226
species richness per macroplots ranged 4–44. The
most frequent species were Melica uniflora (109
macroplots), Deschampsia caespitosa (99), Galium odoratum (94), Milium effusum (92) and
Rubus idaeus (86). Of the 166 understorey species,
26 (16%) were found in just a single macroplot.
Among the macroplots, 70 macroplots were scored as
ancient forest, 25 as B100 m from a burial mound, 23
as reclaimed bog, and 8 as converted conifer stands.
The mean and median stand age was 71 and 65 years,
respectively (range 10–144 years). There was no
difference in stand age between ancient forest or
reclaimed bogs and other forest (Wilcoxon rank sum
chi-square approximation tests), while converted
conifer stands had higher median stand age (88 years)
than other forest (65 years; Wilcoxon rank sum chisquare approximation test, P = 0.05).
The first four axes of the PCA had eigenvalues
accounting for 13.4%, 8.6%, 6.7% and 5.3% of the total
variation, respectively. Among the historical variables,
stand age did not correlate with the sample scores for
the first four PCA axes (Spearman rank correlations,
P [ 0.05). However, Wilcoxon rank sum chi-square
approximation tests indicated differences in PCA
sample scores among plots that were or were not
ancient forest (axis 1, P \ 0.0001; axis 3, P \ 0.05;
axis 4, P \ 0.001), reclaimed bog (axis 1, P \ 0.0001;
axis 3, P \ 0.05; axis 4, P \ 0.05), and B100 m from a
burial mound (axis 1, P \ 0.05), while there were no
significant differences between plots that were or were
not converted conifer plantations. The environmental
variables with the strongest correlations with PCA axis
1 were soil pH (rs = 0.58, P \ 0.0001), forest heterogeneity (rs = 0.52, P \ 0.0001), tree diversity
(rs = 0.44, P \ 0.0001), and tree shade (rs = -0.44,
P \ 0.0001); with axis 2, slope (rs = 0.32,
P \ 0.0001) and tree shade (rs = 0.31, P \ 0.0001);
with axis 3, soil pH (rs = -0.32, P \ 0.0001) and with
axis 4, stem density (rs = -0.28, P \ 0.001) and
forest heterogeneity (rs = -0.25, P \ 0.01). Wilcoxon rank sum chi-square approximation tests also
indicated differences in PCA sample scores among
plots that had or had not organic soil (axes 1 and 3,
P \ 0.05), litter cover (axis 1, P \ 0.0001; axis 1,
P \ 0.05). The spatial variable with the strongest
correlations with PCA axis 1 was Y2 (rs = 0.36,
P \ 0.0001), with axis 3, X2 (rs = 0.33, P \ 0.0001)
and with axis 4, Y2 (rs = 0.27, P \ 0.001). No spatial
variables had rs C 0.20 with axis 2.
A.G. Van der Valk (ed.)
The RDA using all 28 explanatory variables
explained 32.4% of the total variation in understory
species composition across the 156 macroplots (first
canonical eigenvalue = 0.090, P \ 0.0001; sum of all
canonical eigenvalues = 0.324, P \ 0.0001). Note
that the sum of the eigenvalues equal the proportion
of the TVE, since CANOCO sets the total variance = 1 in PCA and RDA (ter Braak and Smilauer
2002). After forward variable selection 15 of the 28
explanatory variables were included in the final model,
which explained 24.9% of the total variation (Table 1;
Figs. 2, 3). The three most important explanatory
variables were the environmental variables soil pH and
tree shade and the historical variable ancient forest
(Table 1). Among the historical variables, reclaimed
bog was also included in the final model, while stand
age, converted conifer and burial mound were not
(Table 1). From the RDA axes 1–2 biplot, it can seen
that understorey species typical of ancient forest
(notably Lamiastrum galeobdolon, Melica uniflora,
Poa nemoralis, Anemone nemorosa, Viola sp. and
Galium odoratum) also tended to be associated with
shady, simply structured forest and relatively dry, acid
soils with a thick litter layer, while those typical of
reclaimed bogs (notably Rubus caesius, Brachypodium sylvaticum, Eupatorium cannabinum, Lythrum
salicaria and Rosa canina) had the opposite associations (Fig. 2).
The RDA-based variation partitioning showed that
the historical, environmental and spatial explanatory
variable groups to a large extent had independent
effects on understorey species composition (Table 2).
While the component purely driven by modern
environment was by far the largest, both the purely
spatial and the purely historical components were also
highly significant and of non-negligible size (Table 2).
In a partial RDA with the five historical variables as
explanatory variables and the 14 environmental variables as co-variables, the forward selection included
first ancient forest (F-ratio = 3.65, P = 0.001) and
then second and finally reclaimed bog (F-ratio = 2.26,
P = 0.001). The resulting RDA model (first canonical
eigenvalue = 0.020, P \ 0.0001; sum of all canonical
eigenvalues = 0.033, P \ 0.0001) accounted for
13.3% of the variation explained by final model
produced by forward selection among all 28 explanatory variables (Table 1). The species most strongly
purely associated with ancient forest were Carex
sylvatica, Viola sp., Lamiastrum galeobdolon and
Forest Ecology
227
Table 1 Redundancy analysis of understorey species composition in 156 100 m2 macroplots in the Engestofte–Søholt
forest complex, Denmark (first canonical eigenvalue = 0.085;
Eigenvalues
F ratio = 12.94; P = 0.0001; sum of all canonical eigenvalues = 0.249; F ratio = 3.09; P = 0.0001)
Axis 1
Axis 2
Axis 3
Axis 4
0.085
0.037
0.028
0.022
Species-explanatory variable correlations
0.809
Cumulative variance of species data (%)
8.5
12.1
14.9
17.1
34.0
48.8
59.9
68.8
Cumulative variance of species-explanatory variable relation (%)
0.709
0.664
0.717
Forward selection of variables
Extra fit
F ratio
P-enter
Soil pH
0.05
8.61
0.001
Tree shade
0.03
5.64
0.001
Ancient forest
0.02
4.12
0.001
Soil moisture
0.02
3.05
0.001
Slope
0.01
2.52
0.001
XY
0.01
2.44
0.001
Forest heterogeneity
0.01
2.27
0.001
Litter cover
0.01
2.19
0.003
X2
0.01
2.18
0.001
3
X
Y3
0.01
0.01
1.76
1.90
0.006
0.004
Reclaimed bog
0.01
1.61
0.033
Organic soil
0.01
1.71
0.015
Canopy openness
0.01
0.01
0.022
Y
0.01
1.63
0.020
Note that the sum of the eigenvalues equal the proportion of the total variation explained (ter Braak and Smilauer 2002). The results
refer to the final model produced by forward variable selection with P-to enter \ 0.05. The explanatory variables are listed with the
extra amount of variance in the species data explained, F ratio, and P-value and sorted according to their order of selection
Anemone nemorosa; while those most associated with
reclaimed bog were Rubus caesius, Iris pseudacorus,
Galium palustre,
Potentilla reptans,
Peucedanum palustre, Bromus ramosus, Fragaria vesca and
Lysimachia thyrsiflora (Fig. 4).
Indicator species analysis showed that 10 species
were strong indicators of ancient forest, while 4
species were so for reclaimed bogs and none for other
forest (Table 3a). Eight of the ten ancient-forest, but
none of the reclaimed bog indicator species have also
frequently been noted as associated with ancient forest
in the literature (Table 3a). Three of the reclaimed
bog indicators have high Ellenberg light values and
two have high Ellenberg moisture values (Table 3a).
To assess which species are indicative of ancient
forest, reclaimed bog and other forest, when environmental differences are controlled for, we reclassified
the macroplots according to their sample scores in the
partial RDA with ancient forest and reclaimed bog as
explanatory variables and the 14 environmental
variables set as co-variables (Table 3b). The list of
ancient forest indicators changed little, while just two
species, neither with high Ellenberg moisture values,
continued to be strong indicators of reclaimed bog,
and two indicators (Mercurialis perennis and
Urtica dioica) were added for other forest (Table 3b).
We note that the latter two species were also
associated with other forest before the reclassification,
according to the RDA biplot (Fig. 2).
Discussion
Relative importance of history, modern
environment and spatial location
The PCA-based results showed that the main gradients
in understorey species composition in the Engestofte–
A.G. Van der Valk (ed.)
0.8
228
Reclaimed bog
Y Y3 rub_ida scu_gal lys_vul
bra_syl rub_cae
cal_sep
eup_can
equ_flu
Litter cover
lyt_sal ros_can
oxa_ace des_cae mil_eff
Forest heterogeneity
vib_opu
agr_ten
arc_lap
ste_hol
Soil moisture
Ancient forest vio_sp
ane_nem
sta_syl
gal_odo
cir_lut
lam_gal poa_nem
urt_dio geu_urb
mel_uni car_syl
pri_ela
soil pH
mer_per
-0.6
1.0
46.3****
Pure history (H)
13.0****
Pure space(S)
22.5****
Mixed E and H
Litter cover
Forest heterogeneity
dry_car
con_maj
maj_bif
the_pal
mel_uni
dry_dil
6.8
Mixed H and S
Mixed E, H and S
3.1
6.5
epi_pal
car_pra
car_pan
car_pse
agr_eup
All 14 environmental, 5 historical, and 9 spatial variables were
used. The significance of the three pure fractions was assessed
by partial RDA using 9,999 permutations. The total variation
explained (TVE) is 32.4%
Reclaimed bog
fra_ves
lys_thy
rub_cae
osm_reg
iri_pse
gym_dry
gal_pal
gal_odo
pot_rep
des_cae
bro_ram
cal_can
peu_pal
vio_sp
car_syl
ane_nem
arc_nem
lam_gal
car_rem
dry_car
Ancient forest
ste_hol
cam_tra
oxa_ace
mer_per
urt_dio mel_uni
aeg_pod
X2
rub_cae
bra_syl
vio_sp
soil pH
ste_hol
sta_syl
des_cae
rub_ida
rub_fru
-0.6
dac_glo
Ancient forest
Canopy openness
-0.4
-1.0
2.2
Mixed E and S
Partial RDA axis 2
0.6
Fig. 2 Redundancy analysis of understorey species composition in 156 100-m2 macroplots in the Engestofte–Søholt forest
complex, Denmark: correlation biplot of the first and second
axes. The final RDA model produced by forward variable
selection is shown. Species with a fit range (% variation
explained by the two ordination axes) between 0 and 5% were
removed, as were explanatory variables with interset correlations between -0.2 and 0.2 (ter Braak and Smilauer 2002).
Species acronyms are the first three letters of the genus
followed by the first three letters of species epithet. Continuous
explanatory variables are represented by arrows (indicating
their direction of steepest increase), while binary variables are
shown by symbols indicating their sample score centroid for
samples with the value of 1 (ter Braak and Smilauer 2002)
RDA axis 4
Pure environment (E)
**** P B 0.0001
RDA axis 1
Reclaimed bog
TVE (%)
1.0
-0.6
Slope
XY
Tree shade
Fractions
-0.4
RDA axis 2
cal_can
Table 2 Partitioning of the variation in understorey species
composition in 156 100-m2 macroplots in the Engestofte–
Søholt forest complex, Denmark into seven mutually exclusive
fractions using RDA and partial RDA
RDA axis 3
Partial RDA axis 1
0.6
0.6
Fig. 3 Redundancy analysis of understorey species composition in 156 100-m2 macroplots in the Engestofte–Søholt forest
complex, Denmark: correlation biplot of the third and fourth
axes. The final RDA model produced by forward variable
selection is shown. See Fig. 2 for further details
Fig. 4 Partial redundancy analysis of understorey species
composition in 156 100-m2 macroplots in the Engestofte–
Søholt forest complex with ancient forest and reclaimed bog as
explanatory variables and the 14 environmental set as
covariables: correlation biplot of the first and second axes.
Species with a fit range between 0% and 2% are not shown
Forest Ecology
229
Table 3 Indicator species analysis for (a) ancient forest (high
forest in 1805 AD), reclaimed bog, and other forest, and (b)
based on the sample scores from the partial RDA with ancient
(a) Ancient forest (n = 69)
Reclaimed bog (n = 23)
Carex sylvatica (56.6****)A
Melica uniflora (46.4**)
A
L, M
A
Oxalis acetosella (44.9**)
Other forest (n = 64)
Rubus caesius (43.0***)L
Lamiastrum galeobdolon (53.6****)
Deschampsia caespitosa (47.0**)
forest and reclaimed bog as explanatory variables and the 14
environmental variables set as covariables (Fig. 4)
Brachypodium sylvaticum (42.7**)a
Iris pseudacorus (29.7***)L,
M
Calamagrostis canescens (24.5*)L,
M
A
Stellaria holostea (44.0**)A
Viola sp. (42.4**)A
Anemone nemorosa (35.2**)A
Poa nemoralis (29.4*)a
Luzula pilosa (24.5*)A
(b) ‘‘Pure’’ ancient forest (n = 81)
‘‘Pure’’ reclaimed bog (n = 35)
Lamiastrum galeobdolon (58.2****)A
Rubus caesius (38.1***)L
A
Mercurialis perennis (37.2****)A
a
Carex sylvatica (52.3****)
Melica uniflora (45.7**)
Other forest (n = 40)
Brachypodium sylvaticum (36.9**)
Urtica dioica (34.1**)M
A
Galium odoratum (45.3**)A
Oxalis acetosella (40.8**)A
Deschampsia caespitosa (33.5*)L,
Stellaria holostea (43.2**)
M
A
Viola sp. (42.7***)A
Anemone nemorosa (28.6*)A
Following Fig. 4, ‘‘Pure’’ ancient forest are plots with positive axis 1 scores, while ‘‘Pure’’ reclaimed bog are plots with negative
scores on axis 1 and positive scores on axis 2. Only species with indicator values C25% perfect indication at P \ 0.05 (based on
9,999 Monte Carlo permutations) are shown
*P \ 0.05
**P \ 0.01
***P \ 0.001
****P B 0.0001
a
Listed as ancient-forest species in Hermy et al. (1999), but cited as such by \5 of the 22 publications considered by that study
A
Listed as ancient-forest species by Hermy et al. (1999) and cited as such by C5 of the 22 publications considered
L
Ellenberg light indicator value C6 (Ellenberg et al. 1992)
M
Ellenberg moisture indicator value C6 (Ellenberg et al. 1992)
Søholt forest complex are strongly related to the
environmental, historical, and spatial descriptors that
were included in the study. The RDA forward selection
results indicated that modern environment is most
important, with four of the five first-selected variables
being environmental (Table 1). The RDA-based variation partitioning further confirmed the dominant role
played by modern environment as a driver of species
distributions, with the pure environmental fraction,
constituting nearly 50% of the TVE (Table 2). Hereby,
our results agree with the many previous studies, which
have shown modern environment to be a strong control
of landscape- and local-scale understorey species
distributions in temperate forests in Europe and North
America (e.g. Motzkin et al. 1999; Svenning and Skov
2002; Gilbert and Lechowicz 2004; Graae et al. 2004;
Borchsenius et al. 2004; Thomsen et al. 2005). Our
results indicate that soil pH and tree shade, i.e. the
extent to which the overstorey is dominated by heavily
shade-producing trees, notably Fagus sylvatica, and
soil moisture are the most important environmental
drivers of understorey species distributions. The
230
importance of soil pH and/or soil moisture is a frequent
pattern at local and landscape scales in temperate
forests (e.g. Motzkin et al. 1999; Richard et al. 2000;
Verheyen and Hermy 2001a; Svenning and Skov 2002;
Graae et al. 2004; Borchsenius et al. 2004; Windeballe
et al. 2004; Thomsen et al. 2005). In the Engestofte–
Søholt forest complex macroplots with a high tree
shade value was generally dominated by Fagus sylvatica, which apart from producing heavy shade also
produces acidic, nitrogen-poor litter (Ellenberg 1988;
Neirynck et al. 2000; Hagen-Thorn et al. 2004). Previous studies have documented clear differences in
understorey species composition between stands dominated by Fagus sylvatica and stands dominated by
other tree species (e.g. Graae and Heskjær 1997;
Svenning and Skov 2002; Graae et al. 2004). In
temperate deciduous forests, overstorey tree species
composition may be more important in managed
forests, such as the Engestofte–Søholt complex,
mainly composed of mono- and oligospecific stands
(Svenning and Skov 2002) than in mixed-canopy
natural forests (Thomsen et al. 2005).
While modern environment emerged as the dominant control of understorey species composition, two
historical and five spatial variables were also
included in the final RDA model produced by
forward selection (Table 1). Notably, the historical
variable ancient forest was the third explanatory
selected by the forward selection procedure
(Table 1). Furthermore, the RDA-based variation
partitioning produced pure spatial and pure historical
fractions that were also highly significant and nonnegligible, constituting 23% and 13% of the TVE,
respectively (Table 2). The spatial variables could
reflect unmeasured broadscale variation in the modern environment or historical factors, e.g. natural
dispersal-generated patterns reflecting internal landscape-scale dispersal dynamics (Svenning and Skov
2002; Miller et al. 2002) or regional-scale migration
history (Popiela 2004; Van der Veken et al. 2007;
Svenning et al. 2008).
Which historical factors are important?
The analyses of the PCA sample scores showed that
ancient forest and reclaimed bog status were strongly
associated with the main gradients in understorey
species composition in the Engestofte–Søholt forest
complex, while being near or far from burial mounds
A.G. Van der Valk (ed.)
were of less importance, and neither stand age nor
converted conifer plantation status were of any
importance at all. These results were confirmed by
the RDA (Table 1) and partial RDA analyses, in
which only ancient forest and reclaimed bog were
selected as significant, non-redundant predictors.
Both analyses indicated ancient-forest status as more
important than reclaimed bog status. In a similar vein,
indicator species analysis showed that many more
species were strong indicators of ancient forest than
of reclaimed bog (Table 3).
The fact that eight of the ten ancient-forest
indicators (Table 3) have frequently been reported
as associated with ancient forest in earlier publications (Hermy et al. 1999) provide strong external
support for the interpretation of the ancient-forest
effect as a being to a large extent a historical legacy
effect (Gilliam 2007). The variation partitioning
shows that the historical effect on species compositions consists of a pure fraction of 13% TVE and
three mixed fractions of totalling 12% TVE, hereof
nearly 9% involving modern environment (Table 2).
These mixed historical–environmental (–spatial) fractions may reflect persistent environmental changes
induced by past history. A number of studies have
shown that past land-use can cause edaphic changes
that persist hundreds or even [1,000 years after
reforestation has taken place (Kristiansen 2001;
Dupouey et al. 2002; Dambrine et al. 2007; also cf.
Brunet 2007). Furthermore, past land-use may cause
long-term changes in overstorey tree species composition (Peterken 1996; Motzkin et al. 1999;
Bellemare et al. 2002), which again may induce soil
changes (Ellenberg 1988; Neirynck et al. 2000;
Hagen-Thorn et al. 2004). However, the large pure
historical fraction of the variation in species composition is suggestive of purely population dynamic
legacies. The association of a number of well-known
ancient-forest species with areas of ancient forest, i.e.
1,805 high forest, even when environmental factors
are controlled for (Table 3b), is also suggestive of
dispersal limitation. Dispersal limitation has been
shown by many studies to pose a strong constraint on
the rate at which understorey plants recolonize
secondary forest (e.g. Peterken and Game 1984;
Matlack 1994; Honnay et al. 2002; Bellemare et al.
2002; Gilliam 2007) and a recent review concluded
that dispersal of diaspores appears to be the most
critical step in the colonization of young forests by
Forest Ecology
ancient-forest plant species (Hermy and Verheyen
2007). Furthermore, except for two species (Galium odoratum, Poa nemoralis) all the ancient-forest
indicators in the present study have dispersal modes
(myrmeco-, baro- or autochory) associated with slow
recolonization rates of secondary forests (Matlack
1994; Brunet and von Oheimb 1998; Bossuyt et al.
1999; Bellemare et al. 2002; Brunet 2007) and
particularly aggregated distributions within older
forests (Svenning and Skov 2002; Miller et al. 2002).
The RDA and partial RDA results suggest that
reclaimed bogs differ from other plots in understorey
species composition beyond what can be explained by
modern environment (Table 1; Figs. 2, 4). Overall,
the species associated with the reclaimed bog plots
after controlling for modern environment are a
mixture of wet-soil and high-light species (Fig. 4).
Considering the reclaimed bog indicator species, the
wet-soil species Iris pseudacorus and Calamagrostis canescens could be interpreted as evidence that
plots on reclaimed bog still somehow differ hydrologically from other plots (Table 3a). However,
among the only two strong indicator species remaining after controlling for modern environment
(Table 3b), Rubus caesius has no specific soil moisture preferences and Brachypodium sylvaticum
prefers fresh, but not wet soils (Ellenberg moisture
value = 5), just as most of the ancient-forest indicators in Table 3 (Ellenberg et al. 1992). Hence, it
seems likely that at least part of the reclaimed bog
effect reflect a non-environmental historical legacy.
Rubus caesius, but not Brachypodium sylvaticum has
relatively high-light requirements (Ellenberg et al.
1992). It is noteworthy that both species are likely to
have relatively efficient dispersal, being endo- and
epizoochorously dispersed by vertebrates, respectively (Hodgson et al. 1995; Hermy et al. 1999) and
are therefore expected to be relatively fast colonizers
of secondary forests (Brunet 2007). The abundance of
Brachypodium sylvaticum is consistent with opportunistic expansion in the absence of more slowly
dispersing competitors, while the abundance of
Rubus caesius and other high-light species (Fig. 4)
could reflect either opportunistically expanding populations or remnant populations, (Eriksson 1996)
slowly declining after the initially open conditions.
In a Swedish study, Brachypodium sylvaticum was
similarly found to be more abundant in plantations
adjacent to ancient forest than in the latter itself
231
(Brunet 2007). The association of a number of wetsoil plants with reclaimed bog even after controlling
for modern environment (Fig. 4) could also reflect
their persistence in formerly wet areas as remnant
populations (Eriksson 1996). Considering the two
indicator species of the other young forests, i.e. forests
that were neither bogs nor high forest in the early
1800s, Urtica dioica is well known to be favoured by
anthropogenic disturbance (e.g. Bossuyt et al. 1999;
Verheyen and Hermy 2001b), while Mercurialis perennis has often been found to be associated with
ancient forest (Hermy et al. 1999). However, in
England Mercurialis perennis has been shown to
have rather efficiently recolonized 19th century
secondary forests within 0.5 km of ancient-forest
source populations (Peterken and Game 1981) and to
be less constrained to ancient forests than many other
ancient-forest species (Peterken and Game 1984).
Notably, in the latter study Mercurialis perennis had
only 54% of its localities in ancient woods, while the
10 ancient forest indicator species in Table 3 that are
also listed as such in Hermy et al. (1999) had 58–97%
(mean = 80%) of their localities in ancient woods.
Furthermore, Mercurialis perennis may sometimes
even occur in tall-herb and grassland vegetation
(Jefferson 2008), so it may well have been present
outside the high-forest areas in 1805. Although the
abundance of Urtica dioica could reflect soil changes
not captured by our environmental descriptors (notably phosphate enrichment: Pigott 1971; Verheyen and
Hermy 2001b), the abundance of Mercurialis perennis is more suggestive of opportunistic expansion, as
discussed for Brachypodium sylvaticum. Most of the
other species associated with other forest according to
the RDA biplots (Figs. 2, 4) are species with
relatively good dispersal abilities, being epizoochorous
(Arctium nemorosum,
Circaea lutetiana,
Geum urbanum, Stachys sylvatica) or having tiny
wind-dispersed spores (Dryopteris carthusiana). Several previous studies have also found forest species
with dispersal modes indicative of good dispersal
abilities to be overrepresented in young forests (e.g.
Matlack 1994, Brunet 2007).
While modern understorey plant species distributions have sometimes been linked to land-use
[1,500 years back, no unique effect of being located
in the vicinity of Bronze Age burial mounds was
detected. This contrasts the evidence for persistent
floristic effects on Roman agriculture on the
232
understory vegetation in French forests, 1,700 years
or more post-abandonment (Dupouey et al. 2002;
Dambrine et al. 2007). Plausible explanations for
these discrepancies include that later anthropogenic
disturbances in the Danish study area erased all
Bronze Age legacies, that the 2,500 years elapsed
since the Bronze Age have been long enough to allow
complete recolonization or, more generally, for any
legacies to persist, or that the Bronze Age disturbances in Denmark were of much milder than the
Roman agricultural impacts in France. While pre1800 forest management clearly has strong impacts on
the understorey species composition via its effects on
modern environment (tree stand structure and composition, edaphic conditions), only artificial drainage,
but neither stand age or the past presence of conifer
plantations had any effect. Apparently these forestryrelated disturbances have not been severe enough to
produce persistent changes in the understorey species
composition, as also found for stand age in another
Danish forest area (Svenning and Skov 2002). Similarly, although some North American studies have
suggested that there are long-term forestry-induced
legacy effects, these results have been questioned and
conflicting patterns have been found, causing a recent
review to conclude that more research is needed to
clarify this issue (Roberts and Gilliam 2003).
Concluding remarks
Our results emphasize the importance of historical
factors and associated legacy effects for understanding modern vegetation patterns in forested
landscapes. While modern environment is the main
determinant of understorey species distributions in
Engestofte–Søholt forest complex, past land-use also
continues to exert important influence. Hereby, our
study adds to the previous studies that have shown
past land-use to be an important determinant of
understorey plant species community composition
and diversity in temperate forests of Europe and
North America (reviewed in Hermy and Verheyen
(2007) and Gilliam (2007)) and other forest ecosystems and regions, e.g. Central American tropical
forest (Svenning et al. 2004). The Engestofte–Søholt
forests have, as the large majority of Danish forests,
been strongly affected by modern forestry. Nevertheless, the strong historical legacies found relate
more to long-term forest continuity than to forest
A.G. Van der Valk (ed.)
management. Many typical forest interior species
were still concentrated in areas that were high forest
in 1805 AD, probably in large part due to dispersal
limitation, despite the fact that most of the areas
studied were already afforested by the mid-1800s.
Among the younger forests, there were clear floristic
differences between those on artificially drained bogs
and those not. Both included understorey species
capable of relatively efficient recolonization, but
apparently remnant populations of wet-soil plants
were still present in the reclaimed bog areas, too. The
observation that strong effects of past deforestation
on the understorey vegetation can persist for several
hundred years even within contiguous forests emphasises the conservation value of ancient forests.
Acknowledgements We thank the Danish Natural Science
Research Council (grants 21-01-0412 and 21-04-0346 to JCS)
for economic support, Birgitte S. Windeballe, Karen C. Larsen,
Henrik Baktoft, Egon Ronge Hansen and the late Jonas E.
Lawesson for assistance and support, Benjamin Øllgaard,
Simon Lægaard and Jens-Christian Schou for their expert help
with species identifications, Ragna Nielsen and the forest
kindergarden of Holeby for accommodation.
References
Aude E, Lawesson J (1998) Vegetation in Danish beech forests:
the importance of soil, microclimate and management
factors, evaluated by variance partitioning. Plant Ecol 134:
53–65. doi:10.1023/A:1009720206762
Bellemare J, Motzkin G, Foster DR (2002) Legacies of the
agricultural past in the forested present: an assessment of
historical land-use effects in rich mesic forests. J Biogeogr
29:1401–1420. doi:10.1046/j.1365-2699.2002.00762.x
Björse G, Bradshaw R (1998) 2000 years of forest dynamics in
southern Sweden: suggestions for forest management. For
Ecol Manage 104:15–26
Borcard D, Legendre P, Drapeau P (1992) Partialling out the
spatial component of ecological variation. Ecol 73:1045–
1055. doi:10.2307/1940179
Borchsenius F, Nielsen PK, Lawesson JE (2004) Vegetation
structure and diversity of an ancient temperate deciduous
forest in SW Denmark. Plant Ecol 175:121–135. doi:
10.1023/B:VEGE.0000048095.29961.c5
Bossuyt B, Hermy M, Deckers J (1999) Migration of herbaceous plant species across ancient-recent forest ecotones
in central Belgium. J Ecol 87:628–638. doi:10.1046/
j.1365-2745.1999.00379.x
Bradshaw R, Holmqvist BH (1999) Danish forest development
during the last 3000 years reconstructed from regional
pollen data. Ecography 22:53–62. doi:10.1111/j.16000587.1999.tb00454.x
Brown N, Jennings S, Wheeler P, Nabe-Nielsen J (2000) An
improved method for the rapid assessment of forest
Forest Ecology
understorey light environments. J Appl Ecol 37:1044–
1053. doi:10.1046/j.1365-2664.2000.00573.x
Brunet J (1992) Impact of grazing on the field layer vegetation
in a mixed oak forest in south Sweden. Sven Bot Tidskr
86:347–353
Brunet J (2007) Plant colonization in heterogeneous landscapes: an 80-year perspective on restoration of
broadleaved forest vegetation. J Appl Ecol 44:563–572.
doi:10.1111/j.1365-2664.2007.01297.x
Brunet J, von Oheimb G (1998) Migration of vascular plants to
secondary woodlands in southern Sweden. J Ecol 86:429–
438. doi:10.1046/j.1365-2745.1998.00269.x
Cushman SA, Wallin DO (2002) Separating the effects of
environmental, spatial and disturbance factors on forest
community structure in the Russian Far East. For Ecol
Manage 168:201–215
Dambrine E, Dupoey JL, Laüt L, Humbert L, Thinon M,
Beaufils T, Richard H (2007) Present forest biodiversity
patterns in France related to former Roman agriculture.
Ecology 88:1430–1439. doi:10.1890/05-1314
Diniz-Filho JAF, Bini LM, Hawkins BA (2003) Spatial autocorrelation and red herrings in geographical ecology. Glob
Ecol Biogeogr 12:53–64. doi:10.1046/j.1466-822X.2003.
00322.x
Dufrêne M, Legendre P (1997) Species assemblages and
indicator species: the need for a flexible asymmetrical
approach. Ecol Monogr 67:345–366
Dupouey JL, Dambrine E, Laffite JD, Moares C (2002) Irreversible impact of past land use on forest soils and
biodiversity. Ecology 83:2978–2984
Dupré C, Ehrlén J (2002) Habitat configuration, species traits
and plant distributions. J Ecol 90:796–805. doi:10.1046/
j.1365-2745.2002.00717.x
Ellenberg H (1988) Vegetation ecology of Central Europe.
Cambridge University Press, Cambridge
Ellenberg H, Weber HE, Düll R, Wirth V, Werner W, Paulißen
D (1992) Zeigerwerte von Pflanzen in Mitteleuropa.
Scripta Geobotanica 18, 2nd edn. Verlag Eich Goltze KG,
Göttingen
Eriksson O (1996) Regional dynamics of plants: a review of
evidence for remnant, source-sink and metapopulations.
Oikos 77:248–258. doi:10.2307/3546063
Fritzbøger B (1994) Kulturskoven: Dansk skovbrug fra oldtid
til nutid. Gyldendal, Copenhagen
Gilbert B, Lechowicz MJ (2004) Neutrality, niches, and dispersal in a temperate forest understory. Proc Natl Acad
Sci USA 101:7651–7656. doi:10.1073/pnas.0400814101
Gilliam FS (2007) The ecological significance of the herbaceous layer in temperate forest ecosystems. Bioscience
57:845–858. doi:10.1641/B571007
Graae BJ, Heskjær VS (1997) A comparison of understorey
vegetation between untouched and managed deciduous
forest in Denmark. For Ecol Manage 96:111–123
Graae BJ, Økland RH, Petersen PM, Jensen K, Fritzbøger B
(2004) Influence of historical, geographical and environmental variables on understorey composition and richness
in Danish forests. J Veg Sci 15:465–474. doi:10.1658/11009233(2004)015[0465:IOHGAE]2.0.CO;2
Hagen-Thorn A, Callesen I, Armolaitis K, Nihlgard B (2004)
The impact of six European tree species on the chemistry
233
of mineral topsoil in forest plantations on former agricultural land. For Ecol Manage 195:373–384
Hansen K (ed) (1984) Dansk Feltflora. Nordisk Forlag,
Copenhagen
Hermy M, Honnay O, Firbank L, Grashof-Bokdam C, Lawesson J (1999) An ecological comparison between ancient
and other forest plant species of Europe, and the implications for forest conservation. Biol Conserv 91:9–22. doi:
10.1016/S0006-3207(99)00045-2
Hermy M, Verheyen K (2007) Legacies of the past in the
present-day forest biodiversity: a review of past land-use
effects on forest plant species composition and diversity.
Ecol Res 22:361–371. doi:10.1007/s11284-007-0354-3
Hodgson JG, Grime JP, Hunt R, Thompson K (1995) The
electronic comparative plant ecology. Chapman & Hall,
London
Honnay O, Hermy M, Coppin P (1999) Impact of habitat
quality on forest plant species colonization. For Ecol
Manage 115:157–170
Honnay O, Verheyen K, Butaye J, Jacquemyn H, Bossuyt B,
Hermy M (2002) Possible effects of habitat fragmentation
and climate change on the range of forest plant species.
Ecol Lett 5:525–530. doi:10.1046/j.1461-0248.2002.
00346.x
Høy T, Dahl J (2000) Danmarks søer—søerne i Storstrøms
Amt og på Bornholm. Strandbergs Forlag, Birkerød
Jefferson RG (2008) Biological flora of the British Isles:
Mercurialis perennis L. J Ecol 96:386–412. doi:10.1111/
j.1365-2745.2007.01348.x
Kristiansen SM (2001) Present-day soil distribution explained
by prehistoric land-use: Podzol–Arenosol variation in an
ancient woodland in Denmark. Geoderma 103:273–289.
doi:10.1016/S0016-7061(01)00044-1
Laursen J (1994) Historien i skoven—skove og fortidsminder i
Århus Amt. Skippershoved, Århus Amtsmuseumsråd
Lawesson JE, Eilertsen O, Diekmann M, Reinikainen A,
Gunnlaugsdóttir E, Fossa AM, Carøe I, Skov F, Groom G,
Økland T, Økland R, Andersen PN, Bakkestuen V (2000)
A concept for vegetation studies and monitoring in the
Nordic countries. TemaNord 2000: 517. Nordic Council
of Ministers, Copenhagen
Legendre P, Gallagher ED (2001) Ecologically meaningful
transformations for ordination of species data. Oecologia
129:271–280. doi:10.1007/s004420100716
Legendre P, Legendre L (1998) Numerical ecology, 2nd edn.
Elsevier, Amsterdam
Matlack GR (1994) Plant species migration in a mixed-history
forest landscape in eastern North America. Ecol 75:1491–
1502. doi:10.2307/1937472
McCune B, Grace JB (2002) Analysis of ecological communities. MjM Software Design, Gleneden Beach, Oregon
McCune B, Mefford MJ (1999) PC-ORD. Multivariate analysis
of ecological data, version 4. MjM Software Design,
Gleneden Beach, Oregon
Miller TF, Mladenoff DJ, Clayton MK (2002) Old-growth
northern hardwood forests: spatial autocorrelation and
patterns of understory vegetation. Ecol Monogr 72:487–
503
Motzkin G, Wilson P, Foster DR, Allen A (1999) Vegetation
patterns in heterogenous landscapes: the importance of
234
history and environment. J Veg Sci 10:903–920. doi:
10.2307/3237315
Neirynck J, Mirtcheva S, Sioen G, Lust N (2000) Impact of
Tilia platyphyllos Scop., Fraxinus excelsior L., Acer
pseudoplatanus L., Quercus robur L. and Fagus sylvatica
L. on earthworm biomass and physico-chemical properties of a loamy topsoil. For Ecol Manage 133:275–286
Nørrevang A, Lundø J (eds) (1980) Danmarks natur 6: Skovene. Politikens Forlag, Copenhagen
Økland RH (1990) Vegetation ecology: theory, methods and
applications with reference to Fennoscandia. Sommerfeltia Supplement 1:1–233
Økland RH (1999) On the variation explained by ordination
and constrained ordination axes. J Veg Sci 10:131–136.
doi:10.2307/3237168
Økland RH, Eilertsen O (1994) Canonical correspondence
analysis with variation partitioning: some comments and
an application. J Veg Sci 5:117–126. doi:10.2307/
3235645
Peterken GF (1996) Natural woodland: ecology and conservation in northern temperate regions. Cambridge
University Press, Cambridge
Peterken GF, Game M (1981) Historical factors affecting the
distribution of Mercurialis perennis in Central Lincolnshire. J Ecol 69:781–796. doi:10.2307/2259636
Peterken GF, Game M (1984) Historical factors affecting the
number and distribution of vascular plant species in the
woodlands of central Lincolnshire. J Ecol 72:155–182.
doi:10.2307/2260011
Pigott CD (1971) Analysis of the response of Urtica dioica to
phosphate. New Phytol 70:953–966. doi:10.1111/j.14698137.1971.tb02597.x
Popiela A (2004) Phytogeographic aspects of the occurrence of
forest vascular plant species in Pomerania (northwest
Poland). Bot Jahrb Syst Pflanzengesch Pflanzengeogr
125:97–228. doi:10.1127/0006-8152/2004/0125-0097
Rasmussen HW (1966) Danmarks geologi. Gjellerup, Tønder
Richard M, Bernhardt T, Bell G (2000) Environmental heterogeneity and the spatial structure of fern species diversity
in one hectare of old-growth forest. Ecography 23:231–
245. doi:10.1111/j.1600-0587.2000.tb00279.x
Roberts MR, Gilliam FS (2003) Response of the herbaceous
layer to disturbance in eastern forests. In: Gilliam FS,
A.G. Van der Valk (ed.)
Roberts MR (eds) The herbaceous layer in forests of
eastern North America. Oxford University Press, Oxford,
pp 302–320
Schierup H-H, Jensen A (1979) Vejledning i kemisk og fysisk
analyse af jordprøver og plantemateriale. Botanical
Institute, Aarhus University, Aarhus
Svenning J-C, Kinner DA, Stallard RF, Engelbrecht BMJ,
Wright SJ (2004) Ecological determinism in plant community structure across a tropical forest landscape.
Ecology 85:2526–2538. doi:10.1890/03-0396
Svenning J-C, Normand S, Skov F (2008) Postglacial dispersal
limitation of widespread forest plant species in nemoral
Europe. Ecography 31:316–326. doi:10.1111/j.0906-7590.
2008.05206.x
Svenning J-C, Skov F (2002) Mesoscale distribution of understorey plants in temperate forest (Kalø, Denmark): the
importance of environment and dispersal. Plant Ecol
160:169–185. doi:10.1023/A:1015885321263
ter Braak CJF, Smilauer P (2002) CANOCO reference manual
and CanoDraw for Windows user’s guide: Software for
canonical community ordination (version 4.5). Microcomputer Power, Ithaca, New York
Thomsen RP, Svenning J-C, Balslev H (2005) Overstorey
control of understorey species composition in a nearnatural temperate broadleaved forest in Denmark. Plant
Ecol 181:113–126. doi:10.1007/s11258-005-3996-7
Van der Veken S, Rogister J, Verheyen K, Hermy M, Nathan R
(2007) Over the (range) edge: a 45-year transplant
experiment with the perennial forest herb Hyacinthoides
non-scripta. J Ecol 95:343–351. doi:10.1111/j.13652745.2006.01208.x
Verheyen K, Hermy M (2001a) The relative importance of
dispersal limitation of vascular plants in secondary forest
succession in Muizen Forest, Belgium. J Ecol 89:829–
840. doi:10.1046/j.0022-0477.2001.00596.x
Verheyen K, Hermy M (2001b) An integrated analysis of the
spatio-temporal colonization patterns of forest plant species. J Veg Sci 12:567–578. doi:10.2307/3237008
Windeballe BS, Svenning J-C, Balslev H (2004) The influence
of past land-use on understory plant distributions in a
near-natural deciduous forest in Denmark. Nord J Bot
23:69–81. doi:10.1111/j.1756-1051.2003.tb00370.x
Short-term responses of the understory to the removal
of plant functional groups in the cold-temperate
deciduous forest
Alexandre Lenière Æ Gilles Houle
Originally published in the journal Plant Ecology, Volume 201, No. 1, 235–245.
DOI: 10.1007/s11258-008-9545-4 Springer Science+Business Media B.V. 2008
Abstract It is widely believed that functional
diversity contributes to the stability of ecosystems.
Indeed, greater redundancy among species within
functional groups and greater complementarity
among functional groups within communities should
increase the resistance and resilience of ecosystems.
In the present study, we tested for functional group
complementarity by examining how the loss of
specific functional groups may alter the role that
other groups play in ecosystem functions. We
removed different functional groups, one at a time,
from the understory of three maple-dominated forests
in southern Québec (Canada) and followed the
understory response over a 2-year period. The
experimental design included a control and five
removal treatments. Five functional groups were
defined: spring-flowering ephemeral species; springflowering persistent species; summer-flowering species; fern species; and seedlings and juveniles of
woody species. Richness, cover, soil pH and organic
matter content were determined after two years of
removal. The results of our experiment revealed that
richness was significantly lower than what we
expected when spring-flowering persistent species
or seedlings and juveniles of woody species were
removed, suggesting not only direct but also indirect
positive effects of both of these groups on understory
richness (mostly through effects on summer-flowering species and fern species). Removal of the
seedlings and juveniles of woody species and, to a
lesser extent, of spring-flowering persistent species
and of fern species lead to a decrease in the cover of
summer-flowering species, implying a positive effect
of the former groups on the latter. The cover–richness
relationship in the control and in each one of the five
removal treatments was positive and well fitted by a
linear regression. Yet, the slope of the relationship
differed among treatments, but not between the
control and any one of the removal treatments
(pair-wise comparisons). Our results suggest that
the different functional groups are complementary
and that positive interactions predominate over
negative ones. Contrary to common belief, understory
plants can respond quite rapidly to changes in
community functional composition. Although we
have not investigated the specific mechanisms
responsible for the short-term responses reported
here, we suggest that complex intergroup interactions
may favour functional diversity and enhance ecosystem functions.
A. Lenière G. Houle (&)
Département de biologie, Université Laval, Québec,
QC, Canada G1V 0A6
e-mail: gilles.houle@bio.ulaval.ca
Keywords Cold-temperate deciduous forest
Complementarity Cover-richness relationship
Experimental removal Functional diversity
Functional group Interdependence
Understory
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_18
235
236
Introduction
Species that share physiological, morphological and/
or phenological characteristics form distinct functional groups (Naeem et al. 1999) and, within such
groups, species can be said to be functionally
redundant. Indeed, the loss of one or a few species
within a given group can be compensated for by other
functionally similar species (Wardle et al. 1999); as a
result, ecosystem functions can be maintained
because of such functional redundancy (but see
Balvanera et al. 2006).
Because of their distinct physiological, morphological and/or phenological characteristics, functional
groups can partition resources among themselves
(Fitter 1982; Hooper et al. 2002); such resource
partitioning may contribute to reduce competitive
exclusion (Cardinale et al. 2000) and increase overall
ecosystem functions (Tilman 1997). However, the
loss of a given functional group can have negative
effects on an ecosystem and may even lead to a
slowing down or a collapse in its functioning (Loreau
et al. 2001). While trying to determine the consequences of the loss of groups of species on the
functioning of ecosystems, several studies have
shown two contrasted properties of functional groups:
complementarity (i.e. positive intergroup interactions) and dominance (i.e. negative intergroup
interactions; Naeem et al. 1999; Wardle et al. 1999;
Cardinale et al. 2000; Dı́az and Cabido 2001). These
two properties are believed to be largely responsible
for the specific form of the relationship between
diversity and production (Grime 1973, 1977; Mittelbach et al. 2001; Tilman et al. 2002; Symstad et al.
2003). However, complementarity and dominance are
not necessarily opposing properties: some functional
groups may be complementary and particular subsets
of complementary species may still be dominant over
others (Loreau et al. 2001; Fox 2005).
Many papers, both empirical and theoretical, have
reported that a decrease of diversity can alter ecosystem properties and functions (Naeem et al. 1995;
Wardle et al. 1999; Loreau et al. 2001; Lepš 2004).
Several empirical studies on this subject have been
performed under controlled conditions, with synthetic, simplified assemblages (Tilman et al. 1997;
Symstad et al. 1998; Hector et al. 1999; Naeem 2001;
Hooper et al. 2002). In these, diversity was manipulated (independent variable) and ecosystem response
A.G. Van der Valk (ed.)
was measured through changes in primary production.
From their review of the literature on the subject,
Loreau et al. (2001) concluded that the relationship
between diversity and ecosystem functions could be
shown under ‘‘natural’’ conditions using two types of
approach: (1) an experimental approach, e.g. one in
which species or groups of species are removed, and
(2) a comparative approach, e.g. one in which
variation in factors other than diversity is controlled
for (Huston 1997; Wardle et al. 1999; Troumbis and
Memtsas 2000; Buonopane et al. 2005).
In the present study, we use an experimental
approach, based on the removal of groups of functionally similar species, to better understand the
relationship between functional diversity and ecosystem functioning. Our objective is to determine how
the loss of functionally similar species can affect the
properties of the understory, in three deciduous
forests of southern Québec (Canada). Our study is
based on the premise that both positive and negative
intergroup interactions may be present and that the
loss of a given group may have significant effects on
the other groups and on understory production.
Material and methods
Study site
The experiment was conducted in three forest fragments of the Bois-Francs region, two at SaintGrégoire (4617.8200 N, 7230.7400 W; 4617.5250
N, 7231.0760 W) and one at Sainte-Françoise
(4629.3410 N, 7156.1560 W), on the south shore
of the St. Lawrence River, between Trois-Rivières
and Quebec City (Québec, Canada). The dominant
tree species is sugar maple (Acer saccharum Marsh.),
along with American beech (Fagus grandifolia
Ehrh.), American linden (Tilia americana L.), American hop-hornbeam [Ostrya virginiana (Mill.) K.
Koch], yellow birch (Betula alleghaniensis Britton),
eastern hemlock (Tsuga canadensis L.) and balsam fir
(Abies balsamea L.). The region is part of the Great
Lakes–St. Lawrence forest region of Rowe (1972),
subsections Mid St. Lawrence (L-2) and High St.
Lawrence (L-3).
At the nearby Trois-Rivières and Quebec City
weather stations, annual precipitation totals 1100 mm
(24% as snow) and 1230 mm (38% as snow),
Forest Ecology
respectively, and annual mean daily temperature is
4.9C and 4.0C, respectively (URL: www.climate.
weatheroffice.ec.gc.ca). The soils of the region are
mostly brunisols except in low areas that are characterized by gleysols (Choinière and Laplante 1948).
All three forests have been lightly managed for sap
over several years, although one of the sites at SaintGrégoire is protected by the Québec government and
has not been exploited since at least 1975. The size of
the larger trees on the different sites suggest a
minimum stand age of *200 years (old, secondgrowth stands). Site selection was based on the
following criteria: relatively uniform topographical
features; easy road access; plant species richness
representative of the region; different plant functional
groups present; permission (from the owners or the
Québec government) to carry out the experiment.
Sampling protocol and variables measured
In the spring of 2003, we established 30 2 m 9 2 m
quadrats at each one of the three study sites. Within
sites, quadrats were positioned under homogeneous
canopy composition and microtopography. Species
richness was determined for each quadrat at three
different times during the growing season (spring,
summer and fall) and assigned to one of five
functional groups, based on flowering period and
growth form (Bratton 1976): spring-flowering ephemeral species, e.g. Erythronium americanum KerGawl.; spring-flowering persistent species, e.g. Trillium erectum L.; summer-flowering species, e.g.
Epipactis helleborine (L.) Crantz; ferns, e.g. Dryopteris spinulosa (O.F. Muell.) Watt; seedlings and
juveniles of woody species, e.g. Acer saccharum
Marsh. (nomenclature follows Marie-Victorin 2002).
We considered spring-flowering ephemeral species
separately from spring-flowering persistent species
because of the important role that they play in the
nutrient dynamics of deciduous forests. Indeed,
spring-flowering ephemeral species are believed to
store in their tissues the nutrients flushed into the
system during snowmelt and to release them progressively as they senesce (vernal dam hypothesis:
Muller and Bormann 1976; Eickmeier and Schussler
1993; Tessier and Raynal 2003). The interest in
considering fern species and seedlings and juveniles
of woody species comes from the significant role that
they play in primary production and in competition in
237
the forest understory. Grasses and sedges were not
abundant on the study sites and were included in the
summer-flowering species.
For the experiment, we chose 18 quadrats per site,
the most similar in terms of species richness and
microtopography, among the 30 initially marked.
Within sites, quadrats were grouped in three blocks
on the basis of their spatial proximity. Treatments
were assigned randomly to the quadrats within each
block and were as follows: control; removal of
spring-flowering ephemeral species; removal of
spring-flowering persistent species; removal of summer-flowering species; removal of fern species;
removal of seedlings and juveniles of woody species.
Only aboveground parts were removed so as to avoid
soil disturbance. Removals took place twice monthly,
between May and August, in 2004 and again in 2005.
We are aware that the persistence of underground
structures (bulbs, rhizomes or corms) and an increase
in the decomposition of roots following the removal
of aboveground parts may influence the responses to
our treatments. However, the method we used is
common in the literature and offers the advantage of
avoiding the confounding effects of soil disturbance
to species removals (Wardle et al. 1999; Buonopane
et al. 2005; Wardle and Zackrisson 2005).
Biological variables
Floristic surveys (vascular plants \1 m in height)
were done three times in 2005 on the experimental
quadrats: in early May, in mid-July, and at the end of
August. Each 4 m2 quadrat was divided into two
sections: a 20 cm buffer zone was established at the
periphery of each quadrat to avoid trampling effects
and no data were collected from this section; the
central section of each quadrat (1.60 m 9 1.60 m)
was divided into 256 plots of 100 cm2 each and the
species present in each plot (there could be more than
one species) were recorded.
Species-specific cover values for each quadrat
were estimated as the frequency of 100 cm2 plots,
over 256, in which a given species was present. As
three surveys were done in 2005, the maximum
species-specific cover value was used to estimate the
annual production of each species. Annual production
per functional group was calculated as the sum of the
maximum species-specific cover values for each
238
A.G. Van der Valk (ed.)
Table 1 F- and P-values (in parentheses) from the ANOVAs for the effects of site (df = 2), treatment (*df = 5; otherwise, df = 4)
and their interaction (*df = 10; otherwise, df = 8) on the variables studied
Variables
Richness (all functional groups)*
Effects
Site
Treatment
Site 9 treatment
1.52 (0.181)
189.16 (<0.001)
5.92 (<0.001)
Richness of spring-flowering ephemeral species
4.27 (0.070)
0.28 (0.884)
1.35 (0.264)
Richness of spring-flowering persistent species
62.60 (<0.001)
0.77 (0.556)
3.20 (0.013)
Richness of summer-flowering species
30.05 (<0.001)
2.64 (0.058)
2.26 (0.058)
Richness of fern species
25.55 (0.001)
1.34 (0.286)
0.58 (0.779)
Richness of seedlings and juveniles of woody species
10.55 (0.011)
0.53 (0.713)
0.99 (0.468)
Cover (all functional groups)*
50.08 (<0.001)
2.89 (0.030)
1.20 (0.327)
Cover of spring-flowering ephemeral species
6.15 (0.035)
0.31 (0.864)
0.53 (0.823)
Cover of spring-flowering persistent species
47.53 (<0.001)
0.48 (0.746)
0.40 (0.909)
4.05 (0.071)
2.81 (0.048)
2.42 (0.045)
24.33 (<0.001)
29.17 (<0.001)
1.37 (0.273)
0.82 (0.519)
2.02 (0.087)
0.74 (0.655)
Cover of summer-flowering species
Cover of fern species
Cover of seedlings and juveniles of woody species
Soil pH*
5.92 (0.038)
1.53 (0.210)
1.00 (0.460)
Soil organic matter content*
5.11 (0.051)
1.30 (0.289)
0.44 (0.910)
Significant values are in boldface characters
quadrat. Species richness per functional group was
also determined for each quadrat.
Environmental variables
In July 2005, four soil samples (each of 90.75 cm3, to
a depth of 10 cm) were collected in each quadrat. All
samples were collected after a period of 2–3 days
without rain. In the laboratory, each sample was
passed through a 2-mm-mesh sieve to remove roots,
twigs and stones, and then dried for 24 h at 75C.
Soil pH was measured in a 1:1 soil:water solution.
Soil organic matter (a good surrogate of soil fertility
in deciduous forests) was estimated as percent mass
loss on ignition (10 ml of soil at 450C for 5 h). For
each one of these environmental variables, an average
value was calculated for each quadrat and used for
the statistical analyses.
Data analysis
The experimental design was made up of six
treatments per block, three blocks per site, and three
sites. The sources of variation of interest were sites
(random, df = 2), blocks within sites (random,
df = 6), treatments (fixed, df = 5 or df = 4), and
the interaction between treatments and sites (fixed,
df = 10 or df = 8; see Table 1).
The pre-experimental data set of 2003 (species
richness only) and the experimental data set of 2005
(species richness of each functional group and of all
five groups together, cover of each functional group and
of all five groups together, soil organic matter content
and pH) were analysed using SAS version 6.12 (SAS
Institute, Inc., Cary, NC, USA). Differences between
treatments were identified with LSD (least significant
difference) tests when the ANOVAs revealed significant differences among treatments (P B 0.05).
The species richness and cover of a given removal
treatment may be expected to be lower than that of
the control treatment, simply because plants were
removed. Thus, we calculated an expected value of
richness and cover for each removal treatment by
subtracting the richness or cover of a given functional
group in the control from the overall richness or
cover of the control. A paired t-test (df = 8, since we
had a total of three blocks for each one of the three
sites) was used to determine the significance of the
differences between the observed and the expected
richness and cover values, i.e. to determine if a
removal treatment had a significant positive or
negative effect on the variable of interest.
Forest Ecology
We determined the strength of the relationship
between cover and richness (linear regression analysis, forced through the origin) for the control and for
each one of the five removal treatments. We tested
for differences among slopes, and since the test
indicated that there were significant overall differences (see below), we compared the slopes two by
two (Sokal and Rohlf 1995).
Results
Pre-experimental data (2003)
In 2003, species richness differed significantly among
sites (F = 257.71; P \ 0.0001), but not among
treatments (F = 0.95, P = 0.4635). This latter result
indicates that quadrats were initially similar within
sites and it validates our randomization procedure
within blocks.
Experimental data (2005)
Richness and cover
Species richness for all sites combined was dominated by herbaceous species (56 species,
representing *76 % of the total pool), whereas only
18 woody species were recorded. The mean number
of species per site was 47.0 ± 11.3 (mean ± SE).
The most important difference in species composition
among sites was observed at Sainte-Françoise, which
had 26 summer-flowering species (in comparison to
13 and 16 for Saint-Grégoire 1 and Saint-Grégoire 2,
respectively). In the control quadrats, richness averaged 15.4 ± 1.9 species and cover amounted to
383.9 ± 48.1%.
As expected, the ANOVAs indicated that total
richness differed significantly among treatments
(P \ 0.001; Table 1): the control had the highest
and the seedlings and juveniles of woody species
removal had the lowest richness (Fig. 1). Observed
richness was significantly lower than expected richness for the spring-flowering persistent species
removal and for the seedlings and juveniles of woody
species removal (P = 0.0304 and P = 0.0232,
respectively; Table 2). However, observed richness
was somewhat higher (17.3%) than expected richness
239
for the summer-flowering species removal
(P = 0.0907).
There was a marginally significant effect of our
treatments on summer-flowering species richness
(P = 0.058; Table 1): indeed, summer-flowering
species richness was low in the quadrats in which
spring-flowering persistent species, fern species, or
seedlings and juveniles of woody species had been
removed (Fig. 1). There were no significant differences among treatments in the richness of springflowering ephemeral species, spring-flowering persistent species, fern species, or seedlings and
juveniles of woody species (Table 1, Fig. 1).
As for richness, there were significant differences
among treatments for the variable ‘‘cover’’
(P = 0.030, Table 1): indeed, cover was higher in
the control, but lower in the spring-flowering persistent species removal, the summer-flowering species
removal, the fern species removal and the seedlings
and juveniles of woody species removal (Fig. 1).
There were, however, no significant differences
between the expected and the observed values of
cover for any of the removal treatments, although
observed cover was somewhat lower than expected in
the spring-flowering persistent species removal
(P = 0.1015; Table 2).
The cover of summer-flowering species differed
significantly among treatments (P = 0.048; Table 1):
there was a marked decrease of summer-flowering
species cover when seedlings and juveniles of woody
species were removed (Fig. 1). However, the significant treatment 9 site interaction indicated that the
intensity of the treatment effect on summer-flowering
species cover varied according to site (P = 0.045;
Table 1). No significant differences were detected
among treatments in the cover of the other functional
groups (Table 1, Fig. 1).
All six cover-richness regressions were significant (P \ 0.0001) with R2 from 0.912 to 0.976
(Fig. 2). The slopes varied from 20.962 (fern
species removal), to 28.692 (spring-flowering persistent species removal) and although there were
significant differences among the six slopes
(P \ 0.05), none of the slopes of the removal
treatments differed from that of the control (pairwise comparison). Log-transforming richness provided a poorer fit (lower R2) for all the regressions
(from 0.876 to 0.942).
240
A.G. Van der Valk (ed.)
Table 2 Observed and expecteda richness and cover as a function of removal treatment. Mean ± SE (n = 9)
Treatments
Spring-flowering
ephemeral species
removal
Spring-flowering
persistent species
removal
Summer-flowering
species removal
Fern species
removal
Seedlings and juveniles
of woody species removal
Richness
Observed
12.9 ± 2.3
10.3 ± 1.4a
12.2 ± 1.7
11.8 ± 1.5
9.7 ± 1.5a
Expected
13.6 ± 1.7
12.6 ± 1.4b
10.4 ± 1.0
13.2 ± 1.6
12.0 ± 1.6b
Cover
Observed
329.2 ± 49.6
291.6 ± 47.7
292.9 ± 51.2
250.9 ± 34.2
267.8 ± 42.7
Expected
308.8 ± 49.1
340.3 ± 42.9
246.3 ± 42.4
262.5 ± 41.5
298.6 ± 41.7
Different letters indicate significant differences between the observed and expected values for each treatment (P B 0.05; paired t-test,
df = 8)
a
We calculated an expected value of richness and of cover for each removal treatment by subtracting the richness or cover of a given
functional group in the control from the overall richness or cover of the control (within each block)
Abiotic variables
The ANOVAs for the soil variables (pH and organic
matter content) did not reveal any significant differences among treatments (Table 1). Soil pH values
varied from 4.48 ± 0.07 to 4.68 ± 0.07 (mean ±
SE). Soil organic matter content was relatively
similar among treatments with means from 11.11 ±
0.38% to 14.24 ± 1.92%.
Inter-site differences
There were significant differences among sites for
most of the variables studied: richness; cover;
richness and cover of almost all of the functional
groups; soil pH and organic matter content (Table 1).
Discussion
Richness
Richness decreased markedly when seedlings and
juveniles of woody species or spring-flowering persistent species were removed (Fig. 1, Table 2): this
was mostly through associated decreases in the
richness of summer-flowering species and, to a lesser
extent, of fern species. Spring-flowering persistent
species and seedlings and juveniles of woody species
complete their aboveground growth in the spring,
mostly before tree canopy closure, but persist through
the entire summer even under a thick overstory
(Rothstein and Zak 2001). They often form a dense
understory that protects the soil against erosion and
contributes to reduce nutrient losses (Tessier and
Raynal 2003); they also help maintain a lower soil
and air temperature and a higher soil and air moisture
(Scholes and Archer 1997). By doing so, they may
contribute to create conditions favourable for the
recruitment and/or maintenance of those species
which complete their cycle in the summer (summerflowering species and fern species), at a period when
drought stress may be limiting. These mechanisms
remain speculative, however, and need to be experimentally tested. Nevertheless, our results confirm
those of George and Bazzaz (1999a, b) who demonstrated that interactions between functional groups
could affect understory recruitment and, in the longer
term, richness.
Our initial null hypotheses stipulated that the
different functional groups had little or no interactions with each other, and that the loss of a group
would not have any serious consequences on the
functional properties of the system. Our results on
richness do not support these hypotheses: some of the
functional groups interacted positively with others
and, thus, were not strictly independent. The loss of a
given functional group had significant consequences
on the understory richness, some groups having a
greater proportional effect than others, however.
Cover
Cover decreased significantly following our removals
and to a similar level regardless of which functional
Forest Ecology
20
500
All species
a
b
bc
cd
400
ab
b
bcd
d
10
Cover (%)
Richness
All species
a
15
b
b
b
300
200
5
100
0
0
2.5
80
Spring-flowering ephemeral species
Spring-flowering ephemeral species
60
Cover (%)
Richness
2.0
1.5
1.0
40
20
0.5
0
0
5
70
Spring-flowering persistent species
Spring-flowering persistent species
60
4
Cover (%)
Richness
50
3
2
40
30
20
1
10
0
0
7
200
Summer-flowering species
a
Summer-flowering species
6
150
Cover (%)
Richness
5
4
3
2
a
ab
ab
100
b
50
1
0
0
4
200
Fern species
2
1
100
50
0
0
6
100
Seedlings and juveniles of woody species
5
Seedlings and juveniles of woody species
80
Cover (%)
4
3
2
60
40
20
1
0
Fern species
150
Cover (%)
Richness
3
Richness
Fig. 1 Richness (all
functional groups together),
individual functional group
richness (left), cover (all
functional groups together)
and individual functional
group cover (right) as a
function of treatment. For
each variable, different
letters indicate significant
differences among
treatments (P B 0.05,
protected LSD).
Mean ? SE. C, control;
SFESR, spring-flowering
ephemeral species removal;
SFPSR, spring-flowering
persistent species removal;
SFSR, summer-flowering
species removal; FSR, fern
species removal; SJWSR,
seedlings and juveniles of
woody species removal. See
Table 1 for details on the
ANOVA results
241
C
SFESR SFPSR SFSR
Tre a tme nts
FSR SJWSR
0
C
SFESR SFPSR SFSR
Tre a tme nts
FSR SJWSR
242
A.G. Van der Valk (ed.)
700
b Fig. 2 Cover-richness relationship for the control and the
different removal treatments. Regression lines were forced
through the origin: C, control: slope = 23.473, P \ 0.0001,
R2 = 0.963; SFESR, spring-flowering ephemeral species
removal: slope = 24.573, P \ 0.0001, R2 = 0.976; SFPSR,
spring-flowering persistent species removal: slope = 28.692,
P \ 0.0001, R2 = 0.970; SFSR, summer-flowering species
removal: slope = 23.688, P \ 0.0001, R2 = 0.912; FSR, fern
species removal: slope = 20.962, P \ 0.0001, R2 = 0.959;
SJWSR, seedlings and juveniles of woody species removal:
slope = 27.329, P \ 0.0001, R2 = 0.969
C
Co ver (%)
600
500
400
300
200
100
0
700
SFESR
Co ver (%)
600
500
400
300
200
100
0
700
SFPSR
Co ver (%)
600
500
400
300
200
100
0
700
SFSR
Co ver (%)
600
500
400
300
200
100
0
700
FSR
Co ver (%)
600
500
400
300
200
100
0
700
SJWSR
Co ver (%)
600
500
400
300
200
100
0
0
5
10
15
Richness
20
25
30
group was removed (except for the removal of
spring-flowering ephemeral species, which did not
differ significantly from the control; Fig. 1): this
result would seem to suggest no short-term compensation in species production following the removal of
a given functional group. Yet, differences in cover
among functional groups were quite large in the
control: [(summer-flowering species * fern species) [ (spring-flowering ephemeral species *
seedlings and juveniles of woody species) [ springflowering persistent species]. In fact, cover decreased
somewhat more than expected (14.3% more) when
spring-flowering persistent species were removed
(Table 2): this was mostly through a slight decrease
in the cover of summer-flowering species and of fern
species relative to the control. In contrast, cover was
somewhat higher than expected (18.9% higher) when
summer-flowering species were removed and this was
mostly through a slight increase in the cover of fern
species relative to the control.
The removal of seedlings and juveniles of woody
species led to a significant decrease in the cover of
summer-flowering species. As mentioned above,
under the cover of seedlings and juveniles of woody
species, soil and air temperature is reduced and soil
and air moisture is higher, conditions that may favour
the growth of summer-flowering species during a
period when drought stress may otherwise limit
growth (Schulz and Adams 1995; Scholes and Archer
1997). Herbivores may also respond to the spatial
structure of the understory cover, some plant species
benefiting from, and others being hindered by, the
presence of some plant species (George and Bazzaz
1999b).
Our results for cover support those presented
above with respect to the effect of the removal of a
functional group on richness: significant interactions
were present among some of the groups.
Forest Ecology
Cover-richness relationship
The linearity of the relationship between cover and
richness, as described for the control, would seem to
suggest strong complementarity among functional
groups: indeed, according to Dı́az and Cabido (2001,
p. 651) ‘‘only when all species have equally complementary niches … the rate of ecosystem processes
should be expected to increase linearly with species
richness’’. However, assuming complementarity
among groups, we could have expected the slope of
the cover-richness regression to decrease significantly
following our removals (Loreau et al. 2001); this was
not the case (no significant differences between the
slope of the control treatment and that of each one of
our five removal treatments; pair-wise comparisons).
Instead, these results suggest that there are no
complementary effects among functional groups and
that intergroup independence may be responsible for
the linear cover-richness relationship reported for the
control (see also Hooper and Dukes 2004).
Our results on cover-richness relationship also
show that removing spring-flowering persistent species or fern species, two groups with comparable
richness although very dissimilar cover, affects
differently the cover-richness relationship (slopes
significantly different between spring-flowering persistent species removal and fern species removal at
P \ 0.01): this suggests that the loss of a given
number of species may have contrasting effects on
ecosystem functioning depending on the characteristics of the species that are lost (e.g. their
productivity), a conclusion already reached by Dı́az
and Cabido (2001). It also emphasizes the significance of functional composition and functional
richness for ecosystem processes (Wardle et al.
2000; Loreau et al. 2001; Hooper and Dukes 2004).
Duration of the experiment
Overall, few significant effects were observed as a
result of our removal treatments and a longer-term
experiment may have revealed more complex intergroup interactions (some effects were marginally
significant). Yet, the effects reported here after only
two years of experiment indicate that even the
understory can respond quite rapidly to removals, as
has been shown by Stinson et al. (2007) only one
243
year after having eradicated the invasive Alliaria
petiolata from plots in the understory of a hardwood
forest of New England. Not unexpectedly, considering the short duration of our experiment, none of the
removal treatments had significant effects on the soil
variables.
Conclusions
Both positive and negative interactions were detected
between functional groups, although positive intergroup interactions prevailed. Spring-flowering
persistent species and seedlings and juveniles of
woody species appeared to be key groups, because of
their positive effects on summer-flowering species
and fern species, two groups having a particularly
significant contribution to understory cover and
richness. Such interactions may increase the stability
of the forest understory and help maintain ecosystem
functions (MacArthur 1955; Tilman and Downing
1994; Johnson et al. 1996): indeed, they may (1) lead
to a higher collective capacity of the species to resist
to minor disturbances and/or (2) allow the system to
re-establish essential functions following more
important disturbances (Walker et al. 1999; Upadhyay et al. 2000; Dı́az and Cabido 2001).
Each one of the five functional groups contributed directly to richness and cover (Tilman et al.
1997), most likely because of the complementary
morphological, physiological and/or phenological
traits of their respective species (Loreau and Behera
1999; Reich et al. 2003). Yet, our results on the
cover-richness relationship suggested a relative
independence among functional groups. Further
experimental work is needed to identify the specific
mechanisms responsible for the interactions outlined
here between the different functional groups of the
understory and to confirm that short-term responses
can be translated into still more significant responses
with time.
Acknowledgements The authors thank V. Bolduc-Tremblay,
G. de Lafontaine, P. Désilets, G. Descôteaux, P. Marchand, and
F. Sahim for field assistance, and S. Boudreau, L. Lapointe,
M.F. McKenna, and S. Payette for comments on an earlier
version of the manuscript. This study was financed by the
Natural Sciences and Engineering Research Council of Canada
through a grant to G. Houle.
244
References
Balvanera P, Pfisterer AB, Buchmann N, He J-S, Nakashizuka
T, Raffaelli D, Schmid B (2006) Quantifying the evidence
for biodiversity effects on ecosystem functioning and
services. Ecol Lett 9:1146–1156. doi:10.1111/j.14610248.2006.00963.x
Bratton SP (1976) Resource division in an understory herb
community: responses to temporal and microtopographic
gradients. Am Nat 110:679–693. doi:10.1086/283097
Buonopane M, Huenneke LF, Remmenga M (2005) Community response to removals of plant functional groups and
species from a Chihuahuan Desert shrubland. Oikos
110:67–80. doi:10.1111/j.0030-1299.2005.13949.x
Cardinale BJ, Nelson K, Palmer M-A (2000) Linking species
diversity to the functioning of ecosystems: on the
importance of environmental context. Oikos 91:175–183.
doi:10.1034/j.1600-0706.2000.910117.x
Choinière L, Laplante L (1948) Étude des sols du Comté de
Nicolet. Ministère de l’Agriculture et Ministère de l’industrie et du Commerce, Québec
Dı́az S, Cabido M (2001) Vive la différence: plant functional
diversity matters to ecosystem processes. Trends Ecol
Evol 16:646–655. doi:10.1016/S0169-5347(01)02283-2
Eickmeier WG, Schussler EE (1993) Responses of the spring
ephemeral Claytonia virginica L. to light and nutrient
manipulations and implications for the vernal-dam
hypothesis. Bull Torrey Bot Club 120:157–168. doi:
10.2307/2996945
Fitter AH (1982) Influence of soil heterogeneity on the coexistence of grassland species. J Ecol 70:139–148. doi:
10.2307/2259869
Fox JW (2005) Interpreting the selection effect of biodiversity
on ecosystem function. Ecol Lett 8:546–856. doi:
10.1111/j.1461-0248.2005.00795.x
George LO, Bazzaz FA (1999a) The fern understory as an
ecological filter: emergence and establishment of canopytree seedlings. Ecology 80:833–845
George LO, Bazzaz FA (1999b) The fern understory as an
ecological filter: growth and filter of canopy-tree seedlings. Ecology 80:846–856
Grime JP (1973) Competitive exclusion in herbaceous vegetation. Nature 242:344–347. doi:10.1038/242344a0
Grime JP (1977) Evidence for the existence of three primary
strategies in plants and its relevance to ecological and
evolutionary theory. Am Nat 111:1169–1194. doi:
10.1086/283244
Hector A, Schmid B, Beierkuhnlein C, Caldeira MC, Diemer M,
Dimitrakopoulos PG, Finn JA, Freitas H, Giller PS, Good J,
Harris R, Högberg P, Huss-Danell K, Joshi J, Jumpponen A,
Körner C, Leadley PW, Loreau M, Minns A, Mulder CPH,
O’Donovan G, Otway SJ, Pereira JS, Prinz A, Read DJ,
Scherer-Lorenzen M, Schulze E-D, Siamantziouras ASD,
Spehn EM, Terry AC, Troumbis AY, Woodward FI, Yachi
S, Lawton JH (1999) Plant diversity and productivity
experiments in European grasslands. Science 286:1123–
1127. doi:10.1126/science.286.5442.1123
Hooper DU, Dukes JS (2004) Overyielding among plant
functional groups in a long-term experiment. Ecol Lett
7:95–105. doi:10.1046/j.1461-0248.2003.00555.x
A.G. Van der Valk (ed.)
Hooper DU, Solan M, Symstad A, Dı́az S (2002) Species
diversity, functional diversity, and ecosystem functioning.
In: Loreau M, Naeem S, Inchausti P (eds) Biodiversity
and ecosystem functioning: synthesis and perspectives.
Oxford University Press, Oxford, pp 195–281
Huston MA (1997) Hidden treatments in ecological experiments: re-evaluating the ecosystem function of
biodiversity. Oecologia 110:449–460. doi:10.1007/s0044
20050180
Johnson KH, Vogt KA, Clark HJ, Schmitz OJ, Vogt DJ (1996)
Biodiversity and the productivity and stability of ecosystems. Trends Ecol Evol 11:372–377. doi:10.1016/01695347(96)10040-9
Lepš J (2004) What do the biodiversity experiments tell us
about consequences of plant species loss in the real
world? Basic Appl Ecol 5:529–534. doi:10.1016/j.baae.
2004.06.003
Loreau M, Behera N (1999) Phenotypic diversity and stability
of ecosystem processes. Theor Popul Biol 56:29–47. doi:
10.1006/tpbi.1998.1408
Loreau M, Naeem S, Inchausti P, Bengtsson J, Grime JP, Hector
A, Hooper DU, Huston MA, Raffaelli D, Schmid B, Tilman
D, Wardle DA (2001) Biodiversity and ecosystem functioning: current knowledge and future challenges. Science
294:804–808. doi:10.1126/science.1064088
MacArthur R (1955) Fluctuations of animal populations and a
measure of community stability. Ecology 36:533–536.
doi:10.2307/1929601
Marie-Victorin (2002) Flore Laurentienne, 3rd edn. Gaétan
Morin Éditeur, Boucherville, Québec
Mittelbach GG, Steiner CF, Scheiner SM, Gross KL, Reynolds
HL, Waide RB, Willig MR, Dodson SI, Gough L (2001)
What is the observed relationship between species richness and productivity? Ecology 82:2381–2396
Muller RN, Bormann FH (1976) Role of Erythronium americanum Ker. In energy flow and nutrient dynamics of a
northern hardwood forest ecosystem. Science 193:1126–
1128. doi:10.1126/science.193.4258.1126
Naeem S (2001) How changes in biodiversity may affect the
provision of ecosystem services. In: Chichilnisky G, Daily
GC, Ehrlich P, Heal G, Miller J (eds) Managing Humandominated Ecosystems. Missouri Botanical Garden, St
Louis, MO, pp 3–33
Naeem S, Chapin FSIII, Costanza R, Ehrlich PR, Golley FB,
Hooper DU, Lawton JH, O’Neill RV, Mooney HA, Sala
OE, Symstad AJ, Tilman D (1999) Biodiversity and
ecosystem functioning: maintaining natural life support
processes. Ecology 4:1–12
Naeem S, Thompson LJ, Lawler SP, Lawton JH, Woodfin RM
(1995) Empirical evidence that declining species diversity
may alter the performance of terrestrial ecosystems.
Philos Trans R Soc London [Biol] 347:249–262
Reich PB, Buschena C, Tjoelker MG, Wrage K, Knops J,
Tilman D, Machado JL (2003) Variation in growth rate
and ecophysiology among 34 grassland and savanna
species under contrasting N supply: a test of functional
group differences. New Phytol 157:617–631. doi:
10.1046/j.1469-8137.2003.00703.x
Rothstein DE, Zak DR (2001) Photosynthetic adaptation and
acclimation to exploit seasonal periods of direct irradiance
Forest Ecology
in three temperate, deciduous-forest herbs. Funct Ecol
15:722–732. doi:10.1046/j.0269-8463.2001.00584.x
Rowe JS (1972) Les régions forestières du Canada. Ministère
de l’Environnement, publication N 1300F, Canada
Scholes RJ, Archer SR (1997) Tree-grass interactions in savannas. Annu Rev Ecol Syst 28:517–544. doi:10.1146/
annurev.ecolsys.28.1.517
Schulz KE, Adams MS (1995) Effect of canopy gap light
environment on evaporative load and stomatal conductance in the temperate forest understory herb Aster
macrophyllus (Asteraceae). Am J Bot 82:630–637. doi:
10.2307/2445421
Sokal RR, Rohlf FJ (1995) Biometry, 3rd edn. Freeman, New
York
Stinson K, Kaufman S, Durbin L, Lowenstein F (2007) Impacts
of garlic mustard invasion on a forest understory community. Northeastern Nat 14:73–88. doi:10.1656/10926194(2007)14[73:IOGMIO]2.0.CO;2
Symstad AJ, Chapin FSIII, Wall DH, Gross KL, Huenneke LF,
Mittelbach GG, Peters DPC, Tilman D (2003) Long-term
and large-scale perspectives on the relationship between
biodiversity and ecosystem functioning. Bioscience 53:89–
98.
doi:10.1641/0006-3568(2003)053[0089:LTALSP]
2.0.CO;2
Symstad AJ, Tilman D, Willson J, Knops J-MH (1998) Species
loss and ecosystem functioning: effects of species identity
and community composition. Oikos 81:389–397. doi:
10.2307/3547058
Tessier JT, Raynal DJ (2003) Vernal nitrogen and phosphorus
retention by forest understory vegetation and soil
microbes. Plant Soil 256:443–453. doi:10.1023/A:1026
163313038
Tilman D (1997) Biodiversity and ecosystem functioning. In:
Daily GC (ed) Nature’s Services. Island Press, Washington, DC, pp 93–112
245
Tilman D, Downing JA (1994) Biodiversity and stability in
grasslands. Nature 367:363–365. doi:10.1038/367363a0
Tilman D, Knops J, Wedin D, Reich P (2002) Plant diversity
and composition: effects on productivity and nutrient
dynamics of experimental grasslands. In: Loreau M,
Naeem S, Inchausti P (eds) Biodiversity and Ecosystem
Functioning: Synthesis and Perspectives. Oxford University Press, Oxford, pp 21–35
Tilman D, Knops J, Wedin D, Reich P, Ritchie M, Siemann E
(1997) The influence of functional diversity and composition on ecosystem processes. Science 277:1300–1302.
doi:10.1126/science.277.5330.1300
Troumbis AY, Memtsas D (2000) Observational evidence that
diversity may increase productivity in Mediterranean
shrublands. Oecologia 125:101–108. doi:10.1007/
PL00008880
Upadhyay RK, Rai V, Iyengar SRK (2000) How do ecosystems
respond to external perturbations? Chaos Solitons Fractals
11:1963–1982. doi:10.1016/S0960-0779(99)00086-7
Walker B, Kinzig A, Langridge J (1999) Plant attribute
diversity, resilience, and ecosystem function: the nature
and significance of dominant and minor species. Ecosystems (NY, Print) 2:95–113. doi:10.1007/s100219900062
Wardle DA, Bonner KI, Barker GM (2000) Stability of ecosystem properties in response to above-ground functional
group richness and composition. Oikos 89:11–23. doi:
10.1034/j.1600-0706.2000.890102.x
Wardle DA, Bonner KI, Barker GM, Yeates GW, Nicholson
KS, Bardgett RD, Watson RN, Ghani A (1999) Plant
removals in perennial grassland: vegetation dynamics,
decomposers, soil biodiversity, and ecosystem properties.
Ecol Monogr 69:535–568
Wardle DA, Zackrisson O (2005) Effects of species and
functional group loss on island ecosystem properties.
Nature 435:806–810. doi:10.1038/nature03611
Host trait preferences and distribution of vascular epiphytes
in a warm-temperate forest
Akiko Hirata Æ Takashi Kamijo Æ Satoshi Saito
Originally published in the journal Plant Ecology, Volume 201, No. 1, 247–254.
DOI: 10.1007/s11258-008-9519-6 Springer Science+Business Media B.V. 2008
Abstract To illustrate the ecological factors and
process leading to the observed diversity patterns of
vascular epiphytes, we examined the effect and
importance of host tree traits on epiphyte richness
and spatial aggregation of epiphytes. The study was
conducted in warm-temperate forest in Japan. The
recorded host traits were diameter, height, species,
habitat topography, and growth rate, and we analyzed
the effects and importance of these traits on three
species groups: total epiphytic species, epiphytic
orchid species, and epiphytic pteridophyte species.
Diameter and species of host trees had the greatest
influence on epiphytes and their magnitudes were
roughly similar in all species groups. Growth rate and
topography were less important than host size and
species. Growth rate had a negative effect on all three
groups, and topography was important for pteridophytes. Epiphyte richness did not exhibit clear spatial
aggregation. Our results suggest that size, stability,
and quality of the host are equally important in
determining epiphyte colonization.
A. Hirata (&) T. Kamijo
Graduate School of Life and Environmental Sciences,
University of Tsukuba, 1-1-1 Tennodai, Tsukuba,
Ibaraki 305-8572, Japan
e-mail: akiko_hirata1845@yahoo.co.jp
S. Saito
Forestry and Forest Products Research Institute,
1 Matsunosato, Tsukuba, Ibaraki 305-8687, Japan
Keywords Diversity Evergreen broad-leaved
forest Orchid Pteridophyte Host specificity
Spatial pattern
Introduction
Vascular epiphytes are an essential component of the
vegetation of forests, in terms of both species diversity
and their role in forest ecosystem functions (Gentry
and Dodson 1987; Benzing 1990; Burns and Dawson
2005; Flores-Palacios and Garcı́a-Franco 2006). Their
importance for forest diversity is based on their
proportionally large contribution to the diversity of
the local flora (Dawson and Sneddon 1969; Gentry and
Dodson 1987; Benzing 1990; Burns and Dawson
2005). However, we still know little about the
mechanisms that maintain epiphyte diversity.
In forests, the distribution pattern of vascular
epiphyte diversity is affected by two major processes:
dispersal and establishment. Host trees provide the
substrate for epiphytes, so establishment seems to be
affected by host tree traits, including area available for
establishment, physical and chemical characteristics
of bark, and architecture, e.g., canopy structure (Frei
and Dodson 1972; Migenis and Ackerman 1993). Size
and species of host trees are indicative traits of these
factors, because an increase in size leads to an increase
in habitat area and in the chance of epiphyte
establishing. Host species determines traits, such as
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_19
247
248
bark characteristics and canopy structure. Many
previous studies have examined the influence of host
size and species separately when examining the
diversity and presence of epiphyte. Most studies have
found a positive relation between host size and
epiphyte diversity (Hietz and Hietz-Seifert 1995; Zotz
and Vollrath 2003; Burns and Dawson 2005). In terms
of the relationship between host species and epiphyte
presence, there is little support for the notion of strict
host-specificity in epiphytes (Benzing 1990). However, failure to find a one-to-one match between
particular species of host trees and epiphytes dose not
confirm ‘neutrality’ of host tree species identity with
respect to the structuring of epiphyte communities
(Laube and Zotz 2006). Some studies have shown that
epiphyte diversity increases markedly with host tree
size, and that the trend differs among host species
(Callaway et al. 2002; Burns and Dawson 2005).
Therefore, a combination of the host tree traits appears
to be important in determining the epiphytes present
and diversity, rather than simply host size or host
species (Laube and Zotz 2006).
In addition, if the vascular epiphyte distribution
shows an aggregative pattern, this suggests the
possibility that another factor, such as dispersal
limitation, affects epiphyte distribution in addition
to habitat preference. In general, epiphytic plants are
expected to be characterized by high seed numbers
and long-distance dispersal, mainly wind-borne
(Gentry and Dodson 1987; Nieder et al. 2000). As a
result, a number of studies that have found a patchy,
rather than random, distribution of epiphytes have
concluded that aggregation was the result of the
presence of preferred hosts with respect to size and
substrate characteristics (Tremblay 1997; Nieder
et al. 2000). However, some studies have shown that
orchid seed dispersal, the most diverse group of
epiphytes, occurs over relatively short distances
(Tremblay 1997; Machon et al. 2003).
The ecological factors affecting epiphyte diversity
may interact with one another. To understand these
factors, it is necessary to examine the effects of host
traits on vascular epiphyte diversity by analyzing all
traits simultaneously. On the other hand, to understand
the process causing the patterns of vascular epiphyte
diversity, it is necessary to examine both the distribution pattern and habitat preference. For non-vascular
epiphytes, i.e., bryophytes and lichens, Löbel et al.
2006 it was found that both environmental conditions
A.G. Van der Valk (ed.)
and spatial aggregation can explain a substantial part
of the variation in species richness. The relevant
factors, however, have been studied rarely in a single
system for vascular epiphytes.
The aims of this study were to assess the relative
importance of host traits, such as size and quality, by
analyzing them simultaneously, and to consider
which habitat preference and dispersal limitation
mechanisms affect the distribution of vascular epiphyte diversity. The host tree traits selected for
examination were habitat topography and growth rate
of the host tree, and host size and species, because we
suspected that these host traits reflect the environment around host trees and the surface stability of the
substrate, which may influence epiphyte colonization.
The questions addressed in this article are:
(1)
(2)
(3)
Which host tree traits affect epiphyte diversity?
How are these traits important?
Is epiphyte diversity patchy in its distribution?
Based on the answers to these questions, we
discuss the mechanisms controlling the pattern of
epiphyte diversity.
Materials and methods
Study area
The study area is in Kyushu Chuo Sanchi SemiNational Park, Southwestern Japan. The natural
vegetation of this area is evergreen broad-leaved
forest. An old-growth forest covering more than
300 ha is well preserved around the study area
(Tanouchi and Yamamoto 1995). Most natural evergreen broad-leaved forests have been logged or
greatly changed by human activity in Japan. The
forest has a canopy that is more than 30 m tall. This
is dominated by evergreen broad-leaved species, such
as Distylium racemosum Sieb. et Zucc. and Quercus
acuta Thunb. (Tanouchi and Yamamoto 1995; Saito
2002). A permanent 4-ha plot (200 m 9 200 m) was
set up on a north to northwest-facing slope on Mt.
Omori (1,109 m asl; 32040 N, 131090 E) at an
elevation ranging from 380 to 520 m in 1989 (Sato
et al. 1999). This plot has been used for long-term
ecological research since then; data of species name,
diameter at breast height (DBH; 1.3 m), and location
of all trees with a DBH greater than 5 cm, plus a
Forest Ecology
249
microtopographical classification of the plot have
been accumulated (Sato et al. 1999). During the
period 1994 to 1998, the mean annual temperature in
this plot at 495 m asl was 14.2C. The mean annual
precipitation from 1951 to 1997 was 3,070 mm at the
nearest recording station, the Ayakita Prefectural
Observatory, 294 m asl (Miyazaki Local Meteorological Observatory 1951–1997; Sato et al. 1999).
Within the 4 ha permanent plot, we set up an
80 m 9 120 m study plot for the epiphyte survey;
this area contained most of the microtopographic
types present in the surroundings (Fig. 1).
Host tree traits and epiphyte species richness
Within the study plot (80 m 9 120 m), DBH, height,
and species name of all the trees with a DBH greater
than 20 cm (i.e., potential host trees) were recorded
in 2007. The microtopography of the habitat in which
the host trees were located was defined on the basis of
data collected previously (Sato et al. 1999), namely:
crestslope, upper sideslope, lower sideslope, headmost wall, head hollow, footslope, and bottomland
(Tamura and Takeuchi 1980; Tamura 1987; Ohnuki
et al. 1997; Fig. 1). The growth rate of each host tree
was estimated by the ratio of 2007 DBH to the 2005.
500m
N
450m
120m
400m
80m
Crestslope
Upper sideslope
Lower sideslope
Headmost wall
Head hollow
Footslope
Bottomland
Fig. 1 Microtopography of the permanent 4-ha plot. An
80 m 9 120 m quadrat shows our study plot
Within the study plot, we recorded the species name
and growing site of all adult epiphytes on a total of 283
host trees. Growing sites were classified into five types
following the revised zonation scheme presented by
Johansson (1974): the basal part of the trunk (0–2 m),
the trunk from 2 m up to the first ramification, the
basal part of the canopy (1/3 of the total length of the
branch), the middle part of the canopy (1/3 of the total
length of the branch), and the outer part of the canopy
(1/3 of the total length of the branch). We searched
epiphytes from the ground using binoculars. The rope
climbing technique was used when searching from the
ground was not possible. With the exception of
mistletoe and an accidental epiphyte (Benzing 1990),
all species were used for the statistical analyses. In
addition to the total epiphyte species, we examined
two species groups: orchids and pteridophytes.
Statistical analysis
The effects of host traits on epiphyte species richness
for the three species groups (total species, orchids, and
pteridophytes) were analyzed using generalized linear
models (GLM; McCullagh and Nelder 1989) with the
log link function, assuming a negative-binomial error
structure. Five host traits were used as predictor
variables: species (Species), DBH (DBH), height
(Height), growth rate (Growth), and habitat microtopography (Topography). The correlation between
predictor variables was tested using the Pearson
correlation coefficient or Cramer coefficient of association. Since all correlation coefficients between each
of the predictor variables were lower than 0.3, all
variables were used as predictors. To evaluate the
importance of each predictor, Akaike information
criterion (AIC) for all possible subsets of the predictor
variables were calculated. The AIC minimizing model
was considered to be the most suitable model, and
predictor variables included in this model were
considered to represent the factors that affected
species richness. To evaluate the relative importance
of each predictor variable, AIC values for the top five
models and the variables included in these models,
were compared for each functional group.
The spatial aggregation of epiphyte species richness was estimated using the semi-variance function.
Semi-variance is a concept widely used to detect
spatial and/or temporal patterns within datasets, e.g.,
in geostatistics (Bailey and Gatrell 1995), and
250
A.G. Van der Valk (ed.)
describes how species richness covaries spatially
among host trees (Löbel et al. 2006). It is defined by:
^cðhÞ ¼
1
2nðhÞ
n
X
zi
2
zj ;
i¼1
where zi and zj represent the observed species richness
on host trees i and j, respectively, h is a vector of
distances between trees, and n(h) is the number of pairs
of host trees located at distance h from each other. The
semi-variance is a measure of dissimilarity for species
richness, and a plot of semi-variance against distance
(i.e., a semi-variogram) can be used to explore whether
the dissimilarity in species richness changes with
distance between trees (Löbel et al. 2006). At short
distance, the values of semi-variance are also small
indicating that the spatial structure is at its strong
intensity (Fortin and Dale 2005). As the distance
increase, the semi-variance values also increase before
leveling off to a plateau (Fortin and Dale 2005). We
plotted the semi-variance values calculated for species
richness against distance, and estimated whether
dissimilarity of species richness changes with distance. In addition to species richness, semi-variance
for the residual from the best model was calculated to
examine the degree of spatial aggregation which could
not be assigned to environmental variables.
All analysis was performed in the R environment
for statistical computing (R Development Core Team
2006).
Results
Excluding species that are accidental epiphytes, 8
orchids, 13 pteridophytes, and 2 mistletoes were
recorded (Table 1). Model selection using AIC
resulted in the variables DBH, Species, and Growth
being included in the best model for all species groups
(Table 2). However, Height was not included in the
best models for any of the three groups. DBH had a
positive effect on epiphyte richness, and growth rate
had a negative effect (Table 2). With respect to host
species, the regression coefficient for Persea japonica
(Sieb. et Zucc.) Kosterm. was the highest, and that for
deciduous trees was the second highest when all
epiphyte species were considered. For the orchids,
Camellia japonica Linn. had the highest regression
coefficient, and deciduous trees the second highest. For
Table 1 Epiphyte species at the study site
Epiphyte species
Family
Number
Neofinetia falcata
Orchidaceae
45
Bulbophyllum drymoglossum
Orchidaceae
43
Bulbophyllum inconspicuum
Orchidaceae
41
Sedirea japonica
Orchidaceae
28
Bulbophyllum japonicum
Orchidaceae
14
Eria reptans
Orchidaceae
13
Saccolabium japonicum
Orchidaceae
4
Dendrobium moniliforme
Orchidaceae
Lepisorus thunbergianus
Polypodiaceae
115
Lemmaphyllum microphyllum
Davallia mariesii
Polypodiaceae
Davalliaceae
110
32
Asplenium wilfordii
Aspleniaceae
24
Pyrrosia lingua
Polypodiaceae
23
Selaginella involvens
Selaginellaceae
20
Gonocormus minutus
Hymenophyllaceae
21
Loxogramme salicifolia
Polypodiaceae
17
Lepisorus onoei
Polypodiaceae
15
Lacosteopsis auriculata
Hymenophyllaceae
Lycopodium sieboldii
Lycopodiaceae
3
Vittaria flexuosa
Vittariaceae
3
Crepidomanes insigne
Hymenophyllaceae
Taxillus yadoriki
Loranthaceae
11
Korthalsella japonica
Loranthaceae
8
3
9
2
the pteridophytes, P. japonica had the highest regression coefficient, and Persea thunbergii (Sieb. et Zucc.)
Kosterm. the second highest. On the other hand, the
regression coefficients for D. racemosum were the
lowest or second lowest for all groups (Table 2).
Topography was included in the pteridophyte model,
and the regression coefficient associated with crestslope
was the lowest of all the topographic categories
(Table 2).
In Table 3, the top five models are ranked
according to their AIC differences (delta AIC), from
best to worst. DBH and Species were included in all
the top five models for all groups. For the orchids and
pteridophytes, AIC difference for model 3, which
includes only the DBH and Species variables, was 2
units less, and these models had substantial support.
In general Growth and Topography were not included
in the top five models (Table 3).
The orchids were found most frequently at growing sites 3 and 4. The pteridophytes showed high
frequency in growing site 1 to growing site 4 (Fig. 2).
Forest Ecology
251
Table 2 The regression coefficients of the generalized linear model (GLM), selected by AIC
Variable
Total
b
Orchid
SE
b
Pteridophyte
SE
SE
b
Intercept
0.113
0.229
-1.177**
0.378
0.451
0.390
DBH
0.635**
0.065
0.684**
0.103
0.453**
0.063
Species
Quercus acuta
0.000
Quercus gilva
0.567
0.347
0.751
0.554
0.334
0.347
Quercus salicina
0.238
0.277
0.627
0.440
-0.197
0.282
Castanopsis cuspidata var. sieboldii
0.015
0.339
-0.198
0.586
-0.233
0.331
Persea thunbergii
0.839**
0.233
Persea japonica
Distylium racemosum
0.000
0.975**
1.170**
0.270
-0.932**
0.290
-1.136*
1.224**
0.000
0.371
0.595**
0.220
0.444
0.605*
0.282
0.516
-1.094**
0.294
0.508
0.397
-0.146
0.856
0.016
0.421
-0.354
0.479
0.393
0.685
-1.373*
0.637
Camellia japonica
0.719
0.432
1.328*
0.642
-0.111
0.500
Deciduous trees
1.164**
0.345
1.311*
0.545
0.561
0.341
0.325
0.075
0.456
-0.271*
0.561
0.135
0.214
-0.102
0.330
0.075
Actinodaphne longifolia
Cleyera japonica
Other evergreen trees
Growth
0.537
-0.157*
Topography
Crestslope
–
–
–
–
-0.764*
0.341
Upper sideslope
–
–
–
–
-0.314
0.308
Lower sideslope
–
–
–
–
-0.309
0.384
Headmost wall
–
–
–
–
-0.040
0.322
Head hollow
–
–
–
–
0.015
0.314
Footslope
–
–
–
–
-0.311
0.391
Bottomland
–
–
–
–
0.000
–
–
–
–
Height
Residual deviance
271.25 (d.f. = 269)
204.45 (d.f. = 269)
–
–
256.74 (d.f. = 263)
Bold text indicates a variable included in the best model for one or more species groups. b = coefficient, SE = Standard error,
d.f. = degrees of freedom, * P \ 0.05, ** P \ 0.01
The semi-variance calculated using species richness and the residual from the best model exhibited
no clear tendency to be small over short distance or to
increase with distance between trees (Fig. 3). The
plot of residuals from the best model exhibited a
more consistent pattern than that of species richness.
Discussion
Factors that affect patterns of vascular epiphyte
species richness
Size and species of host tree were the most important
factors influencing vascular epiphyte richness (Tables 2,
3). The importance of these traits has been demonstrated
individually (Zimmerman and Olmsted 1992; Hietz and
Hietz-Seifert 1995; Callaway et al. 2002; Zotz and
Vollrath 2003; Burns and Dawson 2005), but a quantitative comparison of these traits has not been
previously conducted. Our results show, by analyzing
host traits simultaneously, that they are equally influential. A quantitative comparison of host traits has been
conducted for epiphytic bryophytes and lichens (Löbel
et al. 2006), and it has been shown that host size and
species are the crucial factors affecting richness of nonvascular epiphytes. However, size was only important
for bryophytes, and species was only important for
lichens. This result differed from present study, which
found that these factors exert an equal influence.
252
A.G. Van der Valk (ed.)
relatively rapid changes in the characteristics of the
bark surface. Since vascular epiphytes attach to the
bark surface of their host tree, they may be susceptible to changes to the surface. The influence of
topography was important for pteridophytes
(Table 2). The effect of topography may reflect a
difference in drought survival strategy between
pteridophytes and orchids. The pteridophyte richness
was lowest in crestslope locations that were relatively
dry at our study site. In general, it is known that
vascular epiphytes use the CAM-pathway as a means
of surviving in drought-prone habitats (Winter et al.
1983). However, previous studies have shown that
few pteridophytes are CAM species, unlike the
orchids, and the proportion of these species that are
tolerant to drought increases from shaded trunk to
exposed canopy habitats (Winter et al. 1983; Zotz
and Ziegler 1997). Since orchids were mainly found
in the canopy area in our study (Fig. 2), it is likely
that many of them are highly drought-tolerant.
However, pteridophytes are unlikely to be droughttolerant, possibly explaining the importance of
topography in determining their distribution.
Vascular epiphyte richness did not exhibit clear
spatial aggregation (Fig. 3). This suggests that dispersal ability does not have a crucial effect on
epiphyte richness patterns at the scale of our study.
However, different trends may be exhibited at a
larger scale, since diaspores of orchids and
Table 3 The variables included in the top five models,
according to AIC, and the AIC value for each model
Number
Variables
AIC
Delta
AIC
(a) Total
1
D
S
G
2
D
S
G
3
D
S
G
4
D
5
D
H
T
S
H
T
888.14
0.00
889.67
1.53
890.13
1.99
890.78
2.64
891.32
3.17
S
G
S
G
532.30
0.00
S
G
532.54
0.23
S
534.11
1.80
S
S
G
T
534.25
538.44
1.95
6.14
G
T
746.50
0.00
T
746.52
0.01
746.85
0.35
747.39
0.89
747.75
1.25
(b) Orchid
1
D
2
D
3
D
4
5
D
D
H
H
H
(c) Pteridophyte
1
D
S
2
D
S
3
D
S
4
D
S
5
D
H
G
S
D: DBH; H: height; S: species; G: growth rate; T: topography
On the other hand, growth rate and topography
were less important than host size and species
(Table 3). Growth rate had a negative effect on all
groups. High growth rate of the host tree may cause
Fig. 3 Semi-variance
values for arbitrary
distance. Solid circle,
species richness; open
circle, residual from GLM
Frequency
Total
Orchid
Pteridophyte
100
80
60
40
20
0
60
50
40
30
20
10
0
100
80
60
40
20
0
1
2 3 4 5
Growing site
1
2
Total
Semi-variance
Fig. 2 Frequency of
appearance at each growing
site. Growing site is
indicated by numerals: 1,
the basal part of the trunk;
2, the trunk from 2 m up to
the first ramification; 3, the
basal part of the canopy; 4,
the middle part of the
canopy; 5, the outer part of
the canopy
3
4
5
1
Orchid
2
3
4
5
Pteridophyte
6
10
8
6
4
2
0
1.5
4
1.0
2
0.5
0
0.0
0
10 20
30
Distance (m)
40
0
10
20 30
40
0
10
20
30 40
Forest Ecology
pteridophytes may disperse over a wider range than
the distances examined in our study. On the other
hand, the spatial trend suggested by the residual of
the best model was more consistent than the pattern
associated with species richness. Therefore, it is
possible that habitat quality might have more influence than dispersal ability at our study site, although
we did not compare the two factors directly.
Effect of host size and host species preference
The positive effect of host size seems to reflect
habitat stability in the long-term in addition to habitat
size. Because epiphyte growth and colonization are
slow (Schmidt and Zotz 2002; Laube and Zotz 2003),
it seems that habitat stability is one of the important
factors for epiphyte establishment. It is known that
the diversity of vascular epiphyte species is significantly higher in primary forest than in secondary
forest (Barthlott et al. 2001; Ishida et al. 2005;
Benavides et al. 2006). One possible reason is that
forest structure developed over a long time, results in
more complex, extensive, and stable habitats for
epiphytes.
The host specificity of vascular epiphytes has been
shown in previous studies, and it has been suggested
that this can be explained by bark characteristics,
such as water-holding capacity, nutrient status, and
chemical composition (Frei and Dodson 1972;
Callaway et al. 2002; Mehltreter et al. 2005). The
results of our study also suggest that vascular
epiphytes have preferred host species (Table 2). Most
epiphyte groups preferred deciduous host trees. Not
one group displayed a preference for D. racemosum
(Table 2). Because deciduous trees have thin leaves
than evergreen trees and lose their leaves in the
winter, they provide a lighter habitat. Therefore, light
conditions in the inner canopy may also be an
important factor affecting host species preference.
Although D. racemosum was the most common
species in our study plot, it had the lowest associated
epiphyte diversity. In contrast, there were relatively
few deciduous trees in our study site, because most
are pioneer species and are found on disturbed sites in
temperate evergreen broad-leaved forest (Tanouchi
and Yamamoto 1995). Nevertheless, deciduous trees
provided an important epiphyte habitat. These results
suggest that a diverse forest community helps to
maintain epiphyte richness.
253
The water holding capacity of bark may also be an
important factor affecting host preference by vascular
epiphytes at our study site, since it is not cloud forest
and experiences only moderate humidity. Callaway
et al. (2002) have shown that water availability is a key
factor in determining a good host; they argued that the
host preference of vascular epiphytes may change with
changing humidity. Among the most important factors
for epiphyte richness, the positive effect of host size
probably does not vary with environmental gradient or
location. However, epiphyte host preference may vary
along an environmental gradient.
Conclusion
Size and species of host tree were the most important
factors influencing vascular epiphyte diversity, and it
appears that both habitat suitability and quality are
important for the establishment of vascular epiphytes.
On the other hand, epiphyte diversity did not exhibit
clear spatial aggregation, and it appears that dispersal
limitation is not particularly influential at the scale of
our study.
Acknowledgments We thank H. Nomiya, K. Kawano, N.
Kawano, Y. Cheng, M. Kawagoe, and K. Hashiba for their
support and helpful suggestions for our study. We also thank
the Miyazaki District Forestry Office for allowing the use of
their facilities for our study. For analysis, we used some data of
Forest Dynamics Data Base (FDDB) established by FFPRI
(Forestry and Forest Products Research Institute) and JST
(Japan Science and Technology Agency).
References
Bailey TC, Gatrell AC (1995) Interactive spatial data analysis.
Longman, Essex
Barthlott W, Schmit-Neuerburg V, Nieder J, Engwald S (2001)
Diversity and abundance of vascular epiphytes: a comparison of secondary vegetation and primary montane rain
forest in the Venezuelan Andes. Plant Ecol 152:145–156.
doi:10.1023/A:1011483901452
Benavides AM, Wolf JHD, Duivenvoorden JF (2006) Recovery and succession of epiphytes in upper Amazonian
fallows. J Trop Ecol 22:705–717. doi:10.1017/S02664674
06003580
Benzing DH (1990) Vascular epiphytes. Cambridge University
Press, Cambridge
Burns KC, Dawson J (2005) Patterns in the diversity and distribution of epiphytes and vines in a New Zealand forest.
Austral Ecol 30:883–891. doi:10.1111/j.1442-9993.2005.
01532.x
254
Callaway RM, Reinhart KO, Moore GW, Moore DJ, Pennings
SC (2002) Epiphyte host preferences and host traits:
mechanisms for species-specific interactions. Oecologia
132:221–230. doi:10.1007/s00442-002-0943-3
Dawson JW, Sneddon BV (1969) The New Zealand rain forest: a
comparison with tropical rain forest. Pac Sci 23:131–147
Flores-Palacios A, Garcı́a-Franco JG (2006) The relationship
between tree size and epiphyte species richness: testing
four different hypotheses. J Biogeogr 33:323–330. doi:
10.1111/j.1365-2699.2005.01382.x
Fortin M-J, Dale M (2005) Spatial analysis: a guide for ecologists. Cambridge University Press, Cambridge
Frei JK, Dodson CH (1972) The chemical effect of certain bark
substrates on the germination and early growth of epiphytic orchids. Bull Torrey Bot Club 99:301–307. doi:
10.2307/2997072
Gentry AH, Dodson CH (1987) Diversity and biogeography of
neotropical vascular epiphytes. Ann Mo Bot Gard 74:205–
233. doi:10.2307/2399395
Hietz P, Hietz-Seifert U (1995) Composition and ecology of
vascular epiphyte communities along an altitudinal gradient in central Veracruz, Mexico. J Veg Sci 6:487–498.
doi:10.2307/3236347
Ishida H, Hattori T, Hashimoto Y (2005) Comparison of species
composition and richness among primeval, natural, and
secondary lucidophyllous forests in southeastern Kyushu,
Japan. Veg Sci 22:71–86
Johansson D (1974) Ecology of vascular epiphytes in West
African rain forest. Acta Phytogeogr Suec 59:1–129
Laube S, Zotz G (2003) Which abiotic factors limit vegetative
growth in a vascular epiphyte? Funct Ecol 17:598–604.
doi:10.1046/j.1365-2435.2003.00760.x
Laube S, Zotz G (2006) Neither host-specific nor random:
vascular epiphytes on three tree species in a Panamanian
lowland forest. Ann Bot (Lond) 97:1103–1114. doi:
10.1093/aob/mcl067
Löbel S, Snäll T, Rydin H (2006) Species richness patterns and
metapopulation processes—evidence from epiphyte
communities in boreo-nemoral forest. Ecography 29:169–
182. doi:10.1111/j.2006.0906-7590.04348.x
Machon N, Bardin P, Mazer SJ, Moret J, Godelle B, Austerlitz
F (2003) Relationship between genetic structure and seed
and pollen dispersal in the endangered orchid Spiranthes
spiralis. New Phytol 157:677–687. doi:10.1046/j.14698137.2003.00694.x
McCullagh P, Nelder JA (1989) Generalized linear models.
Chapman & Hall, London
Mehltreter K, Flores-Palacios A, Garcı́a-Franco JG (2005) Host
preferences of low-trunk vascular epiphytes in a cloud
forest of Veracruz, Mexico. J Trop Ecol 21:651–660. doi:
10.1017/S0266467405002683
Migenis LE, Ackerman JD (1993) Orchid-phorophyte relationships in a forest watershed in Puerto Rico. J Trop Ecol
9:231–240
Miyazaki Local Meteorological Observatory (eds) (1951–
1997) Monthly report of meteorology. Japan Weather
Association Miyazaki Branch, Miyazaki (in Japanese)
A.G. Van der Valk (ed.)
Nieder J, Engwald S, Klawun M, Barthlott W (2000) Spatial
distribution of vascular epiphytes (including Hemiepiphytes) in a lowland amazonian rain forest (Surumoni
Crane Plot) of southern Venezuela. Biotropica 32:385–396
Ohnuki Y, Terazono R, Ikuzawa H, Hirata I, Kanna K, Utagawa H (1997) Distribution of colluvia and saprolites and
their physical properties in a zero-order basin in Okinawa,
southwestern Japan. Geoderma 80:75–93. doi:10.1016/
S0016-7061(97)00076-1
R Development Core Team (2006) R: a language and environment for statistical computing. R foundation for
statistical computing, Vienna
Saito S (2002) Effects of a severe typhoon on forest dynamics
in a warm-temperate evergreen broad-leaved forest in
Southwestern Japan. J For Res 7:137–143. doi:10.1007/
BF02762602
Sato T, Kominami Y, Saito S, Niiyama K, Manabe T, Tanouchi
H, Noma N, Yamamoto S (1999) An introduction to the
Aya Research Site, a long-term ecological research site, in
a warm temperate evergreen broad-leaved forest ecosystem in southwestern Japan: research topics and design.
Bull Kitakyushu Mus Nat Hist 18:157–180
Schmidt G, Zotz G (2002) Inherently slow growth in two
Caribbean epiphytic species: a demographic approach. J
Veg Sci 13:527–534. doi:10.1658/1100-9233(2002)013
[0527:ISGITC]2.0.CO;2
Tamura T (1987) Landform-soil features of the humid temperate hills. Pedologist 31:135–146 in Japanese
Tamura T, Takeuchi K (1980) Land characteristics of the hills
and their modification by man—with special reference to
a few cases in the Tama hills, west of Tokyo. Geogr Rep
Tokyo Metropol Univ 14/15:49–94
Tanouchi H, Yamamoto S (1995) Structure and regeneration of
canopy species in an old-growth evergreen broad-leaved
forest in Aya district, southwestern Japan. Vegetatio
117:51–60. doi:10.1007/BF00033258
Tremblay RL (1997) Distribution and dispersion patterns of
individuals in nine species of Lepanthes (Orchidaceae).
Biotropica 29:38–45. doi:10.1111/j.1744-7429.1997.tb00
004.x
Winter K, Wallace BJ, Stocker GC, Roksandic Z (1983)
Crassulacean acid metabolism in Australian vascular
epiphytes and some related species. Oecologia 57:129–
141. doi:10.1007/BF00379570
Zimmerman JK, Olmsted IC (1992) Host tree utilization by
vascular epiphytes in a seasonally inundated forest (Tintal) in Mexico. Biotropica 24:402–407. doi:10.2307/238
8610
Zotz G, Vollrath B (2003) The epiphyte vegetation of the palm
Socratea exorrhiza—correlations with tree size, tree age
and bryophyte cover. J Trop Ecol 19:81–90. doi:10.1017/
S0266467403003092
Zotz G, Ziegler H (1997) The occurrence of crassulacean acid
metabolism among vascular epiphytes from Central Panama. New Phytol 137:223–229. doi:10.1046/j.1469-8137.
1997.00800.x
Seed bank composition and above-ground vegetation
in response to grazing in sub-Mediterranean oak forests
(NW Greece)
Evgenia Chaideftou Æ Costas A. Thanos Æ
Erwin Bergmeier Æ Athanasios Kallimanis Æ
Panayotis Dimopoulos
Originally published in the journal Plant Ecology, Volume 201, No. 1, 255–265.
DOI: 10.1007/s11258-008-9548-1 Springer Science+Business Media B.V. 2008
Abstract We investigate the persistent soil seed
bank composition and its relation to the above-ground
flora of grazed and non-grazed sub-Mediterranean
deciduous oak forests of NW Greece. Twenty-eight
taxa were recorded in the soil seed bank and 83 taxa
(70 taxa in plots of seed bank sampling) in the aboveground vegetation. The dominant tree species and
many woodland species found in the above-ground
vegetation were absent from the soil seed bank.
Similarity between the soil seed bank and the aboveground vegetation decreased with grazing, and
E. Chaideftou A. Kallimanis P. Dimopoulos (&)
Department of Environmental and Natural Resources
Management, University of Ioannina, Seferi 2,
30100 Agrinio, Greece
e-mail: pdimopul@cc.uoi.gr
grazing led to a decrease of species richness in
above-ground vegetation and soil seed bank. Beta
diversity of vegetation among grazed and among nongrazed plots did not differ, but was significantly
higher between grazed and non-grazed areas. Beta
diversity of the soil seed bank declined with grazing.
When applying classification tree and logistic regression analyses, non-grazed forest sites are clearly
differentiated by the presence of Phillyrea latifolia,
Euphorbia amygdaloides and Brachypodium sylvaticum. PCA ordination of above-ground species
composition reflected a gradient from sites grazed
by ruminants to non-grazed sites, but no clear
structure was detected in the seed bank.
Keywords Soil seed bank Wood pasture
Grazing Browsing Deciduous oak forests
Greece Wild boar Ruminants
E. Chaideftou
e-mail: me01390@cc.uoi.gr
A. Kallimanis
e-mail: akallim@cc.uoi.gr
C. A. Thanos
Faculty of Biology, Department of Botany, National and
Kapodistrian University of Athens, Panepistimiopolis,
15784 Athens, Greece
e-mail: cthanos@biol.uoa.gr
E. Bergmeier
Albrecht von Haller Institute of Plant Sciences,
Georg-August University of Göttingen,
Untere Karspüle 2, 37073 Göttingen, Germany
e-mail: erwin.bergmeier@bio.uni-goettingen.de
Introduction
The composition of a seed bank depends on present
and former above-ground vegetation (Rice 1989) and
seed rain from adjacent areas (Hutchings and Booth
1996). The historical composition of above-ground
vegetation has often been identified as a key factor
determining seed bank composition (Bekker et al.
1998). Soil seed bank composition is also influenced
by the surrounding vegetation and former
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_20
255
256
successional stages (Falinska 1999; Godefroid et al.
2006) and can undergo significant changes depending
on the management applied (Wellstein et al. 2007).
However, there is still considerable lack of knowledge on the seed bank characteristics of many
species, including typical forest species (Bossuyt
and Hermy 2001), and seed viability in forest soils
(Grandström 1987; Thompson et al. 1997).
Mediterranean environments have undergone
many changes due to human disturbances, such as
sylvopastoralism (Le Houérou 1990). The effects of
herbivory vary due to parameters such as intensity,
plant taxa and soil properties. Species composition in
European beech forests grazed by high densities of
Cervus elaphus, Cervus dama, Capreolus capreolus,
Ovis musimon and Sus scrofa has been significantly
determined by two pathways: dispersal in time by a
persistent seed bank, and dispersal in space using
ungulates (Naaf and Wulf 2007). In the agricultural
landscape of central Europe wild boars epizoochorously transport large amounts of seeds due to their fur
characteristics, behaviour (wallowing and rubbing on
trees), large local population size and general abundance (Heinken et al. 2006).
Ungulate species such as Cervus elaphus, Capreolus capreolus and Sus scrofa affect natural
regeneration of forests throughout Europe. Selective
browsing promotes changes in forest tree composition
(Kuiters and Slim 2002). However, boars have different grazing behaviour from ruminants. Wild boars
prefer acorns as a food source (Jedrzejewska et al.
1997), and forage on germinating oak seedlings,
saplings and roots by digging, thus affecting both
regeneration and soil properties (Groot Buinderink and
Hazebroek 1996). Ruminants such as Capreolus
capreolus and Cervus elaphus browse on seedlings,
leading and lateral shoots causing structural changes
(Pépin et al. 2006) and rub trunks resulting in tree
damage (Ramos et al. 2006).
The role of browsing and grazing in woodland
regeneration, and the long distance dispersal of seeds
has been studied extensively. However, the impact of
ruminant and boar grazing on soil seed bank composition, and its similarity to above-ground vegetation
has received less attention. In different types of
grasslands, grazing increases (Bakker and de Vries
1992; Ungar and Woodell 1996), decreases (Jutila
1998) or has no effect on (Peco et al. 1998) the
similarity of seed banks and above-ground vegetation
A.G. Van der Valk (ed.)
(Osem et al. 2006). For forests, discrepancies
between above-ground vegetation and soil seed banks
have been recorded (Thompson and Grime 1979;
Bossuyt et al. 2002; Forrester and Leopold 2006;
Roovers et al. 2006) and attributed to disturbance
(Olano et al. 2002; Godefroid et al. 2006).
Earlier studies have investigated the seed bank
composition of either undisturbed forests (Warr et al.
1994; Kjellsson 1992), or the relationship between
seed bank composition and land use (Bossuyt and
Hermy 2001; Brown and Oosterhuis 1981). Few
studies on soil seed banks have investigated the
impacts of differences and changes in management
practices (Wellstein et al. 2007) and none, to our
knowledge, have researched the impacts of overgrazing and different grazing regimes in sub-Mediterranean oak forests.
In the present study, we examine the hypothesis
that long-term (more than 30 years) overgrazing
affected not only the above-ground vegetation but
also the seed bank of the seeds accumulated in the
soil. A secondary goal of this study was to test the
hypothesis that ruminant and non-ruminant grazers
affect the soil seed bank and the above-ground
vegetation in different ways. Above-ground vegetation and soil seed bank were analysed at the levels of
species composition and richness (alpha diversity),
and species turnover (beta diversity). The practical
application of these findings is in the field of restoring
heavily grazed woodlands in the Mediterranean
region. Therefore, our results are discussed from the
restoration point of view, to assess the potential role of
soil seed banks in contributing to vegetation restoration after the cessation of overgrazing pressure.
Materials and methods
Study site
The research area is a deciduous mixed broad-leaved
forest in north-western Greece (Bourazani area,
municipality of Konitsa, Epirus), close to the Albanian
border (40020 N, 20380 E). The forest was coppiced
until a few decades ago. It consists of chiefly deciduous
sub-Mediterranean thermophilous tree species with
high proportions of Quercus frainetto, Q. pubescens,
Carpinus orientalis and Fraxinus ornus, and scattered
Quercus cerris and Q. trojana, while Q. coccifera,
Forest Ecology
Phillyrea latifolia, Cotinus coggygria and Juniperus
oxycedrus are common in the shrub layer (Tsaliki
et al. 2005). The forests represent subtypes of the
south-western Balkan association Verbasco glabratiQuercetum frainetto (Quercion frainetto, Quercetalia
pubescentis) (Bergmeier and Dimopoulos 2008).
The substrate is flysch locally substituted by
limestone. The soils are shallow (15–30 cm) or of
medium depth (30–60 cm). The topography of the
study sites is hilly to mountainous (400–700 m a.s.l.).
The climate is classified to sub-Mediterranean with a
4-month-long dry period (end of May to September)
and about 700 mm average annual precipitation.
Mean monthly temperatures range between 5C in
winter and 24C in summer. Mean monthly rainfall
ranges between 12 mm in July and 135 mm in
December (Tsaliki et al. 2005).
The study site includes (A) a fenced, private
wooded area (112 ha) grazed continuously since
1974 by ruminants and wild boar (Sus scrofa) in
high but varying population densities, and (B) a nonfenced and non-grazed forest.
Site A, the fenced forest area (112 ha), was further
subdivided into site A1 (26 ha) that was continuously
grazed by wild boar and site A2 (86 ha) that was grazed
by ruminants: Dama dama, Cervus elaphus, Capreolus
capreolus, Ovis musimon and Capra hircus cretica.
This subdivision and grazing practice precede our
study by at least 30 years. As a result of over-grazing,
the herb and litter layer of the oak woodland has almost
completely vanished, soils are bare, compressed and
eroded, and tree roots protrude from the ground.
Site B is adjacent to site A. During our study
period, its vegetation cover and other ecological
characteristics were similar to those of site A;
however, the site was not systematically grazed.
Therefore, samples from site B were used as controls
(i.e. ungrazed sites) in the present study. This
combination of overgrazed forests adjacent to undisturbed forests of identical climatic, geological and
topographical conditions allows us to study the effect
of grazing independently of other environmental
parameters and is unique in the wider region.
Sampling
Above-ground vegetation was investigated in 42
permanent plots of 150 m2. The plots were established in both grazed (22 plots, of which nine were
257
grazed by wild boar and 13 by ruminants) and nongrazed (20 plots) forest sites.
In each plot, plant species composition (alpha
diversity) was recorded twice: during spring-summer
of 2004 and autumn of 2005. The data were
combined into one data set so that the maximum
number of species occurring in the above-ground
vegetation of each plot is taken into account.
Of the 42 permanent vegetation plots, six plots
were selected to sample the seed banks, and 20
sample soil cores were taken from each plot. The six
plots were classified into the three types of grazing
regime as follows:
Type 1: grazing by ruminants (R: ruminant regime
sampled at two plots in site A2);
Type 2: grazing by wild boar (B: non-ruminant
regime, sampled at one plot in site A1);
Type 3: no grazing by ruminants or boars (C:
control, i.e. no grazing, sampled at three plots in
site B).
As our research focused on the effects of grazing,
our sampling scheme comprised three grazed plots
(site A) and three ungrazed control plots (site B). In
the grazed treatment there were two discrete subareas grazed by different species. In addition to the
main research topic (grazed versus non-grazed), we
regarded the two grazing treatments as well.
To assess the persistent seed bank composition, we
collected soil samples at the end of May 2004, when
germination had ended and before any new seeds
were dispersed. Soil cores were sampled at two
depths: 0–5 cm (upper layer) and 5–10 cm (deeper
layer). In each plot we collected twenty soil samples,
10 for each depth (i.e. a total of 120 soil samples for
the six plots studied). The quantitative and qualitative
composition of the seed bank was investigated using
the seedling emergence method (Thompson et al.
1997) and with the additional prior application of a
3-month period of artificial stratification, the soil
samples were stored wet in a refrigerator, in the dark
(3–5C). The seedling emergence method, although
laborious, is considered more reliable than elutriation
for determining the species composition of the seed
bank of a plant community (Gross 1990). Emerging
seedlings were counted at regular intervals and, at a
later developmental stage, identified to the closest
taxonomic level possible (about 80% of the soil seed
bank taxa were identified to the species level).
258
A.G. Van der Valk (ed.)
Data analysis
Species turnover (beta diversity) analysis
Similarity in species composition between seed bank
and above-ground vegetation under different grazing
regimes was assessed by Sørensen’s qualitative
similarity index (Kent and Coker 1994; Magurran
2004). We compared the values of the similarity
index using the Mann–Whitney test. To measure the
effect of grazing on the species richness of our
samples, we used the Kruskall–Wallis test.
In order to descriptively display a structure of
species composition possibly related to the grazing
regimes, the indirect linear response model was used.
Explorative ordinations (Principal Component Analysis, PCA) were carried out on the soil seed bank and
above-ground vegetation species data, using CANOCO for Windows (ter Braak and Šmilauer 2002). All
analyses were scaled on inter-species correlations and
species-centred by dividing species scores by their
standard deviation to obtain correlation matrices.
Community parameters such as total cover were not
taken into consideration, since our interest was
focused on the presence of species with respect to
possible grazing effects.
To test if the grazing regimes affect the species
composition, we compared the samples from the
different grazing regimes using two statistical
methods: logistic regression and classification tree
analysis. Classification trees have recently been
applied to the analysis of ecological data (e.g.
De’ath and Fabricius 2000; Kallimanis et al. 2005,
2007). They predict the value of a response variable
(grazing regime in this study), from the values of a
set of explanatory variables, which may be either
numerical or categorical (Witten and Frank 2005).
The basic assumption of this method is that the
functional dependency among system variables is
not uniform in the whole domain, but can be
approximated as such on smaller sub-domains.
Classification trees are induced by recursively
dividing the data set to more homogeneous subsets.
At each repetition, the most informative attribute is
identified, and the data set is divided according to
the values of that attribute. This process is repeated
for each subset until pure datasets (i.e. datasets
where all examples have the same value) or datasets
that cannot be divided further are reached. Those
datasets are the terminal ‘‘leaves’’ of our tree.
Beta diversity represents the spatial turnover of
species and is a measure of changes in the species
composition between two assemblages. There is a
lack of agreement in the literature as to the feature of
the pervasive spatial turnover in the identities of
species that beta diversity is intended to capture,
therefore there are several indices of beta diversity
(see Koleff et al. (2003) for an extensive review on
the subject). In this study, we estimated beta diversity
according to the Colwell and Coddington (1994)
index, which was calculated with the formula:
n
1X
ai
b¼
1
n i¼1
bi
where for every pairwise comparison i we estimated
the number of species simultaneously present in both
plots (ai) and the total number of species recorded in
the two plots (bi), and n the total number of pairwise
comparisons.
We analysed the species composition of six plots
(three grazed and three ungrazed). To compare
the patterns of beta diversity of seed bank and the
patterns of beta diversity of vegetation, we used the
same six plots for both analyses. Among those six
plots there are a total of 15 possible pairwise
comparisons. Three of these comparisons are among
grazed plots and represent the beta diversity within
grazed plots; three comparisons are among the
ungrazed plots and represent the beta diversity within
ungrazed plots. The nine remaining pairwise comparisons among grazed and ungrazed plots represent
the beta diversity of the transition among grazed and
ungrazed areas, i.e. between grazed and ungrazed
treatments. Beta diversity was estimated for both
above-ground vegetation and the soil seed bank.
Results
Effect of grazing on seed bank-vegetation
similarity
She above-ground vegetation under different grazing
regimes comprised 83 taxa, of which 70 taxa
occurred in the above-ground flora of the six seed
bank sampling plots; only 30% of these (21 taxa)
Forest Ecology
were represented in the persistent soil seed bank, in
which a total of 28 taxa were recorded (Table 1).
Thus, 75% of the species found in the seed bank were
also observed in the above-ground vegetation.
When comparing alpha diversity estimated as
species richness of the samples (number of species
per sample) from different grazing regimes, i.e. the
species richness at the finest scale, we found that the
difference was significant for both the above-ground
vegetation (Kruskal Wallis P \ 0.0001) and the soil
seed bank (P = 0.0003). In the samples from control
(non-grazed) plots more species were observed than
in the samples from the grazed plots. Species richness
in samples from the two grazing regimes did not
differ significantly.
Next we analysed the extent of overlap between
the species composition of the above-ground vegetation and the seed bank flora in each plot, i.e. how
many of the species in the above-ground vegetation
were present as seeds in the soil seed bank of each
plot. Similarity of the above-ground vegetation with
the seed bank flora varied considerably among the
different plots (Table 2): it was higher in the nongrazed areas, and significantly lower in the grazed
ones (Mann–Whitney P = 0.046 for the soil seed
bank). In the grazed areas, similarity between aboveground vegetation and seed bank did not exceed 19%,
and in the plots grazed by wild boar the similarity
was 0%. In the non-grazed areas the similarity index
reached 29%. Higher similarity was observed
between ruminant and wild boar sites in aboveground vegetation (Table 2).
By applying logistic regression and classification
tree analyses, we analysed the effect of grazing
regimes on the species composition of the aboveground vegetation and the soil seed bank. Both
approaches yielded similar results. Above-ground
vegetation was clearly distinguished with small
misclassification errors (12% for the tree model and
19% for the logistic regression) and high kappa
statistic (0.8 and 0.7, respectively). The absence of
Phillyrea latifolia indicates grazing by ruminants,
while its presence in combination with the absence of
Euphorbia amygdaloides and Brachypodium sylvaticum indicates grazing by wild boar (Fig. 1). The
combined presence of Phillyrea latifolia and either
Euphorbia amygdaloides or Brachypodium sylvaticum or both indicates non-grazed control plots.
Contrary to the above-ground vegetation, seed bank
259
Table 1 Taxa found in three grazing regimes (R: ruminant, B:
boar, C: control, i.e. non-grazed) for above-ground vegetation
and soil seed banks in the six sampling plots
Grazing regime
R
B
C
Taxa found only in the above-ground vegetation
Acer campestre L.
Acer monspessulanum L.
9
9
9
Arbutus unedo L.
9
Aremonia agrimonoides (L.) DC.
9
Asparagus acutifolius L.
Bituminaria bituminosa (L.) C. H.Stirt.
9 9
9
Brachypodium sylvaticum (Hudson) Beauv.
9
Clematis vitalba L.
9
Clinopodium vulgare L.
9
Colutea arborescens L.
9
Cornus mas L.
Corylus colurna L.
9
9
Cotinus coggygria Scop.
9
9 9
Crocus chrysanthus (Herb.) Herb.
9
Cyclamen hederifolium Aiton
9
Echinops ritro L.
9
Epipactis microphylla (Ehrh.) Swartz
9
Galium lucidum All.
9
Geranium brutium Gasp.
9
Geranium purpureum Vill.
9
Hedera helix L.
Helleborus odorus subsp.
cyclophyllus (A. Braun) Strid
9
9
9
Juniperus oxycedrus L.
9
Lapsana communis L.
9
Lathyrus niger (L.) Bernh.
9
Lathyrus nissolia L.
9
Melittis melissophyllum L.
Muscari neglectum Ten.
9
9
Osyris alba L.
9
Phillyrea latifolia L.
Poa trivialis L. subsp.
sylvicola (Guss.) Lindb. fil.
9 9
9
Potentilla micrantha DC.
Quercus cerris L.
9
9
9
Quercus coccifera L.
9 9
9
Quercus frainetto Ten.
9
Quercus pubescens Willd.
9
9 9
Quercus trojana Webb
9
9 9
9 9
Rosa gallica L.
9
Ruscus aculeatus L.
9
260
A.G. Van der Valk (ed.)
Table 1 continued
Table 1 continued
Grazing regime
Grazing regime
R
R
B
Sorbus domestica L.
C
9
Sorbus torminalis (L.) Crantz
9 9
Tamus communis L.
Sonchus asper (L.) Hill
B
C
?
?
Vicia cassubica L.
?
9
Number of taxa in each grazing regime
2
2
4
Tanacetum corymbosum (L.) Schultz Bip.
9
Total number of taxa in category: 7
Thymus longicaulis C. Presl
9
Total number of taxa in each grazing regime 21
Torilis arvensis (Hudson) Link
9
Total number of taxa in Table: 77
Trifolium ochroleucon Hudson
9
Trifolium pallidum Waldst. & Kit.
9
? Indicates presence in the soil seed bank and 9 indicates
presence in the above-ground vegetation
Trifolium tenuifolium Ten.
9
Vicia sativa L.
9
Number of taxa in each grazing regime
9
10 45
Total number of taxa in category: 49
Taxa common to the above-ground
vegetation and soil seed banks
Campanula spec.
?
Carex flacca Schreb.
9
9
19 70
Table 2 Sørensen similarity index (%) between soil seed bank
and above-ground vegetation in different grazing regimes
Type
Ruminant
Boar
Control
Ruminant
19sv
36v
0s
0sv
24v
32s
13s
29sv
23v
? ?/9
Boar
9 ?/9
Control
Cercis siliquastrum L.
9 ?
Crataegus monogyna Jacq.
9 ?/9
v, Similarity among different types of above-ground
vegetation; s, similarity among different soil seed banks; sv,
similarity between soil seed bank and above-ground vegetation
of the same type
Carpinus orientalis Mill.
Dactylis glomerata L.
?
?/9
Dorycnium hirsutum (L.) Ser.
?/9
Euphorbia amygdaloides L.
?
?/9
Fraxinus ornus L.
9
9 ?/9
Galium aparine L.
?/9
Gramineae
?/9
Inula salicina L.
?/9
Lathyrus laxiflorus (Desf.) O. Kuntze
Luzula forsteri (Sm.) DC.
?/9
?/9
?
Medicago lupulina L.
Silene italica (L.) Pers.
?/9
?
Trifolium arvense L.
?/9
Trifolium campestre Schreb.
?/9
?/9 9 ?/9
Ruminant
Grazing
present
Brachypodium sylvaticum
absent
present
9
Veronica chamaedrys L.
?/9 9 ?/9
Viola alba Besser
?
Number of taxa in each grazing regime
11
Euphorbia amygdaloides
absent
No Grazing
present
?/9
8
21
Wild Boar
Grazing
Total number of taxa in category: 21
Taxa found only in the soil seed banks
Caryophyllaceae
?
Parietaria judaica L.
?
Petrorhagia cf. saxifraga (L.) Link
Phillyrea latifolia
absent
? ?/9
Trifolium physodes Bieb.
Rubus sanctus Schreb.
Solanum nigrum L.
Bold values represent the seed bank-vegetation similarity of
the same type of grazing regime (i.e. between ruminants,
between boar, between control plots)
?
?
?
No Grazing
Fig. 1 Classification tree of the grazing regimes (ruminant,
boar and no-grazing) on the basis of indicator plants. Each
‘‘leaf’’ is labelled according to presence-absence of the species
Phillyrea latifolia, Brachypodium sylvaticum and Euphorbia
amygdaloides in the above-ground vegetation. The misclassification error was minor (12%)
Forest Ecology
261
Table 3 PCA analysis on the above-ground vegetation;
eigenvalues from ordination of plots for axes 1–4
Species turnover (beta diversity)
Axes
Axis 1 Axis 2 Axis 3 Axis 4
Eigenvalues
0.249
For the above-ground vegetation beta diversity among
grazed plots did not differ significantly from beta
diversity among non-grazed plots. However, the
species turnover between grazed and non-grazed plots
was significantly higher than the beta diversity within
both grazing regimes (Kruskal–Wallis P = 0.006).
For the soil seed bank the differences in beta
diversity were found to be statistically significant
(Kruskal–Wallis P = 0.005). More precisely, beta
diversity of grazed plots was significantly higher than
that of non-grazed plots. Beta diversity between grazed
and non-grazed plots was of intermediate value.
Cumulative percentage
24.9
variance of species data
0.107
0.083
0.066
35.6
43.9
50.5
Total inertia: 1.000
species composition was not distinguishable by either
logistic regression or classification tree analyses.
Examining the presence of particular species of the
seed bank in more detail we find that frequent species
like Cotinus coggygria, Juniperus oxycedrus, Phillyrea latifolia, Quercus frainetto and Q. pubescens are
absent from the soil seed bank of the study area.
Other less frequent species like Hedera helix,
Clematis vitalba and Lapsana communis were also
absent from the seed bank of the study area. On the
other hand, some of the above-ground woody species
Carpinus orientalis, Cercis siliquastrum, Crataegus
monogyna, Fraxinus ornus and Rubus sanctus were
found in the soil seed bank.
Ordinations of above-ground vegetation and soil
seed banks
PCA ordinations were performed on the species
presence data of the above-ground vegetation and soil
seed bank separately. The first two PCA axes of the
data set of the above-ground vegetation account for
36% of variance (see Table 3), a relatively low
proportion that reflects a heterogeneous vegetation
gradient structure. The first axis explains 25% of the
variance reflecting a gradient of grazing intensity
from the plots under heavy grazing conditions (left
side of the diagram) to the non-grazed plots (right
side of the diagram) (Fig. 2).
PCA on the soil seed bank composition did not
reveal a clear structure of the identified species in
relation to the differently grazed forest sites. In the
ordination diagram of soil seed bank (not shown),
species of grazed plots (left part along the first axis)
were separated from species of non-grazed plots
(right part along the first axis). The first two axes
explained the bulk of the variance (31%) compared to
the total variance of 49% explained by the first four
axes (eigenvalues for the first four axes: 0.158, 0.150,
0.097 and 0.083, respectively).
Discussion
Seed banks and above-ground vegetation
compared
Approximately two-thirds of the taxa found in the
vegetation did not occur in the soil seed bank of the
study area; on the other hand, three-quarters of the
soil seed bank taxa were found in the above-ground
vegetation. This confirms the generally low similarity
between above-ground vegetation and persistent soil
seed bank floras in forest ecosystems, and that the
above-ground vegetation does not necessarily reflect
the soil seed bank composition (Olano et al. 2002).
To our knowledge, this dissimilarity is reported for
the first time in a sub-Mediterranean woodland. As
expected, small-seeded species dominate the seed
bank flora, while large-seeded species dominate the
woody above-ground vegetation of the studied forest.
Roovers et al. (2006) observed a similar pattern in a
temperate mesophilous deciduous forest.
Looney and Gibson (1995) report that only few
tree taxa of the above-ground vegetation were found
in the soil seed bank, a fact attributed to animal
predation and dormancy (Shen et al. 2007). Similarly, in our study some of the most frequent species
like Cotinus coggygria (anemochorous), Juniperus
oxycedrus, Phillyrea latifolia (both zoochorous),
Quercus frainetto and Q. pubescens (both dispersed
by gravity) are absent from the soil seed bank of the
study area (not surprising of course for the latter two
species, well known to bear recalcitrant seeds). Other
species like Hedera helix are absent from the seed
262
A.G. Van der Valk (ed.)
Fig. 2 Ordination (PCA)
species-samples diagram
(species data set of 42
above-ground vegetation
plots) along axes 1 and 2
(eigenvalues for the first
two axes 0.249 and 0.107,
respectively). The species
are labelled by the first three
letters of the generic name
and the first three letters of
the species epitheta (see
Table 1 for full names).
Plots are displayed as: d
boar-grazed plots; 9
ruminant-grazed plots; j
control (i.e. non-grazed)
plots
Above-ground vegetation
Axis 2 1.0
Sordom
Sortor
Eupamy
Cramon
Carfla
Quepub Potmic
Ptistr
Selsil
Fraorn
Palspi
Sciaut
-0 .6
Helcyc
Vioalb Rosgal
Corcol
Galluc
Torarv
Cotcog
Galapa
Triten Inusal
Carori
Acemon
Vergla
Quefra
Dacglo
Latnis
Dicalb
Gerpur Philat Luzfor
1 .0
Litpur
Vercha
Acecam
Campan
Cersil
Axis
1
Silita Clevit Latnig
Poatri
Trioch
Tripur
Tripal
Triphy Quecoc
Vicsat
Pister Lotspe
Thylon
Quecer Musneg Crochr Tancor
Melmel
Tricam
Orivul
Gerbru
Brasyl
Lapcom Clivul Tamcom
Bitbit
Colarb
Gramin
Ciscre Epimic
Osyalb
Areagr
Dorhir
Echrit
Latlax
Teucha
Aspacu Medlup
Junoxy Arbune Triarv
Rusacu
Cyched
Hedhel
Cormas
Quetro
Grazed
Non-grazed
-1 .0
bank since they rarely produce seeds in shady habitats
(Buckley et al. 1997). Another group of species with
Clematis vitalba and Lapsana communis were absent
from the seed bank of the study area, although they
were found in seed banks of woody and disturbed
habitats elsewhere (Roovers et al. 2006).
Some of the above-ground woody species were
found in the soil seed bank: Carpinus orientalis,
Cercis siliquastrum, Crataegus monogyna, Fraxinus
ornus and Rubus sanctus. Fraxinus ornus and
Carpinus orientalis are common tree species in the
study area with noticeable regeneration in the
sapling layer, thus contradicting Forrester and Leopold’s (2006) observation that most of the dominant
canopy species appearing in the soil seed bank are
absent from the sapling and shrub layer of deciduous forests.
Effect of grazing on seed bank-vegetation
similarity
In the studied sub-Mediterranean forest, grazing
reduced the similarity between seed bank and
above-ground vegetation. Especially in sites with
wild boar, there were no common species between
seed bank and vegetation. This finding is consistent
with the general pattern of decreasing similarity
between seed bank and vegetation under grazing
(Marage et al. 2006; Haretche and Rodriguez 2006).
In temperate forests, Heinken et al. (2006) found large
Forest Ecology
numbers of seeds of chiefly non-forest species and
others which occur both in forests and open habitats
near trees rubbed by wild boar, and concluded that
most plant species were dispersed epizoochorously by
Sus scrofa. In our study, the wild boars were restricted
inside the fenced area and it was therefore impossible
for them to serve as long-distance vectors of diaspores
(from outside the forest).
Effect of grazing on species richness
The impact of grazing on seed bank species richness
and composition has been studied mainly in grasslands
and to a lesser extent in forests, scrub and rangelands.
In most studies species richness was found to decrease
with grazing pressure (Marage et al. 2006; Miller
1999). However, two studies geographically close to
our own (Heinken et al. (2006), temperate forest in
Germany; and Malo et al. (2000), Mediterranean
dehesas) showed the opposite, i.e. increase in the seed
bank diversity under grazing. In our study, we found
that species richness of the above-ground vegetation
and the soil seed bank declined with grazing, thus
confirming the general trend. We presume that
contradictory statements in literature might be due to
different grazing intensities and duration. Furthermore, our results indicate that long-term over-grazing
as such is the determining factor and not the particular
species of mammal, as the effects of ruminants and
boars did not differ significantly.
Effects of grazing on beta diversity
(species turnover)
The effect of grazing on beta diversity has recently
attracted the interest of researchers, but presently no
clear picture emerges from the literature. Although
most studies found no effect (see Harrison 1999;
Zhang 1998; Alrababah et al. 2007; Robson and Clay
2005), a few studies demonstrate increase (e.g.
Bakker and Ruyter 1981) and others decrease (e.g.
Chaneton et al. 2002) of beta diversity with grazing.
Our results show different effects of grazing on the
beta diversity of the vegetation and the seed bank.
Above-ground vegetation displayed no significant
difference in the beta diversity among grazed and
among non-grazed plots, but beta diversity was
significantly higher between grazed and non-grazed
plots. This finding and our community analysis
263
results indicate that grazed and non-grazed plots are
characterized by distinct species assemblages.
Although the grazed plots had fewer species, these
were not characterized by higher species turnover
compared to the control plots. So our finding
contradicts other studies that report increased beta
diversity in areas with decreased alpha diversity
(Kallimanis et al. 2008; Lennon et al. 2001).
Results for soil seed bank reflect a different
picture, since species turnover was significantly
higher in the grazed plots and our statistical analysis
failed to identify distinct communities in the different
grazing regimes, despite the existence of such
communities in the above-ground vegetation. It is
also indicative that the species turnover in the seed
bank between grazed and non-grazed plots is lower
than the respective one in the grazed plots. This
discrepancy might indicate that the main seed
dispersal mode is related to animals. This observation
allows us to suggest that the zoochorous mode of
dispersal and its role in shaping seed bank communities under grazing should be the focus of further
study in the future.
Seed bank and restoration implications
Soil seed bank appeared consistent with the conclusion of Godefroid et al. (2006) that there is no close
relationship between the species composition of the
seed bank and that of the established vegetation.
Thus, the seed bank is ‘capable’ of restoring the
studied forests only to a limited extent. Studies on
the restoration of forests through diaspores stored in
the soil have also been carried out by Oke et al.
(2006) and Warr et al. (1994), and the potential
contribution of the soil seed bank to restoration of
temperate deciduous forests has been recently investigated by Roovers et al. (2006).
Difficulties in the restoration of forests by soil seed
banks are to be expected when the similarity between
the above-ground species composition and that of the
seed bank proves to be poor. Most of the dominant or
frequent species of the herb layer rarely or never
emerged from the soil samples of our forest, or from
temperate forests with Fagus sylvatica, Quercus
robur and Pinus sylvestris (Godefroid et al. 2006).
Our study suggests that ruderal species in forest soil
seed banks increase with grazing and typical nonweedy forest species decrease.
264
The potential of seed banks to restore communities
is rather limited when many species of the community
are either absent from the persistent seed bank or are
not even able to create any seed bank at all (Handlova
and Münzbergova 2006). The use of seed bank as a
tool for restoration depends strongly on which taxa
retain seeds able to recruit in degraded environments.
The results presented in this study have implications
on the restoration of heavily disturbed forests and are
useful for the conservation management of overgrazed sub-Mediterranean forest types.
Acknowledgements We would like to thank ‘Bourazani
Environmental Park Enterprise’ for kindly hosting Evgenia
Chaideftou during the research periods. Special thanks are also
due to Dr. Thomas Raus for his support on the seedlings
identification and confirmation. Funding by the International
Bureau of the BMBF (GRC 01/007) and the Hellenic General
Secretariat for Research and Technology in the framework of
the Greek-German joint Research and Technology Programme
is gratefully acknowledged. Thanks are also due to Sandy
Coles (M.Sc.) for linguistic revision of the manuscript.
References
Alrababah MA, Alhamad MA, Suwaileh A, Al-Gharaibeh M
(2007) Biodiversity of semi-arid Mediterranean grasslands: impact of grazing and afforestation. Appl Veg
Sci 10(2):257–264. doi:10.1658/1402-2001(2007)10[257:
BOSMGI]2.0.CO;2
Bakker JP, de Vries Y (1992) Germination and early establishment of lower salt-marsh species in grazed and mown
salt marsh. J Veg Sci 3:247–252. doi:10.2307/3235686
Bakker JP, Ruyter JC (1981) Effects of five years of grazing on
a salt-marsh vegetation—a study with sequential mapping. Vegetatio 44(2):81–100. doi:10.1007/BF00133005
Bekker RM, Bakker JP, Kalamees R et al (1998) Seed size
shape and vertical distribution in the soil: indicators of
longevity. Funct Ecol 12:834–842. doi:10.1046/j.13652435.1998.00252.x
Bergmeier E, Dimopoulos P (2008) Identifying plant communities of thermophilous deciduous forest in Greece:
species composition, distribution, ecology and syntaxonomy. Plant Biosyst 142(2):228–254. doi:10.1080/112635
00802150357
Bossuyt B, Hermy M (2001) Influence of land use history on
seed banks in European temperate forest ecosystems: a
review. Ecography 24(2):225–238. doi:10.1034/j.16000587.2001.240213.x
Bossuyt B, Heyn M, Hermy M (2002) Seed bank and vegetation composition of forest stands of varying age in central
Belgium: consequences for regeneration of ancient
forest vegetation. Plant Ecol 162:33–48. doi:10.1023/A:
1020391430072
Brown AHF, Oosterhuis L (1981) The role of buried seed in
coppice woods. Biol Conserv 21:19–38. doi:10.1016/00063207(81)90066-5
A.G. Van der Valk (ed.)
Buckley GP, Howell R, Anderson MA (1997) Vegetation succession following ride edge management in lowland plantations and woods 2 The seed bank resource. Biol Conserv
82:305–316. doi:10.1016/S0006-3207(97)00025-6
Chaneton EJ, Perelman SB, Omacini M, Leon RJC (2002)
Grazing, environmental heterogeneity, and alien plant
invasions in temperate Pampa grasslands. Biol Invasions
4(1–2):7–24. doi:10.1023/A:1020536728448
Colwell RK, Coddington JA (1994) Estimating terrestrial
biodiversity through extrapolation. Philos Trans R Soc
345:101–118. doi:10.1098/rstb.1994.0091
De’ath G, Fabricius KE (2000) Classification and regression
trees: a powerful yet simple technique for ecological data
analysis. Ecology 81(11):3178–3192
Falinska K (1999) Seed bank dynamics in abandoned meadows
during a 20-year period in Bialowieza National Park. J
Ecol 87:461–475. doi:10.1046/j.1365-2745.1999.00364.x
Forrester JA, Leopold DJ (2006) Extant and potential vegetation of an old-growth maritime Ilex opaca forest. Plant
Ecol 183:349–359. doi:10.1007/s11258-005-9045-8
Godefroid S, Phartyal SS, Koedam N (2006) Depth distribution
and composition of seed banks under different tree layers
in a managed temperate forest ecosystem. Acta Oecol
29(3):283–292. doi:10.1016/j.actao.2005.11.005
Grandström A (1987) Seed viability of fourteen species during
five years of storage in a forest soil. J Ecol 75:321–331.
doi:10.2307/2260421
Groot Buinderink GWTA, Hazebroek E (1996) Wild boar (Sus
scrofa scrofa L.) rooting and forest regeneration of podzolic soils in the Netherlands. For Ecol Manag 88:71–80.
doi:10.1016/S0378-1127(96)03811-X
Gross KL (1990) A comparison of methods for estimating seed
numbers in the soil. J Ecol 78(4):1079–1093. doi:10.2307/
2260953
Handlova V, Münzbergova Z (2006) Seed banks of managed and
degraded grasslands in the Krkonose Mts., Czech Republic.
Folia Geobot 41:275–288. doi:10.1007/BF02904942
Haretche F, Rodriguez C (2006) Soil seed bank of an Uruguayan grassland under different grazing practices. Ecol
Austral 16(2):105–113
Harrison S (1999) Native and alien species diversity at the
local and regional scales in a grazed Californian grassland. Oecologia 121:99–106. doi:10.1007/s004420050910
Heinken T, Schmidt M, von Oheimb G (2006) Soil seed banks
near rubbing trees indicate dispersal of plant species into
forests by wild boar. Basic Appl Ecol 7(1):31–44. doi:
10.1016/j.baae.2005.04.006
Hutchings MJ, Booth KD (1996) Studies on the feasibility of
re-creating chalk-grassland vegetation on ex-arable land.
I. The potential roles of the seed bank and the seed rain. J
Appl Ecol 33(5):1171–1181. doi:10.2307/2404696
Jedrzejewska B, Jedrzejewski W, Bunevich AN, Milkowski L,
Krsinski ZA (1997) Factors shaping population densities
and increase rates of ungulates in Bialowieza Primeval
Forest (Poland and Belarus) in the 19th and 20th centuries.
Acta Theriol (Warsz) 42(4):399–451
Jutila H (1998) Seed banks of grazed and ungrazed Baltic
seashore meadows. J Veg Sci 9:395–408. doi:10.2307/
3237104
Kallimanis AS, Kunin WE, Halley JM, Sgardelis SP (2005)
Metapopulation extinction risk under spatially
Forest Ecology
autocorrelated disturbance. Conserv Biol 19(2):534–546.
doi:10.1111/j.1523-1739.2005.00418.x
Kallimanis AS, Ragia V, Sgardelis SP, Pantis JD (2007) Using
regression trees to predict alpha diversity based upon
geographical and habitat characteristics. Biodivers Conserv 16:3863–3876. doi:10.1007/s10531-007-9186-2
Kallimanis AS, Halley JM, Vokou D, Sgardelis SP (2008) The
scale of analysis determines the spatial pattern of woody
species diversity in the Mediterranean environment. Plant
Ecol 196:143–151. doi:10.1007/s11258-007-9341-6
Kent M, Coker P (1994) Vegetation description and analysis. A
practical approach. Wiley, Chichester
Kjellsson G (1992) Seed banks in Danish deciduous forests:
species composition, seed influx and distribution pattern
in soil. Ecography 15:86–100. doi:10.1111/j.1600-0587.
1992.tb00012.x
Koleff P, Gaston KJ, Lennon JJ (2003) Measuring beta
diversity for presence-absence data. J Anim Ecol 72:367–
382. doi:10.1046/j.1365-2656.2003.00710.x
Kuiters AT, Slim PA (2002) Regeneration of mixed deciduous
forest in a Dutch forest-heathland, following a reduction
of ungulate densities. Biol Conserv 105:65–74. doi:
10.1016/S0006-3207(01)00204-X
Le Houérou HN (1990) Agroforestry and sylvopastoralism to
combat land degradation in the Mediterranean Basin: old
approaches to new problems. Agric Ecosyst Environ
33:99–109. doi:10.1016/0167-8809(90)90236-7
Lennon JJ, Koleff P, Greenwood JJD, Gaston KJ (2001) The
geographical structure of British bird distributions:
diversity, spatial turnover and scale. J Anim Ecol 70:966–
979. doi:10.1046/j.0021-8790.2001.00563.x
Looney PB, Gibson DJ (1995) The relationship between the
soil seed bank and above-ground vegetation of a coastal
barrier island. J Veg Sci 6:825–836. doi:10.2307/3236396
Magurran AE (2004) Measuring ecological diversity. Blackwell, Oxford
Malo JE, Jiménez B, Suarez F (2000) Herbivore dunging and
endozoochorous seed deposition in a Mediterranean dehesa.
J Range Manag 53(3):322–328. doi:10.2307/4003440
Marage D, Rameau JC, Garraud L (2006) Soil seed banks and
vegetation succession in the Southern Alps: effects of
historical and ecological factors. Can J Bot 84(1):99–111.
doi:10.1139/b05-142
Miller PM (1999) Effects of deforestation on seed banks in a
tropical deciduous forest of western Mexico. J Trop Ecol
15:179–188. doi:10.1017/S0266467499000747
Naaf T, Wulf M (2007) Effects of gap size, light and herbivory
on the herb layer vegetation in European beech forest
gaps. For Ecol Manag 244:141–149. doi:10.1016/j.foreco.
2007.04.020
Oke SO, Oladipo OT, Isichei AO (2006) Seed bank dynamics
and regeneration in a secondary lowland rainforest in
Nigeria. Int J Bot 2(4):363–371
Olano JM, Caballero I, Laskurain NA (2002) Seed bank spatial
pattern in a temperate secondary forest. J Veg Sci
13(6):775–784. doi:10.1658/1100-9233(2002)013[0775:
SBSPIA]2.0.CO;2
Osem Y, Perevolotsky A, Kigel J (2006) Similarity between seed
bank and vegetation in a semi-arid annual community: the
role of productivity and grazing. J Veg Sci 17:29–36. doi:
10.1658/1100-9233(2006)017[0029:SBSBAV]2.0.CO;2
265
Peco B, Ortega M, Levassor C (1998) Similarity between seed
bank and vegetation in Mediterranean grassland: a predictive model. J Veg Sci 9:815–821. doi:10.2307/3237047
Pépin D, Renaud P-C, Boscardin Y, Goulard M, Mallet C,
Anglard F, Ballon P (2006) Relative impact of browsing
by red deer on mixed coniferous and broad-leaved seedlings—an exclosure-based experiment. For Ecol Manag
222:302–313. doi:10.1016/j.foreco.2005.10.034
Ramos JA, Bugalho MN, Cortez P, Iason GR (2006) Selection
of trees for rubbing by red and roe deer in forest plantations. For Ecol Manag 222:39–45. doi:10.1016/j.foreco.
2005.10.017
Rice KJ (1989) Impacts of seed banks on grassland community
structure and population dynamics. In: Leck MA, Parker
VT, Simpson RL (eds) Ecology of soil seed banks. Academic Press, San Diego, pp 212–230
Robson BJ, Clay CJ (2005) Local and regional macroinvertebrate
diversity in the wetlands of a cleared agricultural landscape
in south-western Victoria, Australia. Aquat Conserv: Mar
Freshw Ecosyst 15(4):403–414. doi:10.1002/aqc.675
Roovers P, Bossuyt B, Igodt B, Hermy M (2006) May seed
banks contribute to vegetation restoration on paths in
temperate deciduous forest? Plant Ecol 187:25–38. doi:
10.1007/s11258-006-9130-7
Shen Y, Liu W, Cao M, Li Y (2007) Seasonal variation in density
and species richness of soil seed banks in karst forests and
degraded vegetation in central Yunnan, SW China. Seed
Sci Res 17:99–107. doi:10.1017/S0960258507708139
ter Braak CJF, Šmilauer P (2002) CANOCO reference manual
and CanoDraw for Windows user’s guide: software for
canonical community ordination (version 4.5). Microcomputer Power, Ithaca
Thompson K, Grime JP (1979) Seasonal variation in the seed
banks of herbaceous species in ten contrasting habitats.
J Ecol 67:893–921. doi:10.2307/2259220
Thompson K, Bakker J, Bekker R (1997) The soil seed banks
of North West Europe: methodology, density and longevity. Cambridge University Press, Cambridge
Tsaliki M, Bergmeier E, Dimopoulos P (2005) Vegetation
patterns and plant diversity in mixed oak woodlands in the
region of Bourazani, Epirus (NW Greece). Bot Chron
18:209–235
Ungar IA, Woodell SRJ (1996) Similarity of seed banks to
aboveground vegetation in grazed and ungrazed salt marsh
communities on the Gower Peninsula, South Wales. Int
J Plant Sci 157(6):746–749. doi:10.1086/297398
Warr SJ, Kent M, Thompson K (1994) Seed bank composition
and variability in five woodlands in south-west England.
J Biogeogr 21:151–168. doi:10.2307/2845469
Wellstein C, Otte A, Waldhardt R (2007) Seed bank diversity
in mesic grasslands in relation to vegetation type management and site conditions. J Veg Sci 18:153–162. doi:
10.1658/1100-9233(2007)18[153:SBDIMG]2.0.CO;2
Witten IH, Frank E (2005) Data mining. Practical machine
learning tools and techniques, 2nd edn. Morgan Kaufmann, San Francisco
Zhang W (1998) Changes in species diversity and canopy cover
in steppe vegetation in Inner Mongolia under protection
from grazing. Biodivers Conserv 7(10):1365–1381. doi:
10.1023/A:1008852017493
On the detection of dynamic responses
in a drought-perturbed tropical rainforest in Borneo
M. Lingenfelder Æ D. M. Newbery
Originally published in the journal Plant Ecology, Volume 201, No. 1, 267–290.
DOI: 10.1007/s11258-008-9568-x Springer Science+Business Media B.V. 2009
Abstract The dynamics of aseasonal lowland dipterocarp forest in Borneo is influenced by perturbation
from droughts. These events might be increasing in
frequency and intensity in the future. This paper
describes drought-affected dynamics between 1986
and 2001 in Sabah, Malaysia, and considers how it is
possible, reliably and accurately, to measure both
coarse- and fine-scale responses of the forest. Some
fundamental concerns about methodology and data
analysis emerge. In two plots forming 8 ha, mortality,
recruitment, and stem growth rates of trees C10 cm
gbh (girth at breast height) were measured in a ‘predrought’ period (1986–1996), and in a period (1996–
2001) including the 1997–1998 ENSO-drought. For
2.56 ha of subplots, mortality and growth rates of
small trees (10–\50 cm gbh) were found also for two
sub-periods (1996–1999, 1999–2001). A total of
c. 19 K trees were recorded. Mortality rate increased
by 25% while both recruitment and relative growth
rates increased by 12% for all trees at the coarse scale.
For small trees, at the fine scale, mortality increased
by 6% and 9% from pre-drought to drought and on to
Electronic supplementary material The online version of
this article (doi:10.1007/978-90-481-2795-5_21) contains
supplementary material, which is available to authorized users.
M. Lingenfelder D. M. Newbery (&)
Vegetation Ecology Section, Institute of Plant Sciences,
University of Bern, Altenbergrain 21, 3013 Bern,
Switzerland
e-mail: david.newbery@ips.unibe.ch
‘post-drought’ sub-periods. Relative growth rates
correspondingly decreased by 38% and increased by
98%. Tree size and topography interacted in a
complex manner with between-plot differences. The
forest appears to have been sustained by off-setting
elevated tree mortality by highly resilient stem
growth. This last is seen as the key integrating tree
variable which links the external driver (drought
causing water stress) and population dynamics
recorded as mortality and recruitment. Suitably sound
measurements of stem girth, leading to valid growth
rates, are needed to understand and model tree
dynamic responses to perturbations. The proportion
of sound data, however, is in part determined by the
drought itself.
Keywords Dynamics Perturbation Drought
Stem growth Tree mortality Validity
Introduction
Stochastic fluctuations in the environment are thought
to play an important role in driving the long-term
dynamics of tropical rain forests and in determining
their structure and species composition (Newbery and
Lingenfelder 2004, 2009). Droughts, fires, floods and
hurricanes are notably examples of such external
climatic influences. In South-East Asia, under normally aseasonal climatic conditions, it is droughts
that probably have the most sustained and repeated
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_21
267
268
effects on the forests. These periods of reduced
precipitation are often associated with the El NiñoSouthern Oscillation (ENSO) cycle (Walsh 1996;
Walsh and Newbery 1999), a global process with its
origins in changes in sea-surface temperature and air
pressure in the Pacific Ocean (Trenberth 1997;
Trenberth and Hoar 1997).
The ENSO reaches back 130,000 years and events
are known to have affected Borneo for at least
18,000 years (Walsh and Newbery 1999; Cane 2005).
They result in occasional moderate droughts which
are an integral component of the environment (Walsh
and Newbery 1999). Away from the eastern coast of
Borneo, at the inland site of Danum (in Sabah,
Malaysia) for instance, the tree species of the lowland
dipterocarp forests appear well adapted to the correspondingly moderate perturbations to the ecosystem
that the droughts cause (Newbery et al. 1999;
Gibbons and Newbery 2003; Newbery and Lingenfelder 2004, 2009).
Compared to the preceding century, the last
30 years have shown an increase in the frequency
and intensity of ENSO events (Trenberth et al. 2007).
The trend had been expected by the earlier analyses
of Hulme and Viner (1998), Timmermann et al.
(1999, 2004) and IPCC (2001). Other recent models,
however, lend less support to an increase continuing
in the near future (Cane 2005; McPhaden et al. 2006;
Meehl et al. 2007), even though when ENSO events
do occur they may lead to a higher risk of strong
drought (Christensen et al. 2007). Given that the
prognoses are weak it remains important to be
prepared for either an increase in droughts or a
stabilization of the earlier pattern because it certainly
will have profound implications on how the forests
should be best conserved and managed.
All tropical rain forests can be viewed as being
continuously in various complex and overlapping
states of recovery from past perturbations, whether
these are singular or closely timed multiple events,
happening recently or in the more distant past
(Newbery et al. 1999; Newbery and Lingenfelder
2004). Measuring precisely how the forests respond
to currently occurring perturbations may lead nearer
to reliable models which can estimate how increases
and decreases in frequency and/or intensity of
perturbation might affect their persistence.
After the last strong ENSO-related drought in
1997/1998 several studies have been conducted on
A.G. Van der Valk (ed.)
the ecological effects of such short, but significant
periods of drought stress on trees in South-East Asia
(Nakagawa et al. 2000; Harrison 2001; Potts 2003;
Ichie et al. 2004; Newbery and Lingenfelder 2004;
Slik 2004). They variously concluded that some
forest’s species were well adapted to a moderate
drought regime whereas for those of other forests this
was not so evident. The focus was also mainly on
recording mortality—growth and recruitment receiving less attention—and drought was simply defined to
take effect when the 30-day running total (30-d-rt) of
rainfall fell \100 mm. Plots and tree sample sizes
were sometimes quite small, and the area measured
may not have been fully representative of local
topographic variation.
Drought is a stochastic factor and frequency and
intensity of its complex effects requires careful
consideration. Soil water status before and during
an event, as well as the replacement of depleted soil
water after it, need to be taken in to account using a
dynamic approach based, for example, on the idea of
‘antecedent rainfall history’ proposed by Newbery
and Lingenfelder (2009). In addition, since climatic
variation is occurring at the scale of decades, and not
annually or per century, only long-term measurements over 20–30 years that capture forest dynamics
before and after a drought for several years (at
minimum close to the return time of the event) are
likely to provide enough ecologically meaningful
information. To have such records for repeated
droughts at the same and other replicated locations
would of course be ideal.
From previous work in Borneo, it was concluded
that the forest at Danum is most likely still recovering
from a very strong drought c. 130 years ago and that
it is well adjusted to coping with repeated lesser
droughts that have happened since then. Seen at the
scale of many centuries, the forest is perhaps in a
state of dynamics equilibrium, with variously large
and small ‘set backs’ occurring at different points in
time (Newbery et al. 1992, 1996, 1999). Behind this
process lies the dynamics of the individual species
and how they are adapted to the actual physiological
effects of the drought perturbations (Newbery and
Lingenfelder 2004, 2009). Their responses will
collectively determine the resilience of the ecosystem, i.e., how fast and to what extent it can recover
after perturbation. In this context the pressing question, for both natural and secondary managed forests,
Forest Ecology
is whether species selected under the past environmental history are able to survive as well as before
when droughts come more intensely and closer
together in the future.
Species-specific tree responses to the moderate
perturbation regime at Danum have been presented
recently by Newbery and Lingenfelder (2009). The
present paper describes the structure of the permanent
research plots at Danum and analyses tree dynamics
between 1986 and 2001 in detail at the plot and
subplot levels. The focus here is on the response of
the whole forest to drought, particularly to the 1997/
1998 ENSO-related event, in terms of mortality,
recruitment and growth rates. Several field methodological and data analytical problems are tackled.
These have wider relevance to tropical forest dynamics in general, and highlight some important
limitations to conclusions that can be drawn from
recensussing studies. The aims of the present study
were thus: (1) to quantify the effect of the 1997/1998
main drought on forest dynamics at Danum, (2) to
investigate the interactions between tree size and
topography on dynamics, (3) to refine the treatment
of stem growth estimates for aims 1 and 2, and (4) to
place the dynamics responses into the frame of the
forest ecosystem.
Methods
Study site
Location
The study site lies within the 438-km2 Danum Valley
Conservation Area (DVCA), Sabah, Malaysia, 66 km
inland of Lahad Datu on the N.-E. coast of Borneo.
The DVCA is an uninhabited and unlogged part of
the 9730-km2 Yayasan Sabah Concession Area;
human artefacts suggest that there might have been
some earlier settlement or visitation (Marsh and
Greer 1992). The vegetation around the site is
primary lowland dipterocarp forest of the Parashorea
malaanonan category (Fox 1972). The topography is
gently undulating, and the soils are mainly orthic
acrisols of the Bang association which developed on
sandstone and mudstone of the geological KuamutFormation (Wright 1975). Further details of the site
are given in Newbery et al. (1992, 1996, 1999).
269
Climate
The climate at Danum Valley Field Centre (DVFC,
4570 4800 N, 117480 1000 E, 152 m a.s.l.) is typical of
equatorial rainforest locations (Walsh and Newbery
1999) with the mean daily range of temperature
(8.6C) being larger than the monthly mean range
(1.8C) about an annual mean temperature of 26.8C,
high relative humidity and high annual rainfall (mean
c. 2800 mm). There is no clear dry season indicating
that Danum has a generally aseasonal tropical
climate. Further details can be found in Newbery
and Lingenfelder (2009).
Between 1985 and 2003, Danum experienced 19
low precipitation events of which one was ecologically severe (event centred in 1998), two moderately
strong (1987 and 1992) and five of weaker intensity,
as shown by antecedent rainfall history analysis
(Newbery and Lingenfelder 2009). Severe droughts
across large parts of Borneo that were probably
stronger than the one in 1997/1998 were recorded in
1877/1878, 1914/1915 and just before the setting up
of the main plots at Danum in 1982/1983 (Beaman
et al. 1985; Walsh 1996; Walsh and Newbery 1999)
and most likely affected them. The events of 1877/
1878, 1982/1983, and 1997/1998 were the three
strongest El Niño-events in terms of sea-surface
temperature anomalies in the ‘ENSO 3’-region since
1876, where reliable reconstructions can be made
(IPCC 2001). Since that time droughts of weak to
moderate intensity have occurred frequently across
Sabah (3.25 times per 20 years) and the frequency of
strong droughts at Danum was 1.54 per 20 years on
average (Walsh 1996; Walsh and Newbery 1999).
Design
In 1985–1986, two permanent plots were first set up
and enumerated (Newbery et al. 1992, 1996). They are
located c. 0.8 km NW of DVFC, just north of Main
Trail West on gently undulating terrain with elevations of 208–254 m a.s.l. Plot 2 lies c. 280 m parallel
to, and west of, plot 1. The plots are rectangular in
shape (each 100 m 9 400 m, area = 4 ha) with the
longer sides oriented north-south. Each was divided
into 100 units of 20 m 9 20 m (small subplots) and
their corners marked with belian (ironwood) posts.
Relative differences in elevation and slopes within
the plots are very similar (39–43 m; Fig. 1). Plot 1
270
includes steep east-facing slopes in its northern half: in
plot 2 an episodic small stream cuts two small ridges
with steep slopes. Interpolating from 20-m 9
20-m-grid elevational data, ridge areas (C25 m,
relative to SW-plot corners of 0 m) covered 18%
and 33%, and lower slope areas (\12 m) 32% and
36% of the planimetric surfaces of plots 1 and 2,
respectively. Of plot 1, 31% is flat (\10 inclination)
and 14% is steep (C20; max. 33). In plot 2, the
corresponding values are 26% and 18% (max. 31).
More than half of the area of each plot lies on
intermediate slopes (10–20).
Enumerations
Previous to 2001: background
The first enumeration of the plots was between 24
August 1985 and 15 December 1986 (median 15
Fig. 1 Topographic
variation within the two 4ha permanent plots in the
Danum Valley
Conservation Area, Sabah,
Borneo
A.G. Van der Valk (ed.)
March 1986; Fig. 2). Within each subplot, every
living tree with a minimum stem girth at breast height
(gbh) of 10 cm (C3.2 cm dbh) was mapped, tagged,
(and identified) and its gbh measured at a painted
mark, usually 1.3 m above ground (see Newbery
et al. 1992, for details). The second enumeration was
between 8 November 1995 and 23 February 1997
(median 18 June 1996; Fig. 2), in which all trees were
recorded for alive/dead status and the gbh of survivors remeasured. Surviving trees that had grown to
C10 cm gbh (recruits) were mapped, tagged, identified, and measured (see Newbery et al. 1999). The
first and second enumerations each lasted 1.3 years.
The mean time interval for the two plots was
10.0 years.
Eight 40-m 9 40-m (large) subplots in each main
plot, half of them on lower slopes and the other half
on ridges, had been measured between 20 December
1998 and 29 March 1999 (median 25 January 1999;
Forest Ecology
271
Fig. 2 The sampling scheme at Danum showing the dates of
the three full and one partial enumeration, the timing of the
1998 ENSO drought and the corresponding periods (P1, P2)
and subperiods (P2a, P2b), and the extension of P1 (P1x) and
the estimated immediate post-drought sub-period (P2x)
Fig. 2). Each consisted of four small subplots in a
square, as defined in Newbery and Lingenfelder
(2004). Subplots were selected in a stratified random
manner and represented 2.56 ha (32%) of the main
plot area. In 1999, all 1996-recorded trees C10 cm
gbh were scored for alive/dead status and surviving
small trees (10 – \50 cm gbh) remeasured: recruits
and regressors were not registered in 1999.
buttresses were already influencing stem growth, or
was seen likely to do so in the next 5 years. In the
latter case, the PoM was moved to at least 1 m above
the buttress. This strategy of adding alternative PoMs
was started in 1996 to ensure that at least in two
consecutive enumerations the tree was measured at
the same PoM. Alternative PoMs established in 1996
were utilized in 92 instances in 2001 (Lingenfelder
2005). PoM (original, new, etc.) and condition of the
stem (CoS; e.g., stem normal or deformed) at that
point were recorded in six and 16 classes, respectively (Supplementary materials—Appendix 1).
Height of the PoM (if not at 1.3 m) was also noted.
Girth (gbh, to nearest mm) was measured with a
thin 2-m steel tape, and for larger trees a wider 5-m
one, after lightly cleaning the bark at the PoM. Trees
with multiple stems C5 cm gbh, and of which one
was C10 cm gbh, were included and a single
hypothetical gbh-value found from their combined
basal areas. When it was impossible to insert the tape
under a constricting liana, callipers were used to
measure tree diameter, taking two readings at 90 to
one another. For trees where the PoM had to be
moved upwards to [ c. 2 m, a ladder was used. For
trees with PoMs at c. [4 m (n = 48 trees), stem
diameter was measured optically with a laser ranging
instrument (Criterion 400, Laser Technology Inc.,
Centennial, USA), again with two readings taken at
90 apart. Method of measurement was recorded in
five classes (Supplementary materials—Appendix 1).
Recruits surviving the interval since the second
enumeration, were mapped, tagged, painted, (identified) and measured.
In 2001: advancements
The third full enumeration of the plots was conducted
between 26 February 2001 and 4 February 2002
(median 29 June 2001, Fig. 2), taking nearly 1 year
(109 field days). The mean time interval from the
second enumeration was 5.0 years (Lingenfelder
2005). Dead trees were recorded in five status classes
(e.g., dead standing or dead-broken: missing stems
were assumed to be dead. Status of survivors was
recorded in seven classes, e.g., undamaged or broken
(see Supplementary materials—Appendix 1).
Stems of surviving trees were inspected at the
paint-mark of the previous point of measurement
(PoM). If the paint-mark was lost, a new one was
established at 1.3 m on the uphill side of the tree or at
the nearest suitable point on the stem avoiding stem
deformations and obstructions. The same procedure
was followed if a stem was broken below the old
PoM and a new shoot had to be measured. Unless
deformation was too great to allow remeasurement,
gbh at the old PoM was measured and an alternative
PoM was established at the nearest suitable point, and
measured. An alternative PoM was also established if
272
To make use of the additional 1999 partial enumeration of the large subplots, trees of the same size class
and subplots in the 1986, 1996 and now 2001—
enumerations were selected. Because no recruits were
recorded in 1999, n99 was lowered, and to have used
this value mortality rates in sub-period 2b would have
been overestimated. Accordingly, recruits in 1996 and
2001 were also excluded from the subplot data set.
Individual trees were allocated to the three topographic
classes as defined for the main plots; numbers of trees
in the intermediate locations were c. half those on the
ridges and lower slopes.
Calculations
The three sets of measurements are referred to as the
‘1986-’, ‘1996-’, ‘1999-’ and ‘2001-enumerations’;
the resulting time intervals as ‘period 1’ (1986–1996)
and ‘period 2’ (1996–2001). Period 2 divided into
two sub-periods: 2a (1996–1999, 2.6 years) and 2b
(1999–2001, 2.4 years) when including the 1999enumeration on the subplot level (Fig. 2). These
periods and subperiods might be thought of as ‘predrought’, ‘drought’ and post-drought’, except that
period 1 was not free of any droughts (Newbery and
Lingenfelder 2004), the length of the sub-period
encompassing the 1997–1998 major event is arbitary,
and post-drought effects did begin well before 1999.
Analysis within different sizes was performed for
trees with the following gbh limits: all, C10 cm gbh
(C3.2 cm dbh); small, 10–\50 cm gbh (3.2–
\15.9 cm dbh); medium, 50–\100 cm gbh (15.9–
\31.8 cm dbh); and large, C100 cm gbh (C31.8 cm
dbh). To allow comparison with some other studies,
measures were also found for the population of trees
with a dbh of C10.0 cm (C31.4 cm gbh).
Some trees above the minimum gbh-limit at first
measurement were (due to natural shrinking, bark
loss, slight measurement errors, or because multiplestemmed trees lost one or more of their stems) too
small at the second enumeration, and not being part
of the population they were labelled ‘regressors’.
Between the second and third enumerations regressors either died, remained with gbh less that the
minimum value, or regrew above that value. In the
last case, a regressor was not viewed as a new recruit
because it was a member of the population of trees
C10 cm gbh at an earlier enumeration (original tag
number used). This problem of trees regressing below
A.G. Van der Valk (ed.)
the minimum size and potentially re-growing above
that limit in a subsequent enumeration is addressed in
Supplementary materials—Appendix 2 (‘Losses and
gains’).
Basic dynamic rates
Periodic (mp, rp; %) and annualized (ma, ra; % year-1)
rates were found for mortality and recruitment,
respectively, after Alder (1995) and Sheil et al.
(1995), on the plot or subplot level for different size
classes, using the mean time intervals of each individual group (see Supplementary materials—Appendix 3
for formulae). Confidence limits (95%) of ma and ra
were estimated with an approximation based on the Fdistribution. Correction of ma for the differences in
length of time interval (5-year basis) followed the
method of Sheil and May (1996), as applied to the
Danum data set in Newbery and Lingenfelder (2004).
Absolute (agr; mm year-1) and relative (rgr;
mm m-1 year-1) stem growth rates were similarly
found (Supplementary materials—Appendix 3).
Growth rate calculations were based here on
intervals of each individual tree. As frequency
distributions of rgr values were always very strongly
positively skewed, and no transformation could
normalize, or a suitable probability density function
be found as yet to model them, a bootstrapping
procedure (N = 2000 runs) was used to find means
and 95% confidence limits of these variables. Comparisons on this basis are to be made within each
period separately. The database was handled in
Microsoft Access and statistical analyses performed
with GenStat versions 7 and 8 (Payne et al. 2007).
Individual growth values were used for two reasons:
(1) interest lay in topographic effects and differences
between tree size classes, which were nested within
plots; and (2) the limits would correspond to those
derived for mortality rates which are de facto withinplot estimates also. Where confidence limits did not
overlap means were judged to be significantly
different (a = 0.05). Growth rates were also found
separately for trees that lived, and those that died, in a
successive period.
Validity of growth rates
For each enumeration (except the one of 1986, when
this information was not gathered) every tree was
Forest Ecology
reviewed for suitability of its girth measurements
with regard to calculating growth rates, and assigned
a code accordingly: 1 = suitable, 0 = unsuitable.
Growth rates were considered valid (i.e., sound) only
if both start and end measurements were suitable.
Measurements were unsuitable where (1) the status
code showed that the tree was broken below, halfbroken or dead at the PoM, or had lost one or more of
multiple stems; (2) the CoS indicated major deformations due to buttress growth, cracked or split bark
or stems, excrescence, fluted or hollow stems,
termites or lianas (an irregular stem (CoS = DI)
was not considered a major deformation unless
additional notes in the remarks revealed this, e.g.,
‘heavy’, ‘extremely oval’, or ‘spiral growth’); (3)
POMs were moved or newly established (except on
recruits), or the laser ranging instrument was used.
The use of callipers on liana-fused trees was only
considered a reason for exclusion where it was not
possible to take two measurements or the callipers
were too small.
From the resulting valid rates, some trees had
additionally to be excluded because they had negative
growth rates below an operational threshold. To
separate those values that resulted from faulty
measurements or recording errors from those that
would very likely be part of the population (e.g., due
to slight shrinkage because of low stem water
content, unapparent loss of bark), the approach
developed by Newbery et al. (1999) was followed
and applied to the 1996–2001 data set: relative
frequencies, expressed as proportions, of all growth
rates with agr B0 mm year-1 were logit-transformed
and plotted in increasingly negative agr-classes. Both
plots separately and combined showed an almost
linear decline to -3.5 mm year-1. Below that value
(i.e., B4.0 mm) the distribution increased slightly,
decreased again and then flattened, indicating that
these values were probably not part of the ‘natural’
population (Lingenfelder 2005).
Spatial autocorrelation
As the growth of trees across an area might not be
statistically independent from each other, the data set
was explored for spatial autocorrelation (SAC). The
analysis is based on mean valid relative growth rates
of 10-m 9 10-m subplots to account for the at-places
rapidly changing topography within the main plots; a
273
20-m 9 20-m subplot could be partly located on a
flat ridge area but steeply sloping down into an
intermediate elevation. Moran’s test for SAC was
calculated (Moran’s I; using moran.test of the
spdep package (Bivand 2007 in R 2.6.1, R Development Core Team 2007) and plotted at 5-m intervals
across distances of 0–100 m across the whole plots as
well as per hectare to investigate stationarity. Anisotropy was checked with a routine in S-Plus, Version
7.0 (Kalunsky et al. 1998). The effects of topography
(elevation and slope) on rgr were investigated with a
spatial conditional autoregression (CAR) model estimation by maximum likelihood (spautolm in
spdep; Bivand 2007) and ordinary regression (R
Development Core Team 2007). The models included
linear, quadratic and cubic terms. Based on a
likelihood ratio test (of the spatial coefficient; within
spautolm) it was decided whether CAR spatial
specification improved the model.
Results
Forest structure
Total numbers of trees in the main plots decreased by
299 and 679 in periods 1 and 2, respectively.
Considering only trees with gbh C10 cm, the corresponding decreases were 677 and 642. The large
difference for period 1 was due to regressors being
excluded and gains included (Table 1). On an annual
basis, tree numbers (gbh C10 cm) in period 2 declined
almost twice as fast (128 stems year-1) as in period 1
(68 stems year-1). Results for the individual plots are
given in Supplementary materials—Appendix 3.
With some slight variability between the plots, tree
density decreased for all and for small trees in the two
periods by 4% (Table 2 and Supplementary materials—
Appendix 3). It increased for medium-sized and large
trees in period 1 (by 5%), yet the density of mediumsized trees decreased in period 2 (by 2%) and did not
change for large trees, the latter due to a 3% decrease
in plot 1 but a 3% increase in plot 2. Recruits and dead
trees had lower densities in period 2 than 1, largely
due to the differing interval lengths. On an annual
basis, the density of recruits increased by 7%, while
the density of dead trees increased by 32%. The
contribution of dead trees that had been regressors in
1996 was 1% of all trees (or 9% of all dead trees) in
274
A.G. Van der Valk (ed.)
Table 1 Numbers of trees at Danum for periods 1 (1986–
1996) and 2 (1996–2001), two main plots combined, from
those at the starts (nstart) to the ends (nend), and showing the
numbers that survived (ns), died (nd), recruited (nr), were
gained (ngains) and lost (nlosses)
Period
1
2
nstart
17942
17643
nstart_C10
17942
17265
2655
1938
Table 2 Densities of trees (n trees ha-1) in the main plots at
Danum in 1986, 1996 and 2001 for three size classes of tree
1986
182
nd_reg_p1
1756
nd_C10
All (C 10 cm gbh)
2243
2158
2078
Small (10 to \50 cm gbh)
2033
1939
1863
150
146
153
Large (C100 cm gbh)
63
66
66
C31.4 (C10 cm dbh)
432
452
435
Recruits
295
157
Dead (C10 cm gbh)
332
220
Dead (regressors 1996)
15287
15705
378
341
nlosses
3033
1931
ns_C10
14909
15364
2356
1259
2356
1289
nend
nend_C10
17643
17265
16964
16623
Coarse-scale dynamics
ndiff_C10
-677
-642
Basic rates
ns
nreg
nr
30
nreg_p1_C10
ngains
The subscript C10 refers to numbers of trees with gbh C10 cm
gbh (For details at the plot level, see Supplementary
materials—Appendix 3)
nd_reg_p1: number of regressors of period 1 that were found
dead in period 2; nreg: number of regressors in period 2:
old(remaining) = 166, new = 175; nreg_p1_C10: number of
regressors of previous period, gbh in 2001 C 10 cm
2001 (Table 2). From 1986 to 2001, density decreased
for all and small trees but increased for medium-sized
and large trees. The ratio of densities in three size
classes within each main plot was close to 90:7:3 for
small, medium and large trees, respectively, over the
three enumerations (Supplementary materials—
Appendix 3).
The 16 subplots, with the restrictions applied, had
5190, 4239, 3885 and 3706 in 1986, 1996, 1999 and
2001, respectively. Small trees represented 91% of all
trees in 1986 and 86% at the other three enumerations. During period 1 and sub-periods 2a and 2b,
741, 237 and 213 small trees, respectively, died, so
that the original population in 1986 lost 23% of its
trees by 2001 through mortality, 6% (293 trees) either
regressing to \10 cm gbh or advancing to C50 cm
gbh. Mean density (2027 ha-1) in 1986 was similar
to that of the main plots (cf. Table 2), but mainly as a
consequence of the missing recruits, these values
2001
Size class
Medium (50 to \100 cm gbh)
nd
1996
23
For details at the plot level, see Supplementary materials—
Appendix 3
steadily declined to 1448 ha-1 by 2001. The mean
number of small trees per subplot was 266 (n = 16,
range, 169–386).
In period 1 (1986–1996) almost 15% of trees died
across both plots and in period 2 (1996–2001) 11%
died (Table 3). Differences between the two replicate
plots were apparent in period 1: in plot 1 the periodic
mortality rate was 4% above that in plot 2 (Supplementary materials—Appendix 3). In period 2 the
difference between plots in mp was much less (\1%).
Annualized mortality was 45% higher in period 2 for
both plots combined (Table 3a). The relative increase
of ma in plot 1 was almost twice that in plot 2 (33 vs.
61%) and ma values in period 2 were correspondingly
more similar than in period 1 (Supplementary materials—
Appendix 3).
Based on a 5-year interval, the correction of mall
(the overall average mortality resulting from the taxaand subplot-wise grouped mortalities) produced the
expected result for period 1: shifting the annual
mortality from 10 to 5 years using the correction
factors (1.115 for plot 1 and 1.220 for plot 2)
calculated from the data set where the rarest species
were excluded (nmin = 2), increased mortality rates
by 13% and 24% in plots 1 and 2, respectively. As the
intervals of period 2 were similar for plot 1
(5.06 years) and plot 2 (4.94 years) and both of these
were very close to an average of 5.0 years, mall did
Forest Ecology
275
Table 3 Forest dynamics of the main plots at Danum for
periods 1 (1986–1996) and 2 (1996–2001), trees C10 gbh:
rates of mortality, recruitment and growth, and the estimated
overall mortality based on species’ rates
Period
1
2
mp (%)
14.80
10.98
ma (% year-1)
rp (%)
1.59
13.13
2.30
7.14
ra (% year-1)
1.24
1.39
agr (mm year-1)
3.05
3.12
11.15
12.48
rgr (mm m-1 year-1)
mall (% year-1)
2.34a
–
1.87
mcorr
a
chigher than that in period 1, for both plots combined.
The changes in ra between the periods were, however,
smaller than for ma and the two plots rather diverged
than converged with time: plot 1 had a 16% higher
recruitment rate in period 2 than 1 whereas in plot 2 it
increased by just 7%. Thus plot 1 increased its
prominence in regard to recruitment rate. Absolute
(agr) and relative (rgr) growth rates were 9–17%
higher in plot 1 than plot 2 (Supplementary materials—
Appendix 3). Between periods 1 and 2 agr increased by
2.3% (from 3.05 to 3.12 mm year-1) and rgr by 11.9%
(from 11.2 to 12.5 mm m-1 year-1; plots combined).
The frequency distributions of agr and rgr were
nevertheless strongly positively skewed.
–
For details at the plot level, see Supplementary materials—
Appendix 3
Effects of tree size and topography
mp, ma: periodic and annual mortality (all trees, including
regressors: nd/nstart); rp, ra: periodic and annual recruitment;
agr, rgr: absolute, and relative, growth rate in stem girth; mall:
overall average mortality with species within subplots as
groups; mcorr: mall corrected to 5-year basis (no correction for
period 2)
In periods 1 and 2 ma was higher on intermediate
positions and lower slopes than on ridges (Table 4),
and did not differ greatly between size classes within
topographic classes (Fig. 3). Considering the individual plots, however, the ma of medium-sized trees
in plot 1 was approximately double that in plot 2, a
much larger difference than in the other size classes
(Supplementary materials—Appendix 4). In period 2,
ma increased with size for all topographic classes
combined (Fig. 3d), a reflection especially of the
large (1.6-fold) difference between small and large
trees in plot 1 (Supplementary materials—Appendix
4). This increasing trend with size was most clearly
shown on the ridges (Fig. 3a), while on the lower
slopes ma was highest among the medium-sized trees
(Fig. 3c), and intermediate positions had a complex
pattern in between (Fig. 3b). In the small, medium
and large size classes, ma was overall 31, 42 and
94%, respectively, higher in period 2 than 1. The
increase in ma across periods was strongest for
a
Means of plot 1 and 2 values
not need an interval correction. On this basis of the
foregoing considerations, mortality increased by 20%
in plot 1 and by 31% in plot 2 between periods 1 and
2. The mortality rates for both plots combined were
thus 1.87 and 2.34% year-1 in periods 1 and 2,
respectively, implying an increase by 25% between
the periods. Rates for trees C10 cm dbh are given in
Supplementary materials—Appendix 3 also.
Periodic recruitment rate in period 2 was just over
half of that in period 1, as expected from the differing
time intervals (Table 3). Recruitment was lower than
mortality in both periods and both plots. The
annualized recruitment rate in period 2 was 12%
Table 4 Comparison of annualized mortality (ma, % year-1), recruitment (ra, % year-1) and relative stem growth rates (rgr,
mm m-1 year-1) in plots 1 and 2 combined at Danum in three topographic classes for periods 1 (1986–1996) and 2 (1996–2001)
Topographic
class
Period:
ma
Ridge
1.34 [1.25–1.43] 1.93 [1.78–2.09] 1.19 [1.12–1.26] 1.31 [1.19–1.43] 10.52 [10.10–11.00] 12.71 [12.22–13.30]
1
ra
2
1
rgr
2
1
2
Intermediate
1.63 [1.55–1.71] 2.19 [2.06–2.33] 1.26 [1.20–1.32] 1.27 [1.17–1.36] 11.43 [11.04–11.83] 11.99 [11.59–12.49]
Lower slope
1.75 [1.66–1.85] 2.19 [2.05–2.34] 1.27 [1.20–1.33] 1.67 [1.56–1.78] 11.32 [10.90–11.78] 12.88 [12.36–13.42]
Numbers in square brackets are the 95% confidence limits (bootstrapped in case of rgr). Numbers of trees at the starts of the intervals
(ma) or numbers of valid trees (rgr) are found in Supplementary materials—Appendix 4
276
A.G. Van der Valk (ed.)
medium-sized trees in plot 2 with a 2.25-fold increase
(Supplementary materials—Appendix 4). Again, it
was the intermediate position (averages over size
classes) that showed the largest differences between
plots (plot 2 almost 40% higher than plot 1).
Considering interactions between size class, topography and plot, the most marked changes were the
increase in ma of large trees on ridges plot 1 (period
2 [ 5-fold period 1), and the amelioration for
medium-sized trees in the intermediate position in
plot 2. Small trees were in general much less affected.
Between plot differences were important.
Recruitment was also lowest on ridges although
differences between locations were smaller than for
mortality (Table 4). Furthermore, ra hardly changed
on intermediate locations between periods, increased
moderately (10%) on ridges but rather strongly (32%)
on lower slopes (Table 4). Relative growth rates were
higher in period 2 than 1 by 11, 21 and 18% in small,
medium and large size classes, respectively (Table 4,
Fig. 4d). Except for medium-sized trees in period 1,
growth rates were higher in plot 1 than 2 in both
periods, especially strongly for the small trees
(Supplementary materials—Appendix 4). Relative
growth rates generally decreased with increasing size
class in both plots and both periods, more pronounced
on ridges and intermediate positions (Fig. 4a, b) than
on lower slopes in period 1 (Fig. 4c), while in period
2 the trend was broken by medium-sized trees on the
intermediate positions performing marginally better
(a) ridges
(b) intermediate positions
(c) lower slopes
(d) all topographic positions
4.0
-1
m a (% year )
3.0
2.0
1.0
0.0
4.0
-1
ma (% year )
3.0
2.0
1.0
0.0
small
medium
large
size class
Fig. 3 Mortality rates within size and topographic classes in
the main plots at Danum: ma (% year-1) for period 1 (open
bars) and period 2 (grey bars) in the main classes of small,
small
medium
large
size class
medium and large trees a on ridges, b at intermediate positions,
c on lower slopes, and d for all topographic classes combined.
Bars indicate 95% confidence limits
Forest Ecology
277
(a) ridges
(b) intermediate positions
(c) lower slopes
(d) all topographic positions
16.0
12.0
-1
rgr (mm m year )
14.0
-1
10.0
8.0
6.0
4.0
2.0
0.0
16.0
12.0
-1
rgr (mm m year )
14.0
-1
10.0
8.0
6.0
4.0
2.0
0.0
small
medium
large
size class
small
medium
large
size class
Fig. 4 Relative growth rates within size and topographic
classes in the main plots at Danum: rgr (mm m-1 year-1) for
period 1 (open bars) and period 2 (grey bars) in the main
classes of small, medium and large trees a on ridges, b at
intermediate positions, c on lower slopes, and d for all
topographic classes combined. Bars indicate 95% confidence
limits
relative to the other size classes. Trees on intermediate
positions had 22% and 39% higher mean rgr in plot 1
than 2 in periods 1 and 2, respectively (Supplementary
materials—Appendix 4). Among the plot–plot differences (far fewer than for ma) only one more is
noteworthy: a[50% lower rgr in plot 2 than plot 1 for
large trees in intermediate positions in period 1.
2 (Supplementary materials—Appendix 5). There
was no sign of anisotropy but SAC showed nonstationarity, i.e., varying effect across both plots.
Regular regression models showed very mixed
results. There was a significant fit for main plot 1
of rgr on just elevation in period 1 (P \ 0.01) and on
slope in period 2 (P \ 0.01), however, the variance
accounted for in those cases was very small (2.5%
and 1.7%, respectively). Likewise, for plot 2, the fit
of rgr on elevation was significant in periods 1 and 2
(P \ 0.001) but only for slope in period 2 (P \ 0.01):
r2 lay between 3.2% and 9.2% in those cases. Spatial
specification led to an improvement of the model in
main plot 2: CAR of rgr on elevation and on slope
Growth, topography and spatial autocorrelation
Spatial autocorrelation (Moran’s I) was detected for
distances up to 25 m (period 1) in main plot 1 and up
to 60 m (period 2, when neglecting the significant
cases after some insignificant distances) in main plot
278
A.G. Van der Valk (ed.)
had a significantly improved fit in periods 1 and 2
(LR probability \0.001 in three cases, \0.05 in one
case). In plot 1, CAR led to no significantly improved
fits. Interaction between the periods and elevation and
slope was low.
Fine-scale dynamics
Basic rates
Annualized mortality rate (ma) increased by 42%
between period 1 and sub-period 2a, but by only 7%
between sub-periods 2a and 2b (Table 5a). Correcting
to the basis of t = 5 years and nmin = 2 (correction
factors = 1.109, 0.834 and 0.849 for (sub-) periods 1,
2a and 2b, respectively; see Lingenfelder 2005), mcorr
increased by just 6% between period 1 and sub-period
2a, and by 9% between sub-period 2a to 2b. Mean
relative growth rates across subplots declined by 38%
between period 1 to sub-period 2a but recovered
substantially by 98% in sub-period 2b, 23% higher
than in period 1 (Table 5b). All 16 subplots had lower
rgr (-4 to -82%) in period 2a than in period 1, but
only two decreased further in rgr (-8 to -16%)
during period 2b. Of the 14 subplots with higher rgr,
seven increased by[100% (up to 275%) compared to
period 2a. Against period 1, seven subplots had lower
rgr in period 2b, but in nine subplots growth was still
elevated above the level of the pre-drought period
(three subplots with [100%). Variability of growth
rates was higher in sub-period 2b than before,
pointing—after the more uniform reaction (reduced
growth) immediately after the drought—to a strong
Table 5 Estimates of annualized mortality rate for small trees
in subplots, and the rate corrected for differences in interval
length for period 1 and sub-periods 2a and 2b at Danum, and
positive, but spatially diverse response of trees
starting c. 1 year after the drought.
Effects of tree size and topography
Small trees were divided into four 10-cm size classes,
and in all of these mortality rates of sub-periods 2a
and 2b were higher than in period 1 (Fig. 5a). The
strongest increase in mortality of sub-period 2a over
period 1 was in the 30–40 cm class (80%) with a
smaller increase in sub-period 2b (10%), so that ma in
this size class almost doubled between period 1 and
sub-period 2b. The 20–30 cm size class exhibited the
strongest increase in ma between sub-periods 2a and
2b (22%). All trees were affected immediately in subperiod 2a, most severely those 30 - \50 cm gbh. In
sub-period 2b, trees 20–\40 cm still had increasing
ma but those 10–\20 and 40–\50 cm gbh appeared
to be relatively less affected (Fig. 5a).
Small trees on lower slopes experienced higher ma
in period 1 than those on ridges and intermediate
locations. In sub-period 2a, mortality in all topographic classes increased strongly by 25–65%, but
the differences between classes were smaller than in
period 1, trees on ridges showing a slightly higher
mortality than those on lower slopes. Period 2b
showed a further increase of mortality on the ridges
(9%) and intermediate elevations (48%), but a
decrease (16%) on lower slopes. Trees on intermediate elevations reached the highest mortality rates
across the three topographic classes and periods
(3.05% year-1). Comparing ma of sub-period 2b with
that of period 1, the intermediate class more than
the corresponding mean subplot (±SE) absolute (agr) and
relative (rgr) stem growth rates
Period / Subperiod
1
2a
2b
(a) Annualized mortality (% year-1):
Mean (ma)a
1.53
2.17
Overall mean at t = 5 years (mall)
1.57
2.06
2.18
Corrected overall mean (mcorr)
1.70 ± 0.11
1.81 ± 0.10
1.97 ± 0.23
2.44 ± 0.18
1.60 ± 0.13
2.91 ± 0.33
11.12 ± 0.79
6.90 ± 0.54
13.68 ± 1.56
2.32
(b) Growth rates:
agr (mm year-1)
rgr (mm m-1 year-1)
a
Weighted mean ma values across subplots were almost identical
Forest Ecology
279
(a) size classes
5.0
4.5
4.0
-1
m a (% year )
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
10-20
20-30
30-40
40-50
size class (cm gbh)
5.0
(b) topographic positions
increased with size class: the smallest trees were most
affected by the drought (Fig. 6a). Trees on intermediate topographic locations had the highest rgr in
period 1, and also showed the greatest decline
between period 1 and sub-period 2a (-45%) compared to trees on ridges and lower slopes (Fig. 6b).
Trees on lower slopes grew slightly better than those
on ridges in period 1 and sub-period 2a. Trees on
ridges were more affected in sub-period 2a (40% less
rgr than in period 1) but recovered better than those
on lower slopes in sub-period 2b (135 vs. 65%,
respectively, compared to period 2a) and then
displayed the strongest gain in growth compared to
period 1 (42%) and the highest rates of all topographic classes in all periods (Fig. 6b). Analysis for
4.5
(a) size classes
16
3.5
14
3.0
2.5
12
-1
year )
-1
m a (% year )
4.0
2.0
rgr (mm m
-1
1.5
1.0
0.5
0.0
lower slope intermediate
ridge
10
8
6
4
all
2
topographic classes
0
Fig. 5 Mortality rates of small trees in the subplots at Danum
(ma, % year-1) for period 1 (open bars), and subperiods 2a
(light grey bars) and 2b (dark grey bars): a in four 10-cm gbh
classes, and b for three topographic classes. Bars indicate 95%
confidence limits
10-20
20-30
30-40
40-50
size class (cm gbh)
16
(b) topographic positions
12
-1
-1
rgr (mm m
doubled (increased by 104%) in mortality and ridges
had 80% higher mortality but ma for trees on lower
slopes was elevated by only 6% (Fig. 5b).
The decline of rgr between period 1 and subperiod 2a, and subsequent recovery between subperiods 2a and 2b was apparent across all size classes,
this becoming less pronounced with increasing size
(Fig. 6a). Mean growth rate of the smallest trees
(10–\20 cm gbh) in sub-period 2a was 45% lower
than in period 1 but increased by 129% between
period 1 and sub-period 2b. All size classes had
higher rgr in sub-period 2b compared to period 1 (by
16–25%). Differences in growth between size classes
were small in period 1 and more variable in subperiods 2a and 2b. During sub-period 2a, growth rates
year )
14
10
8
6
4
2
0
lower slope intermediate
ridge
all
topographic class
Fig. 6 Relative growth rates of small trees in the subplots at
Danum (rgr, mm m-1 year-1) for period 1 (open bars), and
subperiods 2a (light grey bars) and 2b (dark grey bars): a in
four 10-cm gbh classes, and b at three topographic classes.
Bars indicate 95% confidence limits
280
In period 1, the growth of trees that died between 1996
and 1999 was a little more than half of that of trees that
were still alive in 1999 (Fig. 7a). This effect was again
visible for trees that died between 1999 and 2001: their
growth rates of period 1 lay still well below those of the
alive-trees in 2001 (29% for rgr, less pronounced with
21% for agr). Growth in sub-period 2a of trees that died
during sub-period 2b was even two-thirds lower than
that of the trees that still lived in 2001 (Fig. 7b). Both
Mann–Whitney U-test and Kolmogorov–Smirnov test
showed highly significant (P \ 0.001) differences
between all combinations. Periodic mortality in period
2 fell nearly 3-fold between trees with \2 and those
with C10 mm m-1 year-1 rgr in period 1 (from 17 to
6%; Fig. 7b).
(a)
12.0
-1
10.0
-1
rgr (mm m year )
Growth and subsequent mortality
14.0
8.0
6.0
4.0
2.0
0.0
alive 99 dead 99 alive 01
dead 01 alive 01
rgr in period 1
25
dead 01
rgr in period 2a
(b)
20
periodic mortality (%)
SAC in subplots showed even fewer significant fits
than on the whole plot level, again with little variance
accounted for. Including the CAR model only proved
to be useful in one case (main plot 1, sub-period 2b).
A.G. Van der Valk (ed.)
15
10
Valid and invalid growth rates
5
Proportions invalid and sources of invalidity
0
< 0.1
0.1 - <2 2 - <5
5 - <10 10 - < 25
-1
≥ =25
-1
rgr classes (mm m year )
The number of trees that were classified as having
unsuitable gbh measurements decreased by 10%
between 1996 and 2001 (Supplementary materials—
Appendix 6; information was not available for the
first enumeration of 1986). This was largely because
of the much smaller (\50%) number of trees where a
new PoM had to be established (or an existing PoM
moved) in 2001 compared to 1996—understandable
given the higher probability of losing a paint mark in
the longer (10-year) interval. Conversely though,
measurements were more affected by unsuitable
stem conditions in 2001 than 1996 (Supplementary
materials—Appendix 6).
During the part-enumeration of 1999 (where only
‘tree status’ and ‘condition of stem’ had been
recorded), the status ‘standing’ (DS) was attributed
to almost half of the dead trees, compared to much
lower proportions in 1996 and 2001. Close to a third
of dead stems were recorded as ‘damaged’ (DB, DA,
DU) in 1999, but this status was attributed to around
half of the trees in 1996 and 2001. Notably, the
proportion of trees with lianas or liana damage
Fig. 7 Growth and subsequent mortality at Danum: a Relative
growth rates of small trees in subplots for period 1 (open bars)
and subperiod 2a (grey bars) categorized according to whether
they lived (wide hatching) or died (narrow hatching) in
subsequent subperiods; bars are SEs of subplot means. b
Change in periodic mortality in period 2 of trees with
increasing (valid) relative growth rate in period 1. Sample
sizes of the six successive classes were 1508, 1950, 2336,
2483, 3447 and 1550. Bars indicate 95% confidence limits
increased steadily between 1996, 1999 and 2001
(Supplementary materials—Appendix 6).
The number of invalid growth rates increased by
40% (from 1754 to 2453), however, because mainly
the newly unsuitable measurements in 2001 were not
all for the same trees as in 1996—in 954 cases
(Table 6). (The remaining difference in unsuitable
measurements versus invalid rates in both periods/
enumerations was due to (a) trees regressing \10 cm
gbh, and (b) growth rates additionally excluded
because agr B-4 mm year-1.) In periods 1 and 2,
12% and 16%, respectively, of the rgr values were
invalid. Losing old PoMs and damage to stems were
Forest Ecology
281
Table 6 Relative contributions (%) of the causes of unsuitability that led to invalid growth rates
Period
Damaged
1
2
26.4
25.1
Lianas
5.1
7.4
Moved PoM
7.8
1.2
56.6
16.4
0.6
0.0
New PoM
Relascope/laser/callipers
Buttresses
0.2
2.1
Irregular stem
2.5
5.8
Absolute growth rate B-4 mm
1.0
Invalid at start of period
Other reasons (regressors, etc.)
1.3
38.4
trees, but much more (20–43%) in sub-periods 2a and
2b. The consequence would have been a decrease in
rgr between period 1 (small) and sub-period 2a by
only 11% and an increase between sub-periods 2a and
2b by only 65%. Including the small negative rates
([-4 and \0 mm year-1) but setting all invalid ones
to zero growth, i.e., adding 304 to 2567 (depending
on the period) rgr values of 0, growth rates would
have been substantially lowered (by 8–17%) with the
effect that the change in rgr would have been only
6% between periods 1 and 2 but similar between the
sub-periods (P1 to P2a: ?38%, P2a to P2b: ?103%)
when compared again to the preferred ‘problem-free’
approach.
2.4
See main text for numbers of invalid rates per period and
Supplementary material—Appendix 6 for totals
the other main reasons why some growth values
became invalid (although this ranking does depend on
the importance given to the individual categories
because the classifications concerned multiple aspects
(CoS, PoM, MeM, etc.), a stem could have been
damaged and been measured at a new PoM: the
ranking chosen here is as shown in Table 6 (from top
to bottom). Across size classes, the proportion of
invalid growth rates increased with size, with a
similar shape in both periods (Fig. 8a).
Comparison of valid with invalid growth rates
Being influenced by very negative values, the mean
of the invalid rates lay well below (by 26–68%) the
mean of the valid rates and in sub-period 2a the mean
invalid rgr was negative (Fig. 8b). Both the increases
in rgr between periods 1 and 2 and between subperiods 2a and 2b, as well as the decrease between
period 1 (subplots, small trees) and sub-period 2a,
were much less pronounced for valid compared to
invalid growth rates, indicating an underestimation of
the changes between periods.
Setting those growth rates to zero where agr was
[-4.0 and \0 mm year-1 and dropping the large
negative values B-4 mm year-1 (which yielded
very similar results as when setting all agr values
\0 mm year-1 to zero), resulted in elevated rgr
compared to the preferred approach, only slightly (up
to 1%) in periods 1 and 2, and in period 1 for small
Immediate effect of 1997/1998-drought
Mortality and growth rates of period 1 (1986–1996)
were assumed to hold constant until the onset of the
1998-drought on 4 April 1998, the date on which the
30-d-rt of precipitation had fallen \100 mm for
10 days. Period 1 was extended to tP1x, by 1.84 years,
leaving a drought sub-period 2x of 0.78 years (see
Fig. 2). The number of trees present at the start of subperiod 2x (n98) was estimated from: n98 = n96 (1 ma_P1)tP1x-tP1. The number of trees dying in extension
was: nd1x = n96–n98, and consequently those dying in
sub-period 2x: nd2x = nd99–nd1x. The resulting ma for
sub-period 2x was 3.64% year-1 (n96 = 4239,
n98 = 4120, nd99 = 237). Using ma estimates uncorrected for interval length (1.53% year-1), ma more
than doubled (increase of 138%) between period 1
(and sub-period 1x) and sub-period 2x. Applying the
correction procedure developed for sub-periods 2a
and 2b (above, and Lingenfelder 2005), and tentatively extrapolating the curve back from 1.0 to
0.78 years, an approximate correction factor which
places ma_2x on a 5-year interval basis is 0.90. This led
to a corrected value of 3.28% year-1, a slightly lessthan-doubling in ma (increase of 93% on 1.70%
year-1 in Table 5). The absolute growth rates of
period 1 were applied at the start of sub-period 1x to
the gbh values of 1996. From the gbh values in 1998
so estimated, rgr for sub-period 2x could be found
(trees 10 - \50 cm gbh). Mean growth rates in subperiod 2x were negative: agr = -0.44 mm year-1,
rgr = -2.38 mm m-1 year-1. During sub-period 2x
small trees on average therefore decreased by
0.34 mm gbh (or 0.11 mm dbh).
282
100
(a)
80
relative frequency (%)
Fig. 8 Valid and invalid
growth rates: a Relative
distribution of invalid
growth rates (trees C10 cm
gbh at start of a period) in
size classes, and for all trees
in period 1 (open bars) and
period 2 (grey bars). b
Mean relative growth rates
(±SE) of invalid (wide
cross-hatching) versus valid
(open bars) values for the
main plots (all trees, periods
1 and 2) and the sub-plots
(trees C10–50 cm gbh at
start of period 1 and subperiods 2a and 2b,
respectively) at Danum
A.G. Van der Valk (ed.)
60
40
20
0
15
25
35
45
55
65
75
85
95
125
175
225 >=250 totals
size classes (cm gbh at start of period; mid points)
(b)
15
ngr (mm m
-1
-1
year )
10
5
0
-5
P1
P2
P1 (small)
P2a
P2b
period / sub-period
Discussion
Methodological and analytical considerations
Mortality and recruitment
The dynamics of the forest at Danum was based on
measurements of two replicate 4-ha plots containing
c. 19 K trees C10 cm gbh spanning 15 years.
Taxonomic identification was almost 100% and done
to a high level, with revisions at later enumerations
reducing the proportion of singletons. Mortality,
recruitment and stem growth rates of all trees from
1986 to 1996, and from 1996 to 2001 (periods 1 and
2) could be estimated at the plot level but only
mortality and growth rates of small trees, from 1996
to 1999 and 1999 to 2001 (sub-periods 2a and 2b)
were achievable at the subplot level (c. 1/3 of the plot
Forest Ecology
area). Care was taken to confirm that trees recorded
as having died really were dead (ideally for any
enumeration this should be checked 3–6 months
after), and that recruits corresponded precisely to
the C10 cm lower gbh limit. At the plot level, rates in
period 2 included trees that recruited at the end of
period 1. By contrast, at the subplot level rates in subperiods 2a and 2b were based on only the survivors of
the previous period or sub-period. Accordingly, the
results given here are slightly different from those of
Newbery and Lingenfelder (2004) where the 1996recruits for sub-period 2a were included. The important contribution of regressors in the dynamics
calculations (leading to alternative calculations of
gains and losses to the population, Supplementary
materials—Appendix 2) was, accordingly, only possible at the plot level and for period 2.
Periodic mortality rates were slightly higher than
recruitment rates in both periods, this being in part
due to the underestimation of true recruitment rates.
Without assuming unrealistic population equilibrium
conditions, even for period 1, there is to date no fully
satisfactory way of accounting for recruits which die
within a period and go unrecorded at the next
enumeration. The best corrected value for ma
increased by 25% between periods 1 and 2, while
ra (uncorrected) increased only 12%, giving the
impression that recruitment lagged behind mortality.
However, the lower ra (than ma) was probably due to
a combination of the evident long-term succession
(Newbery et al. 1992), the influence of the most
recent drought (Newbery and Lingenfelder 2004) and
the methodological underestimation. It means further
that such a data set on tree dynamics—based on plots
remeasured at intervals of several years—cannot be
complete. It is not justified even to assume that ma
and ra are constant with time, which on the one hand
raises a problem for corrections of ma for interval
length (Sheil and May 1996; Newbery and Lingenfelder 2004), and on the other hand questions whether
ra can be similarly corrected (possibly in the way
Lewis et al. 2004 have suggested), given that recruits
of different species will also have their own different
mortality rates.
Tree death is likely in part to be a consequence of
reduced growth rate. Very low to zero, or negative
growth rates, are often associated with trees in their
last months or years before dying (Kobe 1996; Kobe
and Coates 1997). Of particular interest for Danum
283
is—apart from the time lag—that the difference in
rgr of 2001 alive and dead trees was much larger for
sub-period 2a than period 1, the former being directly
associated with the 1998 drought perturbation. The
larger difference for the 1999 than 2001 alive and
dead trees’ rgr in period 1 lends support in the same
direction. A similar association of rgr with mortality
has been shown by Chao et al. (2008).
Local-scale heterogeneity in forest dynamics was
evident from the different responses of the two plots.
Mortality rate (ma) changed more in plot 2 than 1, but
the converse was the case for ra: plots differed less in
ma in period 2 than 1 (a convergence) but differed
more in 2 than 1 for ra (divergence). The plots
differed in important details of topography especially
the small stream running across plot 2, and the more
exposed ridge in plot 1 (Fig. 1). It is interesting that
often medium-sized trees in intermediate positions
showed the largest plot–plot differences, suggesting
that small (understorey species) and large (mostly
canopy species) trees were adapted at the extremes of
the gradient but between them drought caused the
most reactivity.
Including regressors, and using a fixed population
size threshold (to find alternatively losses and gains),
had important consequences for these calculations. Of
the two periods, evaluation of the dynamics was more
complete for period 2 than 1 because information on
regressors at the start of period 1 was lacking. A
critical unknown concerns the dynamics of trees close
to the minimum gbh used in the enumeration. This
may perhaps be overcome in the future by closer study
of subsamples of trees in the c. 7.5–25-cm gbh range
over a series of shorter time intervals. More intensive
sampling (with more persons involved), however,
would mean more interference to the vegetation.
In the present analysis, data from the two replicate
plots have been combined because overall plot
differences were small compared with those over
time (Newbery and Lingenfelder 2009). Confidence
limits on means of ma and rgr in tree-size and
topographic classes approximately indicated the
between-tree variability. Measurements of individuals
will not be spatially or temporally independent from
one another, though, and the true limits are likely to be
slightly larger. Statistical comparisons between classes are inappropriate for another reason—the classes
were arbitrarily defined on a continuous scale. Spatial
auto-correlation was addressed in the analysis of
284
growth in relation to topography with individual tree
elevations and slopes rather than classes. The end
result was that it had a relatively very small effect.
Stem growth
In the analysis reported in this paper attention was
given to the determination of the validity of stem gbh,
and hence rgr, and an extensive system of coding for
invalid trees in the field (CoS, MeM, PoM). In the
calculation of mean growth rates of trees per plot or
subplot almost all other tropical studies have sought
ways of correcting questionable gbh values (those
appearing anomalous due to measurement or recording errors for plausible reasons) or unsuitable pairs of
gbh (due to shift in PoM, poor CoS at start and/or end
of the period) so that all surviving trees had an actual
or estimated growth increment, and any finally
omitted from the data set were those remaining
unexplainable extreme negative and positive values.
A major concern of many researchers has been how
to deal properly with the small negative growth
values, and no standard mathematical probability
density function for tree rgr has been found which
caters for the numerous small negative as well as the
few highly positive growth rates encountered. These
negative rates became important in evaluating
drought effects in period 2 at Danum.
Condit et al. (1993) omitted trees whose dbh
decreased by [5% or had an agr of [75 mm year-1,
and left the smaller decreases in the data set. Later
though after excluding those decreasing[25% and the
same class of extreme positives, negative increments
in dbh were removed by resetting the second dbh of a
pair to the first dbh ? 0.5 mm (Condit et al. 2006). No
mention was made of how increments where PoM,
CoS, and MeM (equivalent to the terminology of this
paper) were dealt with. Condit et al. (2004) excluded
trees where the second dbh was C4 SDs (of a
reference remeasurement) below the first one, which
was equivalent to excluding only trees with growth
rates B-5 mm year-1 dbh (-15.7 mm year-1 gbh;
positives [75 mm year-1 were also again excluded).
Editing the data in this way will raise the mean growth
increment unless the removal of the very few extreme
positive values balances the many small negative
ones. Clark and Clark (1999) moved the PoM when
stem irregularities required it, but seemingly used the
second dbh in finding the last period’s growth
A.G. Van der Valk (ed.)
increment even if the PoM was no longer suitable:
the new PoM was applying to the next period. The
data of Phillips et al. (1998) rest on a method of
standardizing dbh measurements at old and new
(shifted) PoMs using ‘the ratio of diameters at both
PoMs’ (Peacock et al. 2007), but it is not explained
how this was actually achieved. Feeley et al. (2007)
simply changed the growth rate to zero for all trees
where the PoM had changed, presumably replacing in
this way both some negative and some positive values,
and Nakagawa et al. (2000) excluded all growth rates
B-2 mm year-1 in diameter and set those [-2 and
\0 mm to zero growth.
Nevertheless, how frequent stem irregularities
were in leading to new PoMs is not mentioned in
any study we could find and it is not possible from any
of them to ascertain what percentage of values were
edited, rounded up, or omitted. Most authors simply
write the problem off as being of ‘negligible’ consequence, and any details pertaining are sometimes
hidden in appendices. Baker et al. (2004), finding plot
basal area increments in Amazonian forests, also
needed to deal with aberrant dbh values. Those with
agr B-2.0 mm year-1 or C40.0 mm year-1 were
left out (following a recommendation of Sheil 1995,
for one forest site in Africa), and those appearing
unusual were replaced by either a value interpolated
from dbh values before and after the datum in question
or if at the end of a series by the median value of the
other trees in its size class. Chave et al. (2008) applied
a similar procedure but with class limits of -5.0 and
45.0 mm year-1, and using means of dbh classes for
substitution: PoMs were only painted when they
deviated from the standard 1.3 m; a possible source of
inaccuracy. In none of these studies is it explained
objectively why the selected cut-off values were used
or a justification of rounding negative values to zero or
small positives was made. It gives the impression of
practical convenience: Sheil (1995) referred to ‘harmonizing’ his data set on the grounds of ‘common
sense’, and Phillips et al. (2002) call their procedure
‘post measurement data checking’ where so-called
‘false’ negatives are rounded up (to zero usually) but
‘false’ positives are not rounded down. In our analysis
for Danum we have sought to avoid these arbitrary
systems. We excluded only extreme negative values
on the basis of an objective statistical technique
(Newbery et al. 1999) and retained all other negative
values as part of the sample of tree measurements. No
Forest Ecology
extreme positive rgr values were omitted because the
maximum agr was 75.5 mm year-1 gbh (24.0 mm
year-1 dbh), for a dipterocarp in period 2. While this
value is well within the limits used by Condit et al.
(2004, 2006) and Sheil (1995), it is not unexpectedly
large for these species and forests. In the present data
set, modifications of negative or invalid growth rates
would have led to different growth levels (elevated or
lowered) and—in the case of setting slight negative
values to zero—possibly an underestimation of the
response of the forest to the 1997/1998 drought.
By excluding invalid trees, estimates of mean
growth rates of valid ones were highly accurate,
especially for the small trees (10 - \50 cm gbh).
Possible biases as a result of unusual growth (e.g.,
buttresses moving upwards or development of reaction-wood on steep slopes), stem irregularities or
measurement uncertainties through the use of optical
instruments for large trees, were minimized. Nevertheless, trees that were labelled invalid because their
stems were defect or unsuitable might have had
relatively slow growth rates if these features were
indicating damage or a stage prior to death. Conversely, large trees with buttresses, especially those
emerging out of the main canopy, might have had
relatively fast (valid) growth rates. Recording stem
growth rates more accurately and completely could
be achieved by a set of 3–5 (multiple) PoMs spaced
along the bole, so that at least one (preferably more)
gave a valid rgr for any period (Dawkins 1956). This
would be prohibitively intensive in field work and as
a trade-off limit the number of trees and area
enumerated considerably.
Including growth rates down to -4 mm year-1,
and not excluding every rate \0 mm year-1 can be
defended on grounds of (i) physiology and growth,
since it has been shown in the present and other
studies (e.g., Sheil 2003) that shrinkage of trees due
to loss of stem water does occur to this extent; (ii)
there are measurement errors, so that a tree of zero
growth rate can be recorded with an error of ±1 or
2 mm; and (iii) the logit-plot technique of Newbery
et al. (1999) highlighted a very different frequency
distribution below -4 mm compared with above it
where values formed part of an (unknown) exponential-type family function.
In the treatment of growth data there are two
choices: to substitute unmeasured or erroneous rates
by estimates (medians, means, interpolated values,
285
ceven by zeros or small positives), or to leave them
as unmeasured, and accept that where two gbh
values do not meet acceptable accuracy then the rgr
remains unknown. In the present paper, the second
choice has been taken because the forest dynamics
is clearly in a short-term non-equilibrium state and
the response to a perturbation is being studied.
Possibly in a steady-state equilibrium forest some
replacement might be defended but even then it
should not be necessary if ‘errors’ and unmeasured
rates are at random and distributed proportionally
across all size classes and species. To obtain agr
and plot level basal area increments would simply
require here a proportional multiplying up. Nevertheless, substitution must introduce bias and the
more the system is away from a steady state the
stronger the likely bias. This is an important issue
given the increasing recognition that many forests
are recovering from recent perturbations (Wright
2005; Chave et al. 2008).
Was the rgr in period 2 (12.5 mm m-1 year-1)
higher than in period 1 (11.6 mm m-1 year-1) then
because period 2 had a greater proportion of invalid
trees than period 1, that is more trees (of largely low
or negative rgr) were removed from the total sample
in period 2 than 1? It cannot be known empirically
what the valid rates of the invalid class would have
been: they are undetermined. It is not even possible to
reasonably assume, based on current knowledge, that
they were proportional to the invalid rates with a
common conversion equation applying to both periods, or that the invalid sample was a subsample of
similar origins and frequency distribution in both
periods. The same argument applies for the subperiods of period 2. The situation is not satisfactory
but indicates the limits of what can be measured and
how far the dynamics of the system can be reliably
interpreted. We recommend that in future authors
could report how many trees in their samples were
edited and omitted, and for what reasons.
Assuming a dynamic equilibrium in order to
substitute for missing values or make the analysis
tractable has been repeatedly shown to be mistaken in
ecology. It is clearly the case for the forest at Danum,
where the continual readjustment in response to past
perturbations means that the system never comes to a
constant state, remaining in flux and unpredictable. A
fundamental concern is how much the drought
influenced the extent of the recording of valid
286
growth, a problem further compounded by the need to
use fine-scale time resolution to detect the dynamic
response at all.
Dynamics and droughts at Danum and in relation
to other tropical forests
Immediate and lagged mortality and growth
Mortality did increase after 1996 by 25% (interval
corrected rates of all trees for both main plots
combined). Taking the subset of small trees measured
in 1999 into account, a rather moderate, continuous
increase by 6% and 9% in sub-periods 2a and 2b,
respectively, was indicated. However, in the calculation of these values, regressors, gains and recruits
were excluded and thus they are probably overestimated. In a recent work at Danum (Newbery and
Lingenfelder 2004), mortality was shown to have
slightly decreased from 1996 to 1999 (the present
sub-period 2a). If that is taken as the basis for the
‘high drought intensity’ period, then mortality started
to take effect some time after the immediate perturbation—but still within the low precipitation-event—
i.e., in the period between 9 months and 3 years after
the drought. Also increases in growth rates did occur
after 1999 (in period 2b), after a very strong decline
in sub-period 2a. Even though during the partial
enumeration from December 1998 to March 1999—
9 months after the peak of the drought—rainfall was
above average (mean 30-d-rt: 275 mm), measurements were done within the drought event that lasted
until mid-April 1999, with the antecedent rainfall
history still indicating a deficit (Lingenfelder 2005;
Newbery and Lingenfelder 2009). It seems reasonable that under these circumstances, water storage in
the outer tree compartments was not refilled by then
and growth was not substantial enough to result in
positive rates. Sheil (2003) reviewed different studies
and performed an exploratory study on tropical
diurnal tree stem diameter variation. He found that
fluctuations in girth (shrinkage and expansion) of
0.5 mm–2.0 mm day-1 were not exceptional. In
Ghana, Baker et al. (2002) observed dry-season (c.
4 months) shrinkages as much as 2.8 mm in diameter
(8.8 mm in girth). Although a theoretical calculation,
the average shrinkage in tree girth of 0.34 mm in c.
9 months (during sub-period 2x) shown in the present
study is therefore not surprising.
A.G. Van der Valk (ed.)
Two effects successively took place at Danum
during and after the strong drought of 1997/1998: (a)
an immediate response in growth (negative impact)
while mortality did not increase or only slightly
increased (resistance, but possibly weakening), followed by (b) lagged responses in mortality (negative
impact) and increased growth (resilience). Harrison
(2001) hypothesised that even though droughts are
not the direct trigger for flowering, they have an
influence on phenology with a general increase of
leaf production and flowering after droughts (offering
an advantage of not flowering during times of heavy
rain which could possibly damage the flowers and
disrupt pollination and possibly having increased
light levels due to increased mortality). Leaf shedding
and flushing within 2 months of experiencing a short
dry spell was found for trees in Sarawak and the
flushing seemed to have induced cambium growth: 2–
4 months after the flushing or 3–6 months after a dry
spell, growth rates peaked on two occasions in 1996
and 1997 (Ichie et al. 2004). At Danum, extensive
defoliation occurred in March 1998 and growth rates
were very low at least until early 1999 when the
partial enumeration took place. As the 1997/1998drought was more intense than the brief dry periods
described in Sarawak, and it is not known when
flushing recommenced at Danum, it is well possible
that this process of shedding and flushing occurred in
a similar but slowed-down manner. Severe water
stress led to abscission of senescent leaves with
reduced stomatal control (Walsh and Newbery 1999),
bud break and flushing assumingly soon after rainfall
increased again (perhaps when 30-d-rt [100 mm),
but hardly any (detectable) cambium growth until
water storage in the trees was completely refilled in
the first quarter of 1999 (possibly in April, when the
antecedent rainfall history was turning positive
again). Nutrient availability on the forest floor may
have been increased by the defoliation and this
additionally provided the basis for the boost in
growth after April 1999.
The delayed increase in mortality after a severe
drought is in contrast to results of other studies in
Borneo. At two different sites within Lambir Hills
National Park, Sarawak, Nakagawa et al. (2000) and
Potts (2003) estimated mortality for pre-drought
(1993–1997) and drought (1997–1998) periods. They
found more than 3-fold higher mortality rates in the
second interval that ended shortly (5–6 months) after
Forest Ecology
the 1997/1998-event. This drought was possibly more
severe in that region than at Danum, indicated by 30d-rt \100 mm for 89 days (at Danum the equivalent
value was 58 days), although the preceding rainfall
history at the Lambir site is not known. However,
‘true annualised mortality’ (Nakagawa et al. 2000)
and ‘exponential mortality coefficient’ (Potts 2003)
for two time intervals of quite different length were
compared: c. 4 years versus c. 1 year. As the decline
of mortality rate in heterogeneous populations due to
dependence on the interval length is especially strong
from t = 1 to t = 2 (Sheil and May 1996), the high
rates of the short drought period in these two studies
might have been substantially overestimated. Nakagawa et al. (2000) also did not find a large decline in
relative growth rates in their drought period.
A similar pattern of mortality was found in East
Kalimantan (Slik 2004). Although only ‘percentages
of dead standing trees’ were given, these were much
higher shortly (8–13 months) than 4 years after the
drought (15.4% compared to 4.2% in the ‘undisturbed’ plots). Plots in logged areas of that study had
an even higher percentage of dead trees, this also
hinting at the possibly increased risk to disturbed
ecosystems. Sites classified as ‘dry’ had more dead
trees than those which were ‘wet’ (Slik 2004). By
contrast, at Sungai Wain, a site close to that of Slik’s,
lagged mortality was found by van Nieuwstadt and
Sheil (2005): 8 months after the drought the proportion of dead trees was 18.5%, increasing to 26.3% at
21 months. In nearly all of these studies, there was no
correction for interval length and the inferred drought
effect was over-estimated.
Size-related effects
Across both main plots, mortality was highest for
medium-sized and lowest for large trees in period 1.
In the second period, mortality increased with
increasing size, large trees being most affected by
the drought, and this was most pronounced on ridges
(although mortality was lower on ridges than on
lower slopes in both periods). This pattern was not
seen in Sarawak, where mortality decreased with
increasing size; however, increase of mortality in the
drought period was also greatest for large trees
(Nakagawa et al. 2000; Potts 2003). On the other
hand, in East Kalimantan, mortality increased with
size (in the unburned plots) too, and the drought had
287
its largest impact on large trees (van Nieuwstadt and
Sheil 2005). The authors of that study ascribed this
effect to the hydraulic limitation hypothesis, where
water stress increases with the height of trees (all else
staying constant) and imposes a greater risk of
cavitation. During moderate droughts, large trees
with deeper-reaching roots might be less affected, but
if water stress is becoming more severe, cavitation
would in addition to faster depletion of their root
zones affect large trees more than smaller ones (van
Nieuwstadt and Sheil 2005). This generally fits with
the Danum data. Yet, the trend found at Danum that
understorey species followed the general pattern (of
increasing mortality with increasing size), but overstorey-species decreased in mortality with increasing
size (Lingenfelder 2005), is contrary to the findings
of van Nieuwstadt and Sheil (2005). Although small
trees in general seemed to be less affected by the
drought, the impact on large overstorey-trees possibly
was not severe enough to increase their mortality.
Conversely to mortality, relative growth rate
decreased with increasing size in both periods and
growth was higher on lower slopes than on ridges in
period 1. In period 2, however, the recovery (i.e., the
increase in growth) was larger on ridges. Growth of
trees on ridge locations reached similar levels as that
of trees located on lower slopes, with medium-sized
trees on ridges even exceeding those on lower slopes
in growth. Although there was some variation
between plots, topography also showed an influence
on growth rates in the regression models. This seems
to imply that the forest species are largely adapted to
where they are on the gradient of elevation and that
the perturbation (seen on the 5-year scale 1996–2001)
did not have a large effect.
Results of the present work provide strong evidence that forest dynamics at Danum from 1986 to
2001 were influenced by the responses to several mildto-moderate and one severe drought. However, the
perturbations were not major disturbances in the sense
that the forest was vitally damaged. Elsewhere we have
demonstrated highly species-specific dynamics, operating in ways that increased some and decreased other
species and so apparently balancing or compensating
one another (Newbery and Lingenfelder 2009). If the
forest was still recovering from catastrophic droughts
c. 90–130 years ago (Newbery et al. 1999; Newbery
and Lingenfelder 2004), then the event of 1997/1998
could be called a ‘set-back’, one that it seems the
288
forest is capable of overcoming. Potentially threatening could be possible future increases in the intensity
and frequency of droughts with shorter betweendrought intervals for recovery that could lead to
serious changes in the structure and the dynamics of
lowland dipterocarp forests (Walsh 1996).
Conclusions
While recent moderate droughts affected the overall
structure of the forest at Danum only slightly, showing
that the forest can indeed accommodate such perturbations, the upper limits of drought frequency or
intensity to which the forest is resilient remain
uncertain. If, as a result of climatic change, drought
events were to increase in the future, the forest might
respond in either of two hypothetical ways: (1) an
increase in faster growing, light-demanding species,
because the canopy remains open for longer periods; or
(2) an increase in drought-tolerant species, especially
in the understorey, because the atmosphere and soil
become drier also for longer periods. We prefer the
second hypothesis because such a guild of droughttolerant species has been demonstrated at the site, and
increasing drought would presumably select them.
Both outcomes would likely result in lower stature and
biomass forest, with reduced densities of the dominant
dipterocarp species that are largely drought-intolerant
until they reach the sub-canopy. For the primary forest
and its conservation, this would mean a substantially
changed upper canopy, unless the understorey were to
respond effectively enough to nurse the dipterocarps to
the same degree as before, and in secondary logged
forests it might lead to a tendency to replace pioneers
by drought-tolerant understorey species, which could
even increase (through nursing) dipterocarp restocking
(Newbery et al. 1999). Clearly, it is essential to
maintain long-term permanent plots like those at
Danum which have the capability of following these
changes, and use the data to model different drought
scenarios.
The link variable between the external driving
stress (e.g., drought perturbation) and forest dynamics
is rgr. This rate is affected by numerous other factors,
external and internal to the tree, but measured on
stem size it is perhaps the best integrative measure of
tree performance. Trees with very low, zero or even
maintained negative rates tend to die, those with
A.G. Van der Valk (ed.)
positive rates enable recruitment into the population
and movement through the size classes. It is therefore
critically important to measure rgr as accurately as
possible and minimize the number of invalid trees
because these introduce uncertainties and even biases
in the final assessment. One way forward is to employ
multiple PoMs, a second would be to use covariates
of tree growth other than gbh. How the methodological and analytical problems highlighted in this paper
are handled can clearly influence the conclusions
drawn about how perturbations influence the dynamics of the ecosystem under study.
Acknowledgements We are grateful to the Danum Valley
Management Committee (DVMC) and the Economic Planning
Unit, Prime Minister’s Office, Malaysia, for permission to
undertake this research; I. Samat for main field assistance; R.C.
Ong (Sabah Forest Department) and G. Reynolds (Royal
Society) for facilitating the work locally; L. Madani (Sandakan
Herbarium) and C.E. Ridsdale (Rijksherbarium Leiden) for
continued taxonomic work. This project was supported by
grant number 3100-59088 from the Swiss National Science
Foundation, Bern. It is part of the Royal Society of London’s
S.E. Asia Rain Forest Research Programme.
References
Alder D (1995) Growth modelling for mixed tropical forest.
Oxford Forestry Institute, Oxford
Baker TR, Affum-Baffoe K, Burslem D, Swaine MD (2002)
Phenological differences in tree water use and the timing of
tropical forest inventories: conclusions from patterns of dry
season diameter change. For Ecol Manage 171:261–274
Baker TR, Phillips OL, Malhi Y, Almeida S, Arroyo L, Di
Fiore A et al (2004) Increasing biomass in Amazonian
forest plots. Philos Trans R Soc B 359:353–365
Beaman RS, Beaman JH, Marsh CW, Woods PV (1985)
Drought and forest fires in Sabah in 1983. Sabah Soc J
8:10–30
Bivand R (2007) The spdep package. Spatial dependence:
weighting schemes, statistics and models, version 04–13.
http://cran.ch.r-project.org
Cane MA (2005) The evolution of El Niño, past and future.
Earth Planet Sci Lett 230:227–240
Chao KJ, Phillips OL, Gloor E, Monteagudo A, Torres-Lezama
A, Martinez RV (2008) Growth and wood density predict
tree mortality in Amazon forests. J Ecol 96:281–292
Chave J, Condit R, Muller-Landau HC, Thomas SC, Ashton
PS, Bunyavejchewin S et al (2008) Assessing evidence for
a pervasive alteration in tropical tree communities. PLoS
Biol 6:455–462
Christensen JH, Hewitson B, Busuioc A, Chen A, Gao X, Held I
et al (2007) Regional climate projections. In: Solomon S
et al (eds) Climate change 2007: the physical science basis.
Cambridge University Press, Cambridge, UK, pp 847–940
Forest Ecology
Clark DA, Clark DB (1999) Assessing the growth of tropical
rain forest trees: Issues for forest modeling and management. Ecol Appl 9:981–997
Condit R, Aguilar S, Hernadez A, Perez R, Lao S, Angehr G
et al (2004) Tropical forest dynamics across a rainfall
gradient and the impact of an El Niño dry season. J Trop
Ecol 20:51–72
Condit R, Ashton P, Bunyavejchewin S, Dattaraja HS, Davies
S, Esufali S et al (2006) The importance of demographic
niches to tree diversity. Science 313:98–101
Condit R, Hubbell SP, Foster RB (1993) Mortality and growth
of a commercial hardwood ‘el-cativo’, Prioria copaifera,
in Panama. For Ecol Manage 62:107–122
Dawkins HC (1956) Rapid detection of aberrant girth increment of rain-forest trees. Emp For Rev 35:449–454
Feeley KJ, Wright SJ, Supardi MNN, Kassim AR, Davies SJ
(2007) Decelerating growth in tropical forest trees. Ecol
Lett 10:461–469
Fox JED (1972) The natural vegetation of Sabah and natural
regeneration of dipterocarp forests. Ph.D. thesis, University of Wales
Gibbons JM, Newbery DM (2003) Drought avoidance and the
effect of local topography on trees in the understorey of
Bornean lowland rain forest. Plant Ecol 164:1–18
Harrison RD (2001) Drought and the consequences of El Niño
in Borneo: a case study of figs. Popul Ecol 43:63–75
Hulme M, Viner D (1998) A climate change scenario for the
tropics. Clim Change 39:145–176
Ichie T, Hiromi T, Yoneda R, Kamiya K, Kohira M,
Ninomiya I et al (2004) Short-term drought causes synchronous leaf shedding and flushing in a lowland mixed
dipterocarp forest, Sarawak, Malaysia. J Trop Ecol 20:
697–700
IPCC (2001) Contribution of working group I to the third
assessment report of the intergovernmental panel on climate change. In: Houghton JT et al (eds) Climate change
2001: the scientific basis. Cambridge University Press,
Cambridge, UK
Kalunsky SP, Vega SC, Cardosa TP, Shelly AA (1998) S?
spatial stats: user’s manual. Springer, Berlin
Kobe RK (1996) Intraspecific variation in sapling mortality and
growth predicts geographic variation in forest composition. Ecol Monogr 66:181–201
Kobe RK, Coates KD (1997) Models of sapling mortality as a
function of growth to characterize interspecific variation
in shade tolerance of eight tree species of northwestern
British Columbia. Can J For Res 27:227–236
Lewis SL, Phillips OL, Sheil D, Vinceti B, Baker TR, Brown S
et al (2004) Tropical forest tree mortality, recruitment and
turnover rates: calculation, interpretation and comparison
when census intervals vary. J Ecol 92:929–944
Lingenfelder M (2005) Rain forest dynamics and response to
drought in a Bornean primary lowland dipterocarp forest.
PhD thesis. University of Bern, Switzerland. http://www.
stub.unibe.ch/download/eldiss/05lingenfelder_m.pdf
Marsh CW, Greer AG (1992) Forest land-use in Sabah,
Malaysia: an introduction to the Danum Valley. Philos
Trans R Soc B 335:331–339
McPhaden MJ, Zebiak SE, Glantz MH (2006) ENSO as an
integrating concept in Earth science. Science 314:1740–
1745
289
Meehl GA, Stocker TF, Collins WD, Friedlingstein P, Gaye
AT, Gregory JM (2007) Global climate projections. In:
Solomon S et al (eds) Climate change 2007: the physical
science basis. Cambridge University Press, Cambridge,
UK, pp 747–845
Nakagawa M, Tanaka K, Nakashizuka T, Ohkubo T, Kato T,
Maeda T et al (2000) Impact of severe drought associated
with the 1997–1998 El Niño in a tropical forest in Sarawak. J Trop Ecol 16:355–367
Newbery DM, Campbell EJF, Lee YF, Ridsdale CE, Still MJ
(1992) Primary lowland dipterocarp forest at Danum
Valley, Sabah, Malaysia: structure, relative abundance and
family composition. Philos Trans R Soc B 335:341–356
Newbery DM, Campbell EJF, Proctor J, Still MJ (1996) Primary lowland dipterocarp forest at Danum Valley, Sabah,
Malaysia. Species composition and patterns in the understorey. Vegetatio 122:193–220
Newbery DM, Kennedy DN, Petol GH, Madani L, Ridsdale CE
(1999) Primary forest dynamics in lowland dipterocarp
forest at Danum Valley, Sabah, Malaysia, and the role of
the understorey. Philos Trans Roy Soc B 354:1763–1782
Newbery DM, Lingenfelder M (2004) Resistance of a lowland
rain forest to increasing drought intensity in Sabah, Borneo. J Trop Ecol 20:613–624
Newbery DM, Lingenfelder M (2009) Plurality of tree species
responses to drought perturbation in Bornean tropical rain
forest. Plant Ecol. doi:10.1007/s11258-008-9533-8
Payne RW, Harding SA, Murray DA, Soutar DM, Baird DB,
Welham SJ et al (eds) (2007) GenStat Release 10 reference manual. VSN International, Hemel Hempstead, UK
Peacock J, Baker TR, Lewis SL, Lopez-Gonzalez G, Phillips
OL (2007) The RAINFOR database: monitoring forest
biomass and dynamics. J Veg Sci 18:535–542
Phillips OL, Malhi Y, Higuchi N, Laurance WF, Nunez PV,
Vasquez RM et al (1998) Changes in the carbon balance
of tropical forests: evidence from long-term plots. Science
282:439–442
Phillips OL, Malhi Y, Vinceti B, Baker T, Lewis SL, Higuchi
N et al (2002) Changes in growth of tropical forests:
Evaluating potential biases. Ecol Appl 12:576–587
Potts MD (2003) Drought in a Bornean everwet rain forest. J
Ecol 91:467–474
R Development Core Team (2007) R: a language and environment for statistical computing, Version 2.6.1.
http://www.R-project.org. R Foundation for Statistical
Computing, Vienna, Austria
Sheil D (1995) A critique of permanent plot methods and
analysis with examples from Budongo Forest, Uganda.
For Ecol Manage 77:11–34
Sheil D (2003) Growth assessment in tropical trees: large daily
diameter fluctuations and their concealment by dendrometer bands. Can J For Res 33:2027–2035
Sheil D, Burslem DFRP, Alder D (1995) The interpretation and
misinterpretation of mortality rate measures. J Ecol 83:
331–333
Sheil D, May RM (1996) Mortality and recruitment rate
evaluations in heterogeneous tropical forests. J Ecol
84:91–100
Slik JWF (2004) El Niño droughts and their effects on tree
species composition and diversity in tropical rain forests.
Oecologia 141:114–120
290
Timmermann A, Jin FF, Collins M (2004) Intensification of the
annual cycle in the tropical Pacific due to greenhouse
warming. Geophys Res Lett 31:L12208
Timmermann A, Oberhuber J, Bacher A, Esch M, Latif M,
Roeckner E (1999) Increased El Niño frequency in a
climate model forced by future greenhouse warming.
Nature 398:694–697
Trenberth KE (1997) The definition of El Niño. Bull Am
Meteorol Soc 78:2771–2777
Trenberth KE, Hoar TJ (1997) El Niño and climate change.
Geophys Res Lett 24:3057–3060
Trenberth KE, Jones PD, Ambenje P, Bojariu R, Easterling D,
Klein TA (2007) Observations: surface and atmospheric
climate change. In: Solomon S et al (eds) Climate change
2007: the physical science basis. Cambridge University
Press, Cambridge, UK, pp 235–336
A.G. Van der Valk (ed.)
Van Nieuwstadt MGL, Sheil D (2005) Drought, fire and tree
survival in a Borneo rain forest, East Kalimantan, Indonesia. J Ecol 93:191–201
Walsh RPD (1996) Drought frequency changes in Sabah and
adjacent parts of northern Borneo since the late nineteenth
century and possible implications for tropical rain forest
dynamics. J Trop Ecol 12:385–407
Walsh RPD, Newbery DM (1999) The ecoclimatology of
Danum, Sabah, in the context of the world’s rainforest
regions, with particular reference to dry periods and their
impact. Philos Trans Roy Soc B 354:1869–1883
Wright PS (1975) The soils of Sabah. Vol. 3: Western parts of
Tawau and Lahad Datu. Ministry of Overseas Development, Surbiton, UK
Wright SJ (2005) Tropical forests in a changing environment.
Trends Ecol Evol 20:553–560
Changes in tree and liana communities along a successional
gradient in a tropical dry forest in south-eastern Brazil
Bruno G. Madeira Æ Mário M. Espı́rito-Santo Æ Santos D’Ângelo Neto Æ
Yule R. F. Nunes Æ G. Arturo Sánchez Azofeifa Æ G. Wilson Fernandes Æ
Mauricio Quesada
Originally published in the journal Plant Ecology, Volume 201, No. 1, 291–304.
DOI: 10.1007/s11258-009-9580-9 Springer Science+Business Media B.V. 2009
Abstract We investigated changes in species composition and structure of tree and liana communities
along a successional gradient in a seasonally dry
tropical forest. There was a progressive increase in
tree richness and all tree structural traits from early to
late stages, as well as marked changes in tree species
composition and dominance. This pattern is probably
related to pasture management practices such as
B. G. Madeira M. M. Espı́rito-Santo (&)
S. D’Ângelo Neto Y. R. F. Nunes
Departamento de Biologia Geral, Universidade Estadual
de Montes Claros, CP 126, 39401-089 Montes Claros,
MG, Brazil
e-mail: mario.marcos@unimontes.br
B. G. Madeira
Departamento de Biologia Animal, Pós-Graduação em
Entomologia, Universidade Federal de Viçosa,
36570-000 Viçosa, MG, Brazil
G. Arturo Sánchez Azofeifa
Earth Observation Systems Laboratory (EOSL),
Department of Earth and Atmospheric Sciences,
University of Alberta, T6G 2E3 Edmonton, AB, Canada
G. Wilson Fernandes
Ecologia Evolutiva e Biodiversidade/DBG, ICB/
Universidade Federal de Minas Gerais, CP 486,
30161 970 Belo Horizonte, MG, Brazil
M. Quesada
Centro de Investigaciones en Ecosistemas, Universidad
Nacional Autónoma de Mexico, Apartado Postal 27-3
(Xangari), 58089 Morelia, Michoacán, Mexico
ploughing, which remove tree roots and preclude
regeneration by resprouting. On the other hand, liana
density decreased from intermediate to late stages,
showing a negative correlation with tree density. The
higher liana abundance in intermediate stage is
probably due to a balanced availability of support
and light availability, since these variables may show
opposite trends during forest growth. Predicted
succession models may represent extremes in a
continuum of possible successional pathways
strongly influenced by land use history, climate, soil
type, and by the outcomes of tree–liana interactions.
Keywords Forest structure Floristic composition
Succession Liana–tree interactions
Land use history
Introduction
Seasonally dry tropical forests (SDTFs) are considered as one of the most threatened tropical
ecosystems (Janzen 1986) and, in Latin America,
*60% of all SDTFs have already been destroyed
(Miles et al. 2006). Current deforestation rates are
still high and unknown for many regions. Between
1980 and 2000, approximately 11,000 km2 (0.6%) of
SDTFs disappeared yearly in the Americas (Miles
et al. 2006) mainly due to slash-and-burn practices
and conversion to agriculture (Murphy and Lugo
1986; Murphy 1995; Miles et al. 2006). Other
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_22
291
292
potential threats include global climate changes,
habitat fragmentation and increasing human population density (Arroyo Mora et al. 2005; Wright et al.
2007). Conservation efforts are concentrated in the
creation of conservation units, which has a very
limited impact. For example, only 1% of SDTFs in
Central America (Janzen 1988; Sánchez-Azofeifa
et al. 2003) and 3.9% in Brazil (Sevilha et al. 2004)
are under some sort of protection.
During the past few decades, forest restoration
after pasture abandonment has increased in importance to complement conservation strategies such as
creation of conservation units (Janzen 1983), since it
can also minimize global problems, such as climate
change (Prentice et al. 1992). However, a fundamental preliminary requirement is to understand how
successional processes operate in each forest type.
Virtually all the current knowledge in tropical forest
succession was obtained in rain forests and may not
be applicable to STDFs (Vieira and Scariot 2006).
For example, SDTF plant species are predominantly
wind dispersed, compared to a high proportion of
zoochory syndromes in tropical moist forests (Howe
and Smallwood 1982; Gentry 1995; Justiniano and
Fredericksen 2000; Morellato et al. 2000). Also, there
is evidence that resprouting is relatively more
important in SDTF than moist forest natural regeneration (Vieira and Scariot 2006; Vieira et al. 2006;
Sampaio et al. 2007). Thus, information on changes
in forest composition and structure, as well as the
abiotic and biotic interactions driving successional
changes is needed for developing successful restoration programs.
Recent studies in Brazilian SDTFs indicate that
late successional forests are composed mainly by
plant species already present in early succession, due
to their high resprouting capacity (Vieira et al. 2006;
Sampaio et al. 2007). Thus, succession in SDTFs may
not follow the ‘relay floristic model’ (Egler 1954),
which predicts a gradual substitution of pioneer by
late species along forest recovery. Instead, these
ecosystems may conform to the ‘initial floristic
composition model’, with pioneer species remaining
in advanced stages of succession (Egler 1954).
However, it is well documented that previous land
use history affects the speed and pathway of forest
succession in a given area after agricultural land
abandonment (Guariguata and Ostertag 2001; Kennard 2002; Chazdon 2003; Vieira et al. 2006;
A.G. Van der Valk (ed.)
Sampaio et al. 2007). Thus, SDTFs under different
land uses (i.e. agriculture vs. cattle ranching) and
management practices (clear-cutting, ploughing and
burning frequencies) may show contrasting regeneration patterns.
Studies on succession and efforts to promote
natural regeneration of SDTFs also have to recognize
the importance of lianas as a key component of forest
structure. In tropical forests, lianas can account for up
to 40% of leaf area and leaf productivity (Hegarty
and Caballé 1991) and can contribute 10–25% of
plant species richness (Gentry and Dodson 1987;
Gentry 1995; Nabe-Nielsen 2001). Lianas greatly
influence tropical forest dynamics, since they reduce
tree growth and fecundity and increase tree mortality
(Putz 1984; Clark and Clark 1990; Schnitzer and
Bongers 2002), rapidly growing in canopy gaps and
suppressing sapling growth (Putz 1984). Therefore,
they can hinder gap-phase regeneration and impede
forest structure recovery, altering patterns of forest
succession (Putz 1984; Clark and Clark 1990;
Schnitzer et al. 2000; Pérez-Salicrup 2001). Lianas
are usually more common in young, secondary
forests and fragment edges, where light availability
is higher, decreasing in abundance with canopy
closure in mature forests (Clark and Clark 1990;
DeWalt et al. 2000; Laurance et al. 2001). In spite of
that there is very little information on liana community changes with succession (but see DeWalt et al.
2000), specially in SDTFs.
In this study, we compared forest fragments in
different successional stages to describe changes in
tree and liana communities in a Brazilian SDTF. We
used the same approach of Kalácska et al. (2004,
2005) and Arroyo Mora et al. (2005) instead of using
age, and our definition of successional stages was
based on the forest structural characteristics. One of
the main concerns of using age since disturbance or
abandonment to define successional stages is that the
structure and composition of stands of the same age
vary drastically depending on past land use, soil type,
topography and propagule availability (Kellman
1970; Sader et al. 1989; Corlett 1994; Guariguata
and Ostertag 2001; Chazdon 2003; Vieira et al. 2006;
Sampaio et al. 2007). By using a structural approach
to successional stages (i.e. vertical and horizontal
forest structure), we eliminate potential confounding
variables related to the land use history (Arroyo Mora
et al. 2005; Kalácska et al. 2005).
Forest Ecology
We tested the following hypotheses about ecological succession in the studied SDTF: (i) successional
dynamics is dominance controlled. In this case, tree
diversity would be higher in intermediate stages of
succession, due to the competitive exclusion of midsuccessional species as the forest matures (Yodzis
1986; Begon et al. 2006). (ii) Succession pathways
conform to the ‘initial floristic composition model’
(Egler 1954), as proposed by Vieira and Scariot
(2006) and Sampaio et al. (2007). (iii) Lianas are
more abundant in early and intermediate successional
stages, decreasing in late forests due to a negative
interaction with trees. For this purpose, we compared
forest structure and composition among early, intermediate and late successional stages of a SDTF in
south-eastern Brazil, simulating the regeneration
process that would naturally occur in this ecosystem.
Methods
Study area
This study was conducted in the Parque Estadual da
Mata Seca (hereafter PEMS), a conservation unit of
integral protection created by merging of four
farmlands in 2000, and managed by the Instituto
Estadual de Florestas (IEF, State Forestry Institute).
The PEMS has an area of 10281.44 ha and is located
in the valley of the São Francisco River, Minas
Gerais state, Brazil, between 14480 3600 –14560 5900 S
and 43550 1200 –44040 1200 W. The original vegetation
of the park is SDTFs, growing on flat and nutrientrich soils (IEF 2000). These forests are dominated by
deciduous trees, with almost 90–95% of loss in leaf
area during the dry season (May–October). The
climate of the region is considered as tropical semiarid (Köppen’s classification), characterized by the
existence of a severe dry season during the winter.
The average temperature of the study region is 24C
(Antunes 1994), and the average annual precipitation
is 818 ± 242 mm (mean ± standard deviation; data
from the meteorological station in the city of Manga,
10 km from the study area). The main economic
activities in the area before protection were extensive
cattle ranching, and bean and corn plantations inside
two central pivots of 80 ha each. Approximately
1,525 ha of the PEMS is covered with abandoned
pasture fields in early regeneration stages, while the
293
remaining area supports dry forest fragments in
secondary and primary stages (IEF 2000).
Sampling
In January 2006, 20 plots of 20 m 9 50 m (0.1 ha
each, 2.0 ha in total) were delimited in early, intermediate and late forest fragments. To determine the
successional stage of a given forest fragment, we
followed the structural approach of Kalácska et al.
(2004, 2005) and Arroyo Mora et al. (2005). These
authors used the forest vertical structure (i.e. the
number of tree crown layers in a vertical profile of the
forest) and horizontal structure (the horizontal distribution of tree crowns per area) to define stages,
regardless of forest age. In this sense, our early
successional stage is characterized by a forest area
composed of sparse patches of woody vegetation,
shrubs, herbs and grasses with a single stratum of tree
crowns composing a very open canopy up to 4 m. This
area was used as pasture for at least 20 years and
abandoned in 2000, though cattle from adjacent
pastures still use the areas occasionally. Intermediate
successional stages have two vegetation layers: the
first one is composed by deciduous trees with 10–
12 m and some emergent trees up to 15 m. The
second layer is formed by a dense understory with
many young trees and abundant lianas. This area was
used as pasture for an unknown period and was
abandoned at the late 1980s. Pastures where both early
and intermediate successional forests fragments now
occur were managed similarly: after clear-cutting, the
area was ploughed to plant exotic grasses and burned
every 2 years right before the rainy season. The late
successional stage is also characterized by two strata,
but the first stratum was composed by taller deciduous
trees which form a closed canopy 18–20 m high. The
second stratum is formed by a sparse understory with
reduced light penetration and low density of young
trees and lianas. There are no records of clear-cutting
in this area for the last 50 years.
Six plots were established in one early successional
forest fragment, the same occurring for intermediate
succession plots. For late succesional stages, eight
plots were established in two forest fragments
(*3 km from each other). All forest fragments were
located under similar topographic, soil and microclimatic characteristics, thus reducing variation in
physical conditions that could affect succession. The
294
20 plots were located along a 5 km transect encompassing these fragments, between 14500 –14510 S and
43570 –44000 W. All plots were situated inside the
original area of a single farm, in which management
practices were similar for all pasturelands in the last 30
years, when the property belonged to the same owner.
Plots from the same successional stage were located
*0.2–1.0 km from each other.
We identified and measured the diameter at breast
height (DBH) of all living trees with a DBH equal or
greater than 5 cm inside all plots. We also visually
estimated the height of these individuals in each plot,
using a 2 m graduate stick as reference. Moreover, all
independently growing liana stems with a DBH equal
or greater than 2 cm had their DBH measured, and
their height was estimated as the height of their host
tree. Lianas were identified at the morpho-species
level, due to difficulties to collect plant vegetative
and reproductive parts. Voucher specimens were
deposited at the herbarium of the Universidade
Estadual de Montes Claros, in Montes Claros, Brazil.
Data analyses
We compared the forest structural characteristics
(height, basal area and tree density) for both the tree
and liana components among successional stages
using general linear models (GLMs) for each characteristic. Then, all factor levels (stages) were compared
using contrast analysis by aggregating level and
comparing deviance change (Crawley 2002). If the
level of aggregation did not significantly alter the
deviance explained by the model, the levels were
pooled together (amalgamation) simplifying the
model. Rejected amalgamation implied that levels
were indeed different and no further comparisons were
made. Thus, the complete model was simplified by
stepwise omission of non-significant terms. All models
were submitted to residual analyses, so as to evaluate
adequacy of error distribution (Crawley 2002).
We computed the Holdridge complexity index
(HCI) (for the tree component only; Holdridge 1967;
Holdridge et al. 1971) as a measure of community
complexity. This index is calculated by the following
equation: CHCI = (Height 9 Density of stems 9 Basal area 9 Number of species)/1,000. The original
HCI considers only trees with DBH [ 10 cm. Thus,
we used a modified version of the index since we
sampled trees with DBH C 5 cm (Lugo et al. 1978).
A.G. Van der Valk (ed.)
The HCI was compared among successional stages
using the same procedure used for the other structural
variables.
To compare tree richness among successional
stages, observed species richness was calculated for
each plot. Estimated species richness was also
calculated using a non-parametric estimator, the
incidence-based coverage estimator (ICE) using species-by-sample data (Colwell and Coddington 1994;
Chazdon et al. 1998), with the software EstimateS 8.0
(Colwell 2006). We used a GLM to compare the
observed species richness among the three successional stages.
In order to assess the variation in species composition between different successional stages (b
diversity), we calculated the Morisita–Horn index
(quantitative). We also calculated Jaccard’s similarity
coefficient (Cj) (Magurran 2004) to examine the
floristic similarity between all plots of the three
successional stages. Then, we used GLMs to test the
relationship between similarity in species composition (measured as Jaccard’s similarity index between
plots) and the distance between plots within and
between successional stages in order to determine if
plot selection had any influence on species composition within successional stages.
To test the relationship between liana density and
tree structural attributes, we used a GLM. The
complete model was then simplified following the
same procedure used for the other structural variables.
Results
Tree community composition
We identified a total of 1,543 tree individuals,
representing 59 tree species and 23 plant families in
the 20 plots (2 ha) of the three successional stages
(Appendix). In the early stage, we found 296
individuals from 24 species, representing 11 families. Three of these families (Fabaceae = 59.1%,
Anacardiaceae = 23.3%
and
Bignoniaceae =
10.1%) corresponded to 92.6% of all individuals in
this stage. From a total of 13 families and 457
individuals (33 species) in the intermediate stage,
three families (Bignoniaceae = 31.3%, Combretaceae = 25.6% and Fabaceae = 21.0%) constituted
77.9% of the individuals. In the late stage, we
Forest Ecology
295
identified a total of 790 individuals belonging to 42
species and 19 families. Again, three dominant
families
(Bignoniaceae = 35.8%,
Fabaceae =
24.4% and Combretaceae = 19.5%) were responsible for 79.7% of all individuals from this stage.
Individual species dominance changed along the
successional gradient, markedly from the early to the
intermediate stages, but only slightly from the intermediate to late stages (Fig. 1). In the early stage,
Senna spectabilis (Caesalpiniaceae) and Myracrodruon urundeuva (Anacardiaceae) had high
relative abundances, 40.5% and 21.6%, respectively.
While S. spectabilis was present only in the early
Senna spectabilis
5-10 cm
Myracrodruon urundeuva
> 10 cm
Tabebuia ochracea
Mimosa tenuiflora
Acacia cf. polyphylla
Aspidosperma pyrifolium
Platymiscium blanchetii
Schinopsis brasiliensis
Manihot anomala
Piptadenia viridiflora
(a)
0
25
50
75
100
125
150
Tabebuia roseo-alba
5-10 cm
Combretum duarteanum
> 10 cm
Anadenanthera colubrina
Caesalpinia pyramidalis
Sapium obovatum
Commiphora leptophloeos
Machaerium scleroxylon
Cochlospermum vitifolium
Machaerium brasiliense
Machaerium acutifolium
0
(b)
25
50
75
100
125
Tabebuia ochracea
150
5-10 cm
> 10 cm
Combretum duarteanum
Tabebuia roseo-alba
Caesalpinia pyramidalis
Anadenanthera colubrina
Myracrodruon urundeuva
Centrolobium sp.
Machaerium acutifolium
Terminalia eichleriana
Sapium obovatum
(c)
0
25
50
75
100
125
150
Number of individuals
Fig. 1 Number of individuals for the 10 most abundant tree
species in early (a), intermediate (b) and late (c) successional
stages, by DBH classes of 5–10 and C10 cm. Species in bold
were represented in more than one successional stage
stage, M. urundeuva appeared in all three successional
stages and was also a dominant species in the late
stage, although with varying abundance/importance
(relative basal area) (Fig. 1). There was little change in
species dominance from the intermediate to late
stages: from the five species, dominant in both stages,
four were dominant in the intermediate and late stages:
Anadenanthera colubrina (Mimosaceae), Combretum duarteanum
(Combretaceae),
Caesalpinia
pyramidalis (Caesalpiniaceae) and Tabebuia roseoalba (Bignoniaceae) (Fig. 1). The only remarkable
change in composition between the intermediate and
late stages was the dominance of Tabebuia ochracea
(28.1%) in the latter, whereas this species was rare
(relative abundance of 1.3%) in the former.
The early stage presented a lower species richness
than intermediate and late stages, but no significant
difference was observed between intermediate and
late stages (Tables 1, 2). Observed species accumulation curves and the species richness estimator (ICE)
showed contrasting results between successional
stages. No stabilization was observed for the early
stage, and observed richness accumulation curve was
below the estimated richness accumulation curve
(Fig. 2a). Both curves indicated stabilization for the
intermediate stage at 0.4 ha, although observed
species richness showed a slight increase at 0.6 ha
(Fig. 2b). For late stages, both curves tended to
stabilized at 0.8 ha (Fig. 2c). In contrast with the
early stage, the difference between the observed and
estimated species richness was much lower for the
intermediate and late stages (Fig. 2b, c).
The greatest similarity was observed between the
intermediate and late stages, with a Morisita–Horn
index of 0.55 (the probability to find an individual of
the same species in a sample of the two stages). There
was a very low similarity between the early and late
stages (Morisita–Horn = 0.062) and, interestingly, an
even lower similarity was observed between the early
and intermediate stages (Morisita–Horn = 0.014),
meaning a high turnover between successional stages.
On the other hand, we observed a higher similarity in
species composition between plots from the same
successional stage than from different successional
stages (Fig. 3). Besides, similarities in species composition were not influenced by the distance between
plots from the same successional stage (n = 58,
F = 0.263, P = 0.61), but decreased with distance
between plots from different successional stages
296
A.G. Van der Valk (ed.)
Table 1 Mean values (mean ± standard deviation) of the tree and liana structural characteristics (height, basal area and density)
and the Holdridge complexity index (CCHI) in three successional stages in the Parque Estadual da Mata Seca, MG
Stage
Height
Basal area
Density
No. of species
CHCI
Trees
Early
Intermediate
Late
3.4 ± 0.8a
3.1 ± 0.8a
49.3 ± 21.0a
8.3 ± 2.4a
0.6 ± 0.5a
b
15.2 ± 3.4
b
b
b
15.0 ± 8.3b
22.0 ± 6.4
c
98.8 ± 17.2
b
46.1 ± 25.7c
8.2 ± 5.2a
0.56 ± 0.1a
32.0 ± 4.2a
–
b
b
b
–
8.0 ± 4.1
c
11.8 ± 5.6
76.2 ± 10.0
c
16.3 ± 1.9
17.3 ± 2.0
Lianas
Intermediate
Late
13.3 ± 6.0
0.36 ± 0.2
2
15.5 ± 6.9
-1
Height given is measured in metres, basal area in m ha , and density in number of individuals ha
statistical difference (refer to Table 2)
Table 2 Analysis of variance of the forest structural characteristics (height, basal area and density), number of tree species
and of the Holdridge complexity index (CCHI) between three
Response variable
Source
d.f.
Deviance
Height
Stage
1
Basal area
Stage
1
Density
Stage
1
113.569
17
Number of species
Stage
1
22.383
Holdridge index
Stage
1
–
–
-0.1
. Different letters indicate the
successional stages and of the liana structural characteristics
between the intermediate and late stages in the Parque Estadual
da Mata Seca, MG
Residual d.f.
Residual deviance
F
P
Errors
49.782
\0.0001
Normal
30.511
\0.0001
Normal
75.154
13.195
\0.0005
Poisson
17
6.90
26.925
\0.0001
Poisson
17
112.43
30.624
\0.0001
Poisson
Trees
241.629
1227.86
406.56
17
17
41.257
342.07
Lianas
Height
Stage
1
90.493
12
112.484
9.654
0.009
Normal
Basal area
Stage
1
0.176
12
0.394
5.359
0.039
Normal
Density
Stage
1
40.826
12
25.552
40.826
\0.0001
Poisson
No lianas were found in early plots
(n = 132, F = 23.161, P \ 0.0001, Fig. 3). Thus,
there was no effect of spatial autocorrelation in species
composition within successional stages.
Forest structure
All the tree structural variables evaluated in this study
varied among the three successional stages (Table 1).
For tree diameter, this pattern is clearly demonstrated
by the higher frequency of large class diameters
([10 cm) in intermediate and late stages, whereas
lower class diameters predominate in early stages
(Fig. 1). Although changes in forest structure from
intermediate to late stages were statistically significant (Table 2), more dramatic changes were observed
from the early to intermediate stage for all structural
variables (Table 1). Overall, the HCI increased 30
times from the early (0.5 ± 0.2, mean ± standard
error) to intermediate (15.0 ± 2.1) stage and only 3.1
times between intermediate and late (46.1 ± 9.2)
stages (Table 1).
In contrast, liana structure showed a reverse
pattern along the successional gradient (Table 1).
We sampled 192 liana individuals in 6 intermediate
plots and 124 individuals in 8 late plots, representing
a significant decrease in stem density along the
successional gradient. No lianas were found in the
early successional stage. Although the height where
lianas were found increased from intermediate to late
stages, basal area and liana density were significantly
higher in the intermediate stages (Tables 1, 2). Liana
density was negatively correlated with tree density,
and the correlation was statistically significant for
both the intermediate and late stages, indicating that
liana density decreases as succession unfolds
(Table 3, Fig. 4).
Forest Ecology
297
80
Cumulative species sampled
Fig. 2 Species
accumulation curves of
trees for the early (a),
intermediate (b) and late (c)
successional stages.
ICE = incidence-based
coverage estimator
60
40
20
0
Cumulative species sampled
(a)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0
0.1
0.2
0.8
150
125
100
75
50
25
0
(b)
Cumulative species sampled
0
0.8
160
120
(c)
80
40
0
0.3
0.4
0.5
0.6
0.7
0.8
Cumulative area sampled (ha)
Jaccard similarity index
Observed
0.7
within stage
between stages
0.6
ICE
Discussion
Tree structure and diversity: successional changes
0.5
0.4
0.3
0.2
0.1
0
0
1000
2000
3000
4000
5000
Distance between two plots (m)
Fig. 3 Linear regression between Jaccard’s similarity index
(response variable) and distance between plots (explanatory
variable) from the same (full line, closed points) and different
(dashed line, open circles) successional stages
Forest structure changed along the successional
gradient according to the general pattern of secondary
succession described for tropical forests, with a
gradual increase in height and basal area (Guariguata
and Ostertag 2001; Kalácska et al. 2004; Ruiz et al.
2005). On the other hand, variation on stem density
along successional stages observed here did not
conform to general patterns observed for both wet
and dry forests. Usually, there is a high density of
stems with low DBH in early and intermediate stages,
298
A.G. Van der Valk (ed.)
Table 3 Analysis of variance of the complete and minimal
adequate linear models of the density of lianas (response
variable) and the forest structural characteristics (tree species
richness, basal area of trees and density of trees) between the
two sucessional stages in the Parque Estadual da Mata Seca,
MG
Source
Errors
d.f.
Deviance
Residual d.f.
Residual deviance
F
P
Quasipoisson
12.16
0.00686
Complete model
Density of trees
1
26.687
12
39.691
Tree species richness
1
0.447
11
39.243
0.204
0.662
Basal area of trees
1
5.884
10
33.359
2.681
0.136
Stage
1
11.778
9
21.581
5.367
0.0457
Minimal adequate model
Density of trees
Stage
Quasipoisson
1
26.687
12
39.691
13.804
0.00341
1
16.246
11
23.445
8.403
0.0145
Fig. 4 Linear regression between the density of lianas
(response variable) and the density of trees (explanatory
variable) in intermediate and late successional stages
and as DBH increases with forest growth, stem
density decreases (Mizrahi et al. 1997; Saldarriaga
et al. 1988; Denslow and Guzman 2000; Kennard
2002; Kalácska et al. 2004; Ruiz et al. 2005). In spite
of the gradual increase in average stem density
observed here, the pattern described above is clearly
illustrated for one species of the PEMS. M. urundeuva, a very common species in Brazilian STDFs
(Oliveira-Filho et al. 1998; Pereira et al. 2003; Salis
et al. 2004), is the second most abundant species in
the early stages, with low DBH values. It was also
encountered in the late stages, where this species was
the sixth most abundant tree, but with much higher
DBH values (see Fig. 1). Nevertheless, the successional changes in stem density observed at the PEMS
were also detected for a STDF in Chamela, Mexico
(Kalácska et al. 2005).
In spite of the high species dominance observed
for all successional stages, the succession gradient
observed here did not conform to the dominancecontrolled community model, which predicts higher
tree diversity in mid-successional forests (Yodzis
1986; Begon et al. 2006). According to this model,
tree species diversity is low in early successional
stages, which are colonized by a limited group of
pioneers. As succession progresses, other species
invade the area and intermediate regeneration stages
are composed by a high number of mid- and late
successional tree species. As the forest matures
towards the climax, late, efficient competitor species
oust mid-successional species, causing a decrease in
tree community diversity, which is dominated by a
lower number of late species (Yodzis 1986; Begon
et al. 2006). However, a higher tree diversity in late
compared to intermediate stages was described for
other STDFs, such as Chamela, Palo Verde (Kalácska
et al. 2005) and Providence Island (Ruiz et al. 2005),
whereas the successional gradient observed in Santa
Rosa conforms partially to the predicted for dominance-controlled tree communities (Kalácska et al.
2005). One likely alternative explanation is that the
forest sites considered as late stages in these studies
(including the present) are more similar in structure
and diversity to secondary forests in an advanced
stage of regeneration than to mature forests. In this
case, we would expect a decrease in tree diversity in
the late stages in the next decades. Indeed, some
estimates indicate that the recovery time for lowland
dry forest ecosystems is around 150 years (Opler
et al. 1977). Nevertheless, the structure of the late
Forest Ecology
stages in the above-mentioned forests is consistent
with that described for mature STDFs (see Murphy
and Lugo 1986; Ruiz et al. 2005). Thus, more longterm studies are necessary to test whether succession
patterns in STDFs conform to the predicted for
dominance-controlled communities.
The relative importance of seed colonization and
resprouting in forest recovery is controversial, originating opposing succession models. In the ‘relay
floristics model’, a gradual species substitution is
expected across time, whereas the ‘initial floristic
composition model’ predicts that pioneer species
remain in advanced stages of succession (Egler
1954). The successional gradient analysed in the
present study corroborates the former model, since
there are striking changes in tree community composition from early to intermediate and late stages. On
the contrary, some recent studies on Brazilian SDTFs
found that a vast majority of the plant species are
present in recently abandoned pastures and remain
later in the succession (Vieira et al. 2006; Sampaio
et al. 2007). These authors related this pattern to the
great resprouting capacity of dry forest tree species,
which can be affected by the intensity of pasture
management practices such as fire, clear-cutting and
tractor use. In our areas, ploughing probably removed
the majority of plant roots after clear-cutting, preventing resprouting and reducing the presence of late
species in early sites. In areas recovering from this
type of pasture management, succession is more
likely to conform to the ‘relay floristic model’. When
resprouting is intense, succession may resemble the
predicted by the ‘initial floristic composition model’.
However, the presence of multi-stemmed trees in
early plots and the occurrence of M. urundeuva in all
three successional stages of the PEMS suggest that
regeneration by resprouting is also occurring. Thus,
succession models may represent extremes in a
continuum of possible successional pathways
strongly influenced by land use history.
The great tree species substitution from early to
intermediate and late stages is probably related to
changes in light penetration through the forest canopy
along the successional gradient. Light quality and
quantity may have a profound effect in determining the
survivorship of shade-intolerant pioneer trees. For
instance, S. spectabilis, the most abundant species in
the early stage, can be considered a pioneer species
according to the classification proposed by Swaine and
299
Whitmore (1988). The plant is a short-lived heliophyte,
with rapid growth in height, especially under direct
sunlight (Lorenzi 1992). This can explain the absence
of this species from the shaded understory of intermediate and late forest fragments. Also, light availability
can affect seed germination and seedling growth in
early successional stages. Though there is only scattered information on the ecophysiology of the majority
of the tree species encountered in this study, some of
the dominant species in early successional plots have
either positive photoblastic or neutral seeds. For
instance, although M. urundeuva is considered a
climax, shade-tolerant species, it was the second most
abundant species in early stage plots, and its seeds may
be able to germinate both in gaps, exposed to direct
sunlight and daily temperature fluctuations, and in the
understory, where diffuse light and lower daily temperature variations predominate (Silva et al. 2002).
Similarly, the germinative behaviour of A. polyphylla,
fifth most abundant species in early stage plots, also
indicates that it can germinate in different-sized gaps,
exposed to diverse temperature and light conditions
(Araújo Neto et al. 2003). Thus, it is likely that most
species encountered in early plots are adapted to open
canopy conditions, though some late, shade-tolerant
species can also regenerate by sprouting in these areas.
Liana structure and liana–tree interactions
Lianas represented a very important structural component of the SDTFs in PEMS, with marked changes
along the successional gradient considered in this
study, probably related to changes in light and support
availability. In the early stages light is not limiting, but
there are few branches thick enough to support the
growth of lianas with more than 2 cm. In fact, the
dependence of large lianas on large trees has been
reported in other studies (Clark and Clark 1990; NabeNielsen 2001; Phillips et al. 2002). In the intermediate
stages, light availability decreases, but the canopy is
still open enough to allow successful liana establishment, due to the presence of adequate support
(Sánchez-Azofeifa, unpubl. data). As succession progresses, the canopy increases in height (from 8.2 m in
intermediate stages to 13.3 m in late stages, on
average) and continuity, reducing the habitat suitability for lianas for two reasons: first, energetic costs
associated with ascent may reduce liana capacity to
climb a great distance to the canopy (DeWalt et al.
300
A.G. Van der Valk (ed.)
2000); second, lianas are light-demanding (Castellanos 1991; Teramura et al. 1991) and, in closed
canopies of late forests, they are only able to establish
and grow in tree gaps (Putz 1984; Schnitzer and
Carson 2001). These factors may be responsible for the
decline in abundance observed here from intermediate
to late stages, as well as for the negative relationship
verified between liana density and tree density,
considering all the sampled plots from both stages.
Our results corroborate other studies, mostly from
wet forests, which reported a higher density of lianas
in younger forests (DeWalt et al. 2000; Schnitzer and
Bongers 2002; Kuzee and Bongers 2005; Schnitzer
2005). We are not aware of any other study comparing liana structural characteristics along a succesional
gradient of SDTFs. However, Kalácska et al. (2005)
found indirect evidence that lianas are also more
abundant in intermediate successional stages. They
reported that the proportion of liana leaves collected
on leaf traps along a successional gradient in a SDTF
in Santa Rosa, Costa Rica, were higher in the
intermediate than early and late stages. Thus, there
is an urgent need for more studies concerning
successional changes in the liana component of
SDTFs, in order to understand the regeneration
processes in these ecosystems and allow comparisons
with the better studied wet forests.
Conclusions
Studies with successional gradients can be very useful
to understand natural regeneration patterns in SDTFs,
and to compare the consequences of different land use
histories for forest recovery. Resprouting can be a very
common mechanism of SDTF regeneration, but its
intensity may depend on previous land management
practices. Pasture colonization through seed germina-
tion is more likely in ploughed areas, which can lead to
the classical succession pattern characterized by a
gradual but marked change in community composition. However, a mixture of both processes is probably
the rule for most SDTFs. Many others factors are
thought to affect forest regeneration, such as climate,
soil type and the abundance of lianas. To our
knowledge, this was the first study that analysed
changes in liana structure along a successional gradient in SDTFs, providing a possible explanation for
their higher abundance at intermediate secondary
forests based on trade-offs in support and light
availability. Lianas certainly play an important role
in SDTF recovery, and the strength of their influence
on tree growth needs further attention, with long-term
and experimental studies to allow comparisons with
the better known wet forests.
Acknowledgements The authors thank Anna Paola Biadi
Bicalho, Elton Bordoni, Rodrigo Braga Nunes, Hisaı́as
Almeida, Mariana Rodrigues Santos, Diego Oliveira Brandão
and Gládson Borges for their help during field work. We thank
all the staff of the Instituto Estadual de Florestas (IEF) for
allowing us to stay and work at the PEMS, and for logistical
support. We specially thank José Luı́s Vieira (IEF) for his
invaluable field assistance. We are also very grateful to three
anonymous reviewers for their comments on the early versions
of this manuscript. This work was carried out with the aid of a
grant from the Inter-American Institute for Global Change
Research (IAI) CRN II # 021, which is supported by the US
National Science Foundation (Grant GEO 0452325), and from
the Fundação de Amparo à Pesquisa de Minas Gerais
(FAPEMIG CRA 2288/07). Logistical support by the
University of Alberta is also acknowledged. Geraldo Wilson
Fernandes acknowledges a grant provided by CNPq (304851/
2004-3). Bruno Gini Madeira greatly acknowledges a
scholarship from CNPq (140250/2004-2). This study was in
partial fulfilment for the PhD requirements of Bruno Gini
Madeira.
Appendix
Table 4 List of tree species (DBH C 5 cm) identified in the 20 plots in three successional stages in the dry forest of the Parque
Estadual da Mata Seca, MG
Family
Species
Stage
Early
Anacardiaceae
Astronium fraxinifolium Schott
x
Myracrodruon urundeuva All.
x
Schinopsis brasiliensis Engl.
x
Spondias tuberosa Arruda
Intermediate
Late
x
x
x
x
Forest Ecology
301
Table 4 continued
Family
Species
Stage
Early
Apocynaceae
Aspidosperma pyrifolium Mart.
Intermediate
x
x
Aspidosperma polyneuron Mull.Arg.
x
Aspidosperma subincanum Mart.
x
Araliaceae
Aralia warmingiana (Marchal) J. Wen
x
Arecaceae
Syagrus oleracea (Mart.) Becc.
Asclepiadaceae
Calotropis procera (Aiton) W.T. Aiton
x
Asteraceae
Vernonia sp.
x
Bignoniaceae
Tabebuia impetiginosa (Mart. ex DC.) Standl.
Tabebuia ochracea (Cham.) Standl.
Bombacaceae
Burseraceae
x
x
x
x
Tabebuia roseo-alba (Ridl.) Sandwith
Zeyheria tuberculosa Bureau ex Verlot
Late
x
x
x
x
x
x
x
Cavanillesia arborea K. Schum.
x
x
Chorisia glaziovii (Kuntze) E. Santos
x
x
Pseudobombax longiflorum (Mart. & Zucc.) A. Robyns.
x
x
Commiphora leptophloeos (Mart.) J.B. Gillett
x
x
x
Cactaceae
Cereus jamacaru DC.
x
Cactaceae
Pereskia grandifolia Haw.
x
x
Cochlospermaceae
Cochlospermum vitifolium Spreng.
x
x
Combretaceae
Combretum duarteanum Cambess.
x
x
Terminalia eichleriana Alwan & Stace
x
x
Cnidoscolus pubescens Pax
x
Euphorbiaceae
Manihot anomala Pohl
x
x
Maprounea guianensis Aublet
x
Sapium obovatum Klotzsch ex Mull. Arg.
Fabaceae
Acacia cf. polyphylla DC.
x
Acacia sp. 1
x
x
x
x
x
Acacia sp. 2
x
Anadenanthera colubrina (Vell.) Brenan
Bauhinia sp.
x
Caesalpinia pyramidalis Tul.
x
Cassia multijuja Rich.
x
Centrolobium sp. Mart. Ex Benth.
x
x
x
x
x
x
x
x
Chloroleucon tortum (Mart.) Barneby & J.W.
Grimes
x
Enterolobium contortisiliquum Vell. (Morong.)
x
Goniorrhachis marginata Taub.
x
x
Machaerium acutifolium Vog.
x
x
Machaerium brasiliense Vog.
x
x
Machaerium cf. floridum (Mart.) Ducke
x
x
Machaerium scleroxylon Tul.
x
Mimosa tenuiflora Benth.
x
Piptadenia viridiflora Kunth. (Benth.)
x
x
Plathymenia reticulata Benth.
x
x
Platymiscium blanchetii Benth.
x
x
Pterocarpus rohrii Vahl
Senna spectabilis (DC.) HS. Irwin & Barneby
x
x
Meliaceae
Cedrela odorata L.
Myrtaceae
Myrtaceae sp.
x
x
x
Nyctaginaceae
Ramisia brasiliensis Oliv.
x
302
A.G. Van der Valk (ed.)
Table 4 continued
Family
Species
Stage
Early
Picramniaceae
Picramnia sellowii Planch.
Polygonaceae
Coccoloba schwackeana Lindau
Rhamnaceae
Zizyphus joazeiro Mart.
x
Rubiaceae
Randia armata DC.
x
Sterculiaceae
Sterculia striata A. St.-Hil. & Naudin
Ulmaceae
Celtis iguanaea (Jaqc.) Sarg.
Vochysiaceae
Callisthene major Mart.
Total
References
Antunes FZ (1994) Caracterização Climática—Caatinga do
Estado de Minas Gerais. Informe Agropec 17:15–19
Araújo Neto JC, Aguiar IB, Ferreira VM (2003) Efeito da
temperatura e da luz na germinação de sementes de
Acacia polyphylla DC. Revta Bras Bot 26:249–256
Arroyo Mora JP, Sánchez-Azofeifa GA, Rivard B, Calvo J
(2005) Quantifying successional stages of tropical dry
forests using Landsat ETM?. Biotropica 37:497–507. doi:
10.1111/j.1744-7429.2005.00068.x
Begon M, Harper JL, Townsend CR (2006) Ecology: individuals, populations and communities. Blackwell,
Oxford
Castellanos AE (1991) Photosynthesis and gas exchange of
vines. In: Putz FE, Mooney HA (eds) The biology of
vines. Cambridge University Press, Cambridge, pp 181–
202
Chazdon RL (2003) Tropical forest recovery: legacies of
human impact and natural disturbances. Perspect Plant
Ecol 6:51–71. doi:10.1078/1433-8319-00042
Chazdon RL, Colwell RK, Denslow JS, Guariguata MR (1998)
Statistical methods for estimating species richness of
woody regeneration in primary and secondary rain forests
of NE Costa Rica. In: Dallmeier F, Cominsky JA (eds)
Forest biodiversity research, monitoring and modelling:
conceptual background and Old World case studies. Parthenon Publishing, Paris, pp 285–309
Clark DB, Clark DA (1990) Distribution and effects on tree
growth of lianas and woody hemiepiphytes in a Costa
Rican wet forest. J Trop Ecol 6:321–331
Colwell RK (2006) EstimateS: statistical estimation of species
richness and shared species from samples, version 8.0.
Available via http://purl.oclc.org/estimates. Accessed 12
August 2008
Colwell RK, Coddington JA (1994) Estimating terrestrial
biodiversity through extrapolation. Philos T R Soc B
345:101–118. doi:10.1098/rstb.1994.0091
Corlett R (1994) What is secondary forest? J Trop Ecol 10:
445–447
Crawley MJ (2002) Statistical computing: an introduction to
data analysis using S-Plus. Wiley, Chichester
Denslow JS, Guzman S (2000) Variation in stand structure,
light, and seedling abundance across a tropical moist
Intermediate
x
x
Late
x
x
x
x
x
x
24
33
42
forest chronosequence, Panama. J Veg Sci 11:201–212.
doi:10.2307/3236800
DeWalt SJ, Schnitzer SA, Denslow JS (2000) Density and
diversity of lianas along a chronosequence in a central
Panamanian lowland forest. J Trop Ecol 16:1–19. doi:
10.1017/S0266467400001231
Egler FE (1954) Vegetation science concepts I. Initial floristic
composition: a factor in old field vegetation development.
Vegetatio 4:412–417. doi:10.1007/BF00275587
Gentry AH (1995) Diversity and floristic composition of
Neotropical dry forests. In: Bullock SH, Mooney HA,
Medina E (eds) Seasonally dry tropical forests. Cambridge University Press, Cambridge, pp 146–194
Gentry AH, Dodson C (1987) Contribution of nontrees to
species richness of a tropical rain forest. Biotropica
19:149–156. doi:10.2307/2388737
Guariguata MR, Ostertag R (2001) Neotropical secondary forest
succession: changes in structural and functional characteristics. For Ecol Manage 148:185–206. doi:10.1016/
S0378-1127(00)00535-1
Hegarty EE, Caballé G (1991) Distribution and abundance of
vines in forest communities. In: Putz FE, Mooney HA
(eds) The biology of vines. Cambridge University Press,
Cambridge, pp 313–336
Holdridge LR (1967) Life zone ecology. Tropical Science
Center, San Jose
Holdridge LR, Grenke WC, Hatheway WH, Liang T, Tosi JA
Jr (1971) Forest environments in tropical life zones: a
pilot study. Pergamon Press, New York
Howe HF, Smallwood J (1982) Ecology of seed dispersal.
Annu Rev Ecol Syst 13:201–228. doi:10.1146/annurev.es.
13.110182.001221
IEF—Instituto Estadual de Florestas (2000) Parecer técnico
para a criação do Parque Estadual da Mata Seca. Instituto
Estadual de Florestas, Belo Horizonte
Janzen DH (1983) No park is an island: increase in interference
from outside as park size decreases. Oikos 41:402–410.
doi:10.2307/3544100
Janzen DH (1986) Tropical dry forests: the most endangered
major tropical ecosystem. In: Wilson EO (ed) Biodiversity.
National Academy Press, Washington, DC, pp 130–137
Janzen DH (1988) Management of habitat fragments in a tropical dry forest: growth. Ann Mo Bot Gard 75:105–116. doi:
10.2307/2399468
Forest Ecology
Justiniano MJ, Fredericksen TS (2000) Phenology of tree
species in Bolivian dry forests. Biotropica 32:276–281
Kalácska M, Sánchez-Azofeifa GA, Calvo-Alvarado JC,
Quesada M, Rivard B, Janzen DH (2004) Species composition, similarity and diversity in three successional
stages of a seasonally dry tropical forest. For Ecol Manag
200:227–247
Kalácska M, Sánchez-Azofeifa GA, Calvo-Alvarado JC, Rivard
B, Quesada M (2005) Effects of season and successional
stage on leaf area index and spectral vegetation indices in
three Mesoamerican tropical dry forests. Biotropica
37:486–496. doi:10.1111/j.1744-7429.2005.00067.x
Kellman MC (1970) Secondary plant succession in tropical
montane Mindanao. Australian National University,
Department of Biogeography and Geomorphology,
Canberra
Kennard DK (2002) Secondary forest succession in a tropical
dry forest: patterns of development across a 50-year
chronosequence in lowland Bolivia. J Trop Ecol 18:
53–66. doi:10.1017/S0266467402002031
Kuzee ME, Bongers F (2005) Climber abundance, diversity
and colonization in degraded forests of different ages in
Côte d’Ivoire. In: Bongers F, Parren MPE, Traoré D (eds)
Forest climbers of West Africa: diversity, ecology and
management. CABI Publishing, Wallingford, pp 67–84
Laurance WF, Pérez-Salicrup D, Delamônica P, Fearnside PM,
D’angelo S, Jerozolinski A, Pohl L, Lovejoy TE (2001)
Rain forest fragmentation and the structure of Amazonian
liana communities. Ecology 82:105–116
Lorenzi H (1992) Árvores brasileiras: manual de identificação
e cultivo de plantas arbóreas nativas do Brasil, vol 2.
Plantarum, São Paulo
Lugo AE, Gonzalez-Liboy JA, Cintron B, Dugger K (1978)
Structure, productivity and transpiration of a subtropical
dry forest in Puerto Rico. Biotropica 10:278–291. doi:
10.2307/2387680
Magurran AE (2004) Measuring biological diversity. Blackwell Science, Malden
Miles L, Newton AC, Fries RS, Ravilious C, May I, Blyth S,
Kapos V, Gordon JE (2006) A global overview of the
conservation status of tropical dry forests. J Biogeogr
33:491–505. doi:10.1111/j.1365-2699.2005.01424.x
Mizrahi A, Ramos-Prado JM, Jimenez-Osorio J (1997) Composition, structure, and management potential of secondary
dry tropical vegetation in two abandoned henequen plantations of Yucatan Mexico. For Ecol Manage 96:273–282.
doi:10.1016/S0378-1127(97)00008-X
Morellato LPC, Talora DC, Takahasi A, Bencke CC, Romera
EC, Zipparro VB (2000) Phenology of Atlantic rain forest
trees: a comparative study. Biotropica 32:811–823. doi:
10.1111/j.1744-7429.2000.tb00620.x
Murphy PG (1995) Dry forests of Central America and the
Caribbean. In: Bullock SH, Mooney HA, Medina E (eds)
Seasonally dry tropical forests. Cambridge University
Press, Cambridge, pp 9–34
Murphy PG, Lugo AE (1986) Ecology of tropical dry forests.
Annu Rev Ecol Syst 17:67–88. doi:10.1146/annurev.
es.17.110186.000435
Nabe-Nielsen J (2001) Diversity and distribution of lianas in a
neotropical rain forest, Yasunı́ National Park, Ecuador. J
Trop Ecol 17:1–19. doi:10.1017/S0266467401001018
303
Oliveira-Filho AT, Curi N, Vilela EV (1998) Effects of canopy
gaps, topography, and soils on the distribution of woody
species in a central Brazilian deciduous dry forest. Biotropica 30:362–375. doi:10.1111/j.1744-7429.1998.tb00071.x
Opler PA, Baker HG, Frankie GF (1977) Recovery of tropical
lowland forest ecosystems. In: Cairns J Jr, Dickson KL,
Herricks EE (eds) Recovery and restoration of damaged
ecosystems. University of Virginia Press, Charlottesville,
pp 379–421
Pereira IM, Andrade LA, Sampaio EVSB, Barbosa MRV
(2003) Use-history effects on structure and flora of caatinga. Biotropica 35:154–165
Pérez-Salicrup DR (2001) Effect of liana cutting on tree
regeneration in a liana forest in Amazonian Bolivia.
Ecology 82:389–396
Phillips OL, Martı́nez RV, Arroyo L, Baker TR, Killeen T,
Lewis SL, Malhi Y, Mendoza AM, Neill D, Vargas PN,
Alexiades M, Cerón C, Di Fiore A, Erwin T, Jardim A,
Palacios W, Saldias M, Vinceti B (2002) Increasing
dominance of large lianas in Amazonian forests. Nature
418:770–774. doi:10.1038/nature00926
Prentice IC, Cramer W, Harrison SP, Leemans R, Monserud
RA, Solomon AM (1992) Special paper: a global biome
model based on plant physiology and dominance, soil
properties and climate. J Biogeogr 19:117–134. doi:
10.2307/2845499
Putz FE (1984) The natural history of lianas on Barro Colorado
Island, Panama. Ecology 65:1713–1724. doi:10.2307/193
7767
Ruiz J, Fandino MC, Chazdon RL (2005) Vegetation structure,
composition, and species richness across a 56-year chronosequence of dry tropical forest on Providencia Island,
Colombia. Biotropica 37:520–530. doi:10.1111/j.17447429.2005.00070.x
Sader S, Waide RB, Lawrence WT, Joyce AT (1989) Tropical
forest biomass and successional age class relationships to
a vegetation index derived from Landsat TM data. Remote
Sens Environ 28:143–156. doi:10.1016/0034-4257(89)
90112-0
Saldarriaga JG, West DC, Tharp ML, Uhl C (1988) Long-term
chronosequence of forest succession in the Upper Rio
Negro of Colombia and Venezuela. J Ecol 76:938–958.
doi:10.2307/2260625
Salis SM, Silva MP, Matos PP, Silva JSV, Potti VJ, Pott A
(2004) Fitossociologia de remanescentes de florestas estacionais deciduais em Corumbá, Estado do Mato Grosso
do Sul, Brasil. Rev Bras Bot 27:671–684. doi:
10.1590/S0100-84042004000400008
Sampaio AB, Holl KD, Scariot A (2007) Regeneration of
seasonal deciduous forest tree species in long-used pastures in Central Brazil. Biotropica 39:655–659. doi:
10.1111/j.1744-7429.2007.00295.x
Sánchez-Azofeifa GA, Castro K, Rivard B, Kalascka M, Harriss RC (2003) Remote sensing research priorities in
tropical dry forest environments. Biotropica 35:134–142
Schnitzer SA (2005) A mechanistic explanation for global
patterns of liana abundance and distribution. Am Nat
166:262–276. doi:10.1086/431250
Schnitzer SA, Bongers F (2002) The ecology of lianas and their
role in forests. Trends Ecol Evol 17:223–230. doi:10.1016/
S0169-5347(02)02491-6
304
Schnitzer SA, Carson WP (2001) Treefall gaps and the maintenance of species diversity in a tropical forest. Ecology
82:913–919
Schnitzer SA, Dalling JW, Carson W (2000) The impact of
lianas on tree regeneration in tropical forest canopy gaps:
evidence for an alternative pathway of gap-phase regeneration. J Ecol 88:655–666. doi:10.1046/j.1365-2745.
2000.00489.x
Sevilha AC, Scariot A, Noronha SE (2004) Estado atual da
representatividade de unidades de conservação em Florestas Estacionais Deciduais no Brasil. In: Sociedade
Brasileira de Botânica (org) Biomas florestais. Editora da
Universidade Federal de Viçosa, Viçosa, pp 1–63
Silva LM, Rodrigues TJD, Aguiar IB (2002) Efeito da luz e da
temperatura na germinação de sementes de aroeira (Myracrodruon urundeuva Allemão). Rev Arvore 26:691–697
Swaine MD, Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests. Vegetatio
75:81–86. doi:10.1007/BF00044629
A.G. Van der Valk (ed.)
Teramura AH, Gold WG, Forseth IN (1991) Physiological
ecology of mesic, temperate woody lianas. In: Putz FE,
Mooney HA (eds) The biology of vines. Cambridge
University Press, Cambridge, pp 245–285
Vieira DLM, Scariot A (2006) Principles of natural regeneration of tropical dry forests for restoration. Restor Ecol
14:11–20. doi:10.1111/j.1526-100X.2006.00100.x
Vieira DLM, Sampaio AB, Scariot A, Holl K (2006) Tropical
dry forest regeneration from root suckers in Central Brazil. J Trop Ecol 22:1–5. doi:10.1017/S0266467405003135
Wright JC, Sánchez-Azofeifa GA, Portillo C, Davies D (2007)
Poverty and corruption compromises tropical forest
reserves. Ecol Appl 17:1259–1266. doi:10.1890/06-1330.1
Yodzis P (1986) Competition, mortality and community
structure. In: Diamond J, Case TJ (eds) Community
ecology. Harper and Row, New York, pp 480–491
Woody plant composition of forest layers: the importance
of environmental conditions and spatial configuration
Maya Gonzalez Æ Marc Deconchat Æ
Gérard Balent
Originally published in the journal Plant Ecology, Volume 201, No. 1, 305–318.
DOI: 10.1007/s11258-009-9572-9 Springer Science+Business Media B.V. 2009
Abstract The species–environment relationships for
woody species may vary according to the forest layers
considered. In fragmented forest, spatial configuration
may also influence forest layer composition. We
investigated the relationships between four forest
layer compositions and environmental conditions,
and spatial variables accounting for forest fragmentation, in 59 forest stands. Field and shrub layer
compositions were mainly linked to environmental
conditions, particularly to soil pH and slope aspect,
while the upper layer compositions were principally
correlated to the spatial configuration. The distance
from the forest edge was correlated with all the forest
layer compositions. Our results suggest that woody
species respond to factors acting at different spatial
and temporal scales, depending on the forest layer they
belong to. The species–environment relationship
seems to weaken from the lower to upper layer, the
upper layer being more closely linked to the spatial
configuration and probably to the past management.
This study underlines the importance of taking spatial
configuration in addition to environmental conditions
M. Gonzalez (&) M. Deconchat G. Balent
INRA, UMR 1201 DYNAFOR, Chemin de Borde Rouge,
BP 52627, 31326 Castanet-Tolosan Cedex, France
e-mail: m-gonzalez@enitab.fr
Present Address:
M. Gonzalez
ENITAB, UMR 1220 TCEM, 1 cours du Général de
Gaulle, CS 40201, 33175 Gradignan Cedex, France
into account when studying woody plant diversity for
different forest layers in stands located in deciduous
fragmented forests. Moreover, stand history seems to
have a lasting effect on woody plant composition,
particularly for the tree layer.
Keywords Coppice-with-standards
Land-use history Fragmented forest
Slope aspect Soil pH South-western France
Introduction
Woody plant species is a key biological group for
forest ecosystems since it is responsible for their
architecture (Stapanian et al. 1997), which subsequently determines many of the ecological conditions
found within forest. The factors that explain the
diversity of woody plants differ from those of non
woody species because of their larger size, their
stratification, and their longevity; the study of woody
plant diversity needs to account for these specificities.
Diversity is a scale-dependent concept (Magurran
1988); thus, the scale at which the factors are studied
needs to be defined. We know that factors acting on
plant diversity at fine scales are not the same as those
acting at larger spatial scales (Whittaker et al. 2001;
Decocq 2002; Weiher and Howe 2003). Besides,
within the possible range of spatial scales at which
we can study diversity and its determinants, it seems
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_23
305
306
to be important to study the diversity patterns at the
spatial scale at which biodiversity is usually managed
(Bestelmeyer et al. 2003), with the perspective of
preserving such diversity. Woody plant diversity and
their driving factors are likely to be affected by forest
logging interventions at the stand scale. Forest
fragments in agricultural landscapes, particularly
those owned by farmers, frequently pertain to several
owners (Guyon et al. 1996), and, as a consequence,
different management practices can be found in a
given forest fragment (De Warnaffe et al. 2006).
According to the traditional niche-based approach,
many authors have sought for species–environment
relationships to understand the determinants of community composition (Svenning and Skov 2002). These
studies commonly found a predominant effect of soil
pH on plant diversity (i.e., richness and composition)
(Brunet et al. 1997; Sagers and Lyon 1997; Diekmann
et al. 1999; Augusto et al. 2003; Borchsenius et al.
2004; Schuster and Diekmann 2005; Lenière and
Houle 2006), and also of slope aspect in forested
landscapes with dissected topography (Cantlon 1953;
Small and MacCarthy 2002; Gracia et al. 2007). These
studies usually ignored the potential effect of spatial
configuration on the observed patterns (but see de
Blois et al. 2001). However, there is now a growing
interest in taking spatial configuration into account
when studying the drivers of forest stand diversity
since recent studies in plant ecology have reported
important effects of seed dispersal limitations in forest
plant communities (Svenning and Skov 2002; McEuen
and Curran 2004). For stands located in fragmented
forests, spatial variables that account for forest
fragmentation (i.e., forest fragment area, distance
from the forest edge and isolation) may explain the
residual variation of species–environment relationships (Schuster and Diekmann 2005). At the forest
fragment scale, several studies have compared the
effects of site conditions and spatial configuration on
forest plant species richness and composition
(Dzwonko and Loster 1988, Grashof-Bokdam 1997,
Honnay et al. 1999; Butaye et al. 2001; Jacquemyn
et al. 2003). At the stand scale (i.e., spatial units within
forest fragments with uniform canopy composition,
structure, age and management), the influence of the
forest edge has been extensively studied (see Murcia
1995; Ries et al. 2004; Harper et al. 2005 for reviews
on edge effect). Few studies, however, have investigated the relationships between forest fragment area,
A.G. Van der Valk (ed.)
or isolation, and forest plant species diversity (but see
Petersen 2002 and Guirado et al. 2007).
Woody plants, given their longevity and the
important differences in size between individuals
found in the different forest layers, have been shown
to present different relationships with environmental
variables (Bratton 1975; Burnett et al. 1998; Lyon
and Gross 2005) and also spatial variable (distance
from edge) (Ranney et al. 1981; Gehlhausen et al.
2000), depending on the forest layers considered.
Moreover the composition and the amount of cover in
the understory are also dependent on the identity of
the trees present in the overstory (Augusto et al.
2003; Legare et al. 2002).
The purpose of our article is to examine the factors
determining woody plant composition of stands
located in forest fragments. We investigated both
species–environment relationships and relationships
between woody species composition and spatial
variables in 59 stands located in fragmented forests
for four forest layers. We also investigated the
relationship between the compositions of the four
forest layers.
Methods
Study area
Field research was conducted in a 632-km2 area
located in the ‘‘Coteaux de Gascogne,’’ in southwestern France (43130 N, 0520 E). The climate is
mild, with a mean annual temperature of 12.5C and
mean annual precipitation of 750 mm. This hilly area
(200–400 m a.s.l.) has short and steep slopes along the
river valleys descending from the Pyrenees Mountains. The forests have three types of soil: superficial
calcareous soils (rendosols or calcosols), brown acid
soils (brunisols), and brown washed soils (neoluvisols)
(Duchaufour 1983). Forests cover 15% of the area
with numerous small forested fragments ranging from
\1 to 50 ha, and a few larger forests (max = 600 ha)
(Balent and Courtiade 1992). The vegetation shows
both Atlantic and Mediterranean influences on medioEuropean type flora, where oaks (Quercus robur L.,
Quercus pubescens Willd., Quercus petraea Liebl.)
are the main tree species, often in combination with
hornbeam (Carpinus betulus L.), cherry (Prunus avium L.), wild service tree (Sorbus torminalis (L.)
Forest Ecology
Crantz), chestnut (Castanea sativa Mill.), and field
maple (Acer campestre L.). Most of these forest
fragments are owned or managed by farmers who
produce firewood, and to a lesser extent, timber.
Stands are mostly coppice, with 30–50 oak standard
trees per ha (Deconchat and Balent 2001) retained to
produce timber. Coppice trees are cut every 30 years
and standard trees are cut approximately every
60 years. They are regenerated by natural seeding
while coppice is mainly regenerated by resprouting.
Vegetation sampling
We selected 39 forest fragments in the study area, i.e.,
woodlots of various sizes, ranging from 0.32 to 693 ha
(median ± SD = 5.75 ± 136). We chose a total of 59
mature stands within these 39 forest fragments, after a
thorough survey of the area, with mature stand
selection based on two criteria: stands with percent
of canopy openness\20% and a coppice layer close to
harvest age (i.e., about 30 years). We thus avoided,
recently, cut areas or young stands with thin canopies.
We controlled the percentage of canopy openness for a
subset of 43 stands, using hemispherical photographs
taken at the center of the plot, at 1.0 m above ground,
and analysed with ‘Gap Light Analyser’ software
(Frazer et al. 1999) (median = 14.6%, SD = 1.57),
thus confirming our visual selection of the target
stands. Within a given forest fragment, the number of
stands selected was determined according to the forest
fragment heterogeneity (i.e., the number of different
stands in terms of their composition and structure)
observed during the survey of the entire forest
fragment. The higher the heterogeneity of the forest
fragment, the higher the number of stands inventoried
in the forest fragment. Our sampling procedure was, in
fact, designed so as to cover the maximum range of
variation in composition in stands with closed canopies and a maximum range of distances from the forest
fragment edge. These selection criteria led to the
following sampling scheme: 26 forest fragments with
one stand inventoried, 9 forest fragments with two
stands inventoried, 2 forest fragments with three
stands inventoried, 1 forest fragment with four stands
inventoried, and 1 forest fragment with five stands
inventoried.
For each stand, we inventoried all woody individuals in a 400-m2 square plot (Harcombe et al. 2002),
as this area corresponds to the optimal plot size
307
established in these forests for woody species. The
inventories were made for four forest layers: field
layer (\1.30 m), shrub (1.30 B height \ 7 m), coppice (7 B height \ 15 m), and canopy (C15 m
height). We added the coppice layer to the three
strata (tree, shrub, and field layer) commonly used in
forest studies since the type of sylvicultural management used in the forest studied (i.e., coppicing)
produces a supplementary layer (Leroyer 2002). Each
individual was identified to species level and assigned
to one forest layer according to its height. We thus
obtained the abundance of all the species found
during the inventories in each layer. For species
which were coppice, we counted the number of
stumps. Nomenclature follows Flora Europaea (Tutin et al. 1983). Individuals were also classified
according to their d.b.h. (diameter at breast height,
i.e., 1.3 m), allowing for the calculation of stand
basal area for each plot.
Explanatory variables
Environmental conditions
For each plot, we recorded the slope inclination (%)
using an inclinometer and the slope aspect using a
compass. The aspects were grouped into two categories as follows: ‘‘north-facing’’ (North, Northeast,
Northwest, and one plot facing East; n = 42),
corresponding to cool and wet climatic conditions
and ‘‘south-facing’’ (South, Southeast and Southwest;
n = 15), corresponding to warm and dry climatic
conditions (Bratton 1975; Gonin 1993; Gracia et al.
2007). Two stands located in the valley had no
aspect.
We collected one soil core sample at the centre of
the square plot, in the A horizon, with an auger (8 cm
in diameter) for a subset of 30 plots, selected so as to
cover a wide range of different types of stands in
terms of the composition of dominant species,
structure, and overall species richness. These samples
were analysed for pH-H2O, total C and total N, C/N
ratio and available phosphorus content (Duchaufour
and Bonneau 1959).
Spatial configuration
For each stand, we measured the distance from the
nearest forest edge from the centre of the square plot
308
(in meters) using a tape measure. Hereafter, the
distance from the nearest forest edge is referred to as
‘‘distance from the forest edge.’’
All the 39 forest fragments (containing the plots
inventoried) were digitised in ArcView v.3.2 (ESRI
1999) from aerial photographs (scale 1/25,000, year
2002, distributed by the French National Geographic
Institute). Forest fragment area (in ha) was calculated
using Patch Analyst v.3 (Elkie et al. 1999) extension
of ArcView v.3.2 (ESRI 1999).
All other woodlots present within a radius of
1000 m around the perimeter of each of the 39 forest
fragments (containing the plots) were also digitised
based on the same aerial photographs. Using Patch
Analyst v.3 (Elkie et al. 1999) extension of
ArcView v.3.2 (ESRI 1999), the following four
variables were calculated to describe landscape
context: D = distance to the nearest woodlot (in
meters; measured from edge to edge) and Cov100,
Cov500, Cov1000 = the percentage of wood cover
within radii of 100, 500, and 1000 m, respectively,
around the perimeter of the target forest fragment
(containing the plots inventoried).
Data analysis
We first performed a Correspondence Analysis (CA)
(Benzécri 1973) for each of the four forest layer
matrices: plots 9 species. The plot scores on the two
first axes were used as species composition variables
(Okland et al. 2003; Bennie et al. 2006). We then
analysed the relationship between the composition of
each forest layer (plots scores on the two first
principal axes of the CA) and environmental (canopy
openness only for the two lowest forest layers, slope
inclination, pH, C and N contents, C/N and phosphorus content) and spatial (distance from the forest
edge, forest fragment area, distance to the nearest
woodlot and the three wood covers) variables using
Spearman rank correlation analyses. We tested the
differences in forest layer composition (plots scores
on the two first principal axes of the CA) between
slope aspect categories using the Mann–Whitney U
test. We then used GLM analysis, after transforming
the variables (to achieve normality tested with a onesample Kolmogorov–Smirnof test), to determine for
each forest layer which variables (environmental and
spatial) were the most important to explain species
A.G. Van der Valk (ed.)
composition, using a stepwise forward selection
procedure. We also investigated whether the composition of the forest layers was similar using Spearman
rank correlation analysis between plots’ scores of the
CA performed for each forest layer.
Correspondence analyses were performed using
the ade4 package developed by Thioulouse et al.
(1997) in the R free software (Ihaka and Gentleman
1996). Spearman rank correlations analyses, Mann–
Whitney U tests, and GLM analyses were carried out
in SYSTAT v 9.0 (SYSTAT 1999).
Results
Species richness and composition of the forest
layers
We recorded a total of 34 woody species in this
survey (Appendix): 30 species in the field layer, 31
species in the shrub layer, 16 species in the coppice
layer, and 14 species in the canopy layer. The shrub
and the field layers included saplings of tree species
that occurred also in canopy and coppice layers, with
a tendency of shade-tolerant tree species to be more
abundant in the understory layers compared to more
light-demanding species (Fig. 1).
The total variance in the species data explained by
the two first axes of the CA performed on the four
forest layer data matrices ranged from 12.24 (shrub
layer) to 20.00% (canopy layer) for the first axis, and
from 9.74 (field layer) to 16.38% (canopy layer) for
the second axis. The first axes of the three lower
forest layers (i.e., field, shrub, and coppice) presented
the same pattern of species distribution (Fig. 2), with
Fagus sylvatica L., Ilex aquifolium L., C. betulus L.,
and Q. petraea Liebl., at one end of the axis, and
Q. pubescens Willd., Q. robur L., and Fraxinus excelsior L. and S. torminalis (L.) Crantz, at
the opposite end of the axis. For the canopy layer, the
first axis separated a group of six plots containing
Populus tremula L. and A. campestre L. from the rest
of the plots inventoried. The second axis presented
some similarities with the first axis of the CA
performed on the other forest layers regarding the
location of F. sylvatica and C. betulus at one end,
and of Q. pubescens and F. excelsior at the other end
of the axis.
Forest Ecology
Fig. 1 Frequencies
(N = 59) of the tree species
(i.e., species able to be
present in the canopy layer)
in the four forest layers,
from field to canopy layer.
Species are grouped
according to their shade
tolerance (see Appendix for
complete names of species
and shade tolerance source)
309
60
50
Field
Shrub
Coppice
Canopy
40
30
20
10
Light-demanding
Fig. 2 Factorial maps
(axes 1 and 2) of the
Correspondence Analysis
performed: a field layer
matrix (59 plots 9 30
species), inertia axis
1 = 12.84% and axis
2 = 9.74%; b shrub layer
matrix (59 plots 9 31
species), inertia axis
1 = 12.24% and axis
2 = 10.82%; c coppice
layer matrix (59 plots 9 16
species), inertia axis
1 = 14.26% and axis
2 = 12.83%; and d canopy
layer matrix (59 plots 9 14
species), inertia axis
1 = 20% and axis
2 = 16.38%. The species
labels correspond to the first
three letters of the genus
plus the first three letters of
the species name (for
example, POPTRE refers to
P. tremula) and are given in
Appendix. The numbers
correspond to the plots
labels
Light-tolerant
Shade-tolerant
FAGSYL
CARBET
PRUAVI
QUEPET
ACEPLA
CASSAT
FRAEXC
ACECAM
ULMMIN
SORTOR
QUEROB
QUEPUB
POPTRE
0
Shadedemanding
310
A.G. Van der Valk (ed.)
Relationships with the environmental variables
correlated to C and N contents and positively to soil
C/N ratio. Plot scores on the second axis for the shrub
layer were positively and significantly correlated to
both C and N contents but not to C/N ratio. No
relationships were found with phosphorus content for
any of the four forest layers.
The composition of the stands inventoried (plot
scores in the first axis of the CA) was particularly
different between the two categories of slope aspect
(‘‘north-facing’’ vs. ‘‘south-facing’’ slope) for the two
lower forest layers studied (Table 1). The differences
were less significant for the coppice layer (axis 1) and
not significant for the canopy layer (axes 1 and 2:
same results). Forest layer compositions were not
significantly correlated to the slope inclination, and
the two understory layers (field and shrub) were not
significantly correlated to the canopy openness
(Table 1). Plot scores on the first axis for the field
layer and plot scores on the second axis for the shrub
layer were significantly and positively correlated with
soil pH, and plot scores on the first axis for the
coppice layer was significantly and negatively correlated with soil pH (Table 2). Plot scores on the first of
the coppice layer were significantly and negatively
Relationships with the spatial variables
The spatial variable showing the strongest relationship with forest layer composition was the distance
from the forest edge (Table 3). The correlation was
significant for the four forest layers investigated.
Concomitantly, forest layer compositions were correlated significantly with the forest fragment area.
The distance from the forest edge was positively
correlated with the forest fragment area in our
sampling scheme (R2 = 0.670; P \ 0.001). For the
canopy layer, plot scores on the first axis showed
several significant relationships with the spatial
canopy layers) with the Mann–Whitney U testa; N = 57b.
Spearman rank correlations between plots scores and slope
inclination (in %; N = 59) and canopy opennessc (in %;
N = 43) are also given
Table 1 Differences in composition (plots scores on the two
first axes of the correspondence analysis) between the two
types of aspects (‘‘north-facing’’ and ‘‘south-facing’’ slopes)
investigated for the four forest layers (field, shrub, coppice, and
Field layer
Axis 1
Shrub layer
Axis 2
Axis 1
Coppice layer
Axis 2
Axis 1
Slope aspect
10.289***
0.579
10.878***
3.154
Slope inclination
-0.2
0.116
-0.05
0.107
0.086
0.06
-0.104
Canopy openness
0.087
Canopy layer
Axis 2
Axis 2
1.202
3.831
3.83
-0.088
0.023
0.048
–
–
5.256*
-0.055
Axis 1
–
–
a
Chi-square approximation are given with 1 d.f., and level of significance: * P \ 0.05, ** P \ 0.01, *** P \ 0.001
b
Two stands located in the valley with no aspect are not considered in this analysis
c
For the canopy openness only relationships with field and shrub layers are investigated as they are the layers susceptible to be
linked to this variable
the soil variables: pH-H2O, Carbon (C) and Nitrogen (N)
content, C/N ratio and phosphorus content for the A horizon
(N = 30)
Table 2 Spearman rank correlations between plot scores on
the two first axes of the Correspondence Analysis performed on
the matrices of the field, shrub, coppice, and canopy layers and
Field layer
Axis 1
pH
C
N
C/N
Phosphorus
0.529**
0.166
Shrub layer
Axis 2
Axis 1
0.101
0.202
0.107
-0.051
Coppice layer
Axis 2
Canopy layer
Axis 1
Axis 2
Axis 1
Axis 2
0.345*
0.552**
20.491**
20.337*
-0.205
0.126
-0.117
-0.305
0.251
-0.035
0.560**
20.338*
-0.052
-0.285
0.012
0.076
-0.259
-0.034
-0.006
0.175
0.218
-0.114
-0.167
0.080
0.075
-0.189
0.317*
0.016
0.023
-0.135
-0.151
0.247
Bold type indicates significant relationships, with *: P \ 0.05 and **: P \ 0.01
0.449**
-0.235
Forest Ecology
311
Table 3 Spearman rank correlations between plot scores on
the first two axes of the Correspondence Analysis performed
for the field, shrub, coppice and canopy layers and the spatial
variables: forest fragment area (FF area, in ha), distance to the
Field layer
Axis 1
FF area
Shrub layer
Axis 2
0.258*
20.282*
0.022
D
Cov100
Cov500
nearest woodlot (D, in meters), percentage of wood cover
within three radii: 100, 500, and 1000 m (Cov100, Cov500,
and Cov1000, respectively) and distance from the forest edge
(DFE, in meters) (N = 59)
20.243*
0.079
-0.048
0.269
0.142
Cov1000
-0.032
0.072
DFE
20.292*
0.143
Coppice layer
Axis 1
Axis 2
Axis 1
20.301*
20.313**
0.008
0.280*
0.430**
20.234*
Canopy layer
Axis 2
Axis 1
0.045
Axis 2
0.146
20.357**
-0.184
20.230*
0.155
-0.037
0.061
-0.118
-0.142
0.177
0.030
0.153
0.078
0.220*
0.258*
0.168
-0.122
0.009
0.327**
0.310**
0.003
20.406**
20.294*
0.442**
0.029
0.228*
20.431**
-0.009
0.009
Bold type indicates significant relationships, with * P \ 0.05 and ** P \ 0.01
variables accounting for landscape context (i.e.,
distance to the nearest woodlot and wood covers
within the different radii) (Table 3).
Relative effects of environmental versus spatial
variables on forest layer compositions
forest edge (Fig. 5). Finally, the composition of the
canopy layer was best explained by the percentage of
wood cover in the radii of 1000 m around the forest
fragment containing the stand (Fig. 6).
Relationships between the compositions
of the four forest layers
The composition of the field layer was principally
linked to the soil pH (Fig. 3), while the shrub layer
composition was more influenced by the distance
from the forest edge and by slope aspect (Fig. 4 and
Table 4). The coppice layer composition was both
influenced by the soil pH and the distance from the
All the scores of the stands on the first axis of the CA
performed on the three dominated layers’ matrix, i.e.,
the coppice, shrub, and field layers, had a significant
2
1
Plot scores axis 1
1
Plot scores axis 1
0
-1
-2
0
-1
-2
-3
-3
-4
4
5
6
7
8
9
pH
Fig. 3 Relationship between the plot scores on axis 1 of the
CA performed on the field layer matrix (height \ 1.30 m) and
soil pH. The line was obtained with a logarithmic smoothing
method
0
10
0
20
0 0 0
30 40 50
Distance from the forest edge
Fig. 4 Relationship between the plot scores on axis 1 of the CA
performed on the shrub layer matrix (1.30 \ height \ 7 m) and
distance from the forest edge (in meters, in a logarithmic scale).
The line was obtained with a logarithmic smoothing method
312
Table 4 Final model of the
GLM analysis obtained after
a forward stepwise procedure
performed for the four forest
layers with plots scores on
axis 1 of the CA used as the
dependent variable
A.G. Van der Valk (ed.)
Forest layer
Effect
Std Coef
P(2Tail)
N
0.561
0.001
30
-0.348
0.008
57
0.433
0.001
Field
F = 12.843, adjusted r2 = 0.290, P = 0.001
pH
Shrub
F = 12.923, multiple r2 = 0.324, P = 0.0001
DFE
Slope aspect
Coppice
F = 11.319, adjusted r2 = 0.416, P = 0.0001
pH
-0.483
0.002
DFE
0.489
0.002
Cover1000
0.328
0.011
30
Canopy
Fig. 5 Relationship
between the plot scores on
axis 1 of the CA performed
on the coppice layer matrix
(7 \ height \ 15 m); and
left: the distance from the
forest edge (in meters, in a
logarithmic scale) and right:
the soil pH. The lines were
obtained with a logarithmic
smoothing method
Plot scores axis 1
F = 6.893, adjusted r2 = 0.092, P = 0.011
2
2
1
1
0
0
-1
-1
-2
4
5
6
7
pH
relationship with the scores of the stands on the
second axis of the CA performed on the canopy layer
(Table 5). The plot’s scores on the first axis of the
canopy layer were less correlated with the plot’s
scores of the different forest layers investigated.
Discussion
Environmental conditions
The composition of the forest layers was significantly
correlated to the environmental variables, and particularly to soil pH and slope aspect. The prevailing
effect of soil pH on vegetation composition found
here is in accordance with other forest studies (Brunet
et al. 1997; Sagers and Lyon 1997; Augusto et al.
2003; Borchsenius et al. 2004; Schuster and Diekmann 2005; Lenière and Houle 2006). As is the case
in other hilly landscapes, we also found a significant
8
9
-2
0
10
59
0 0 0
0
20 30 40 50
Distance from the nearest forest edge
relationship between slope aspect (Cantlon 1953;
Burnett et al. 1998; Gracia et al. 2007) and stand
composition. Stand composition differences between
slopes aspects decreased as the height of the layer
investigated increased, as previously reported by
Cantlon (1953). The pattern for the relationship
between species and soil pH across the different
layers investigated was less clear. This environmental
variable had a significant relationship with field,
shrub, and coppice forest layer composition but not
with the canopy layer composition. In an old forest,
Borchsenius et al. (2004) found a relationship with
soil pH for both the field and the tree layer floristic
gradients.
These observed patterns of species–environment
relationships are in accordance with previous works
in which it was reported that the strength of the
species–environment relationship decreased from
field to canopy layer (Stohlgren et al. 1998; Collins
and Carson 2004). The reduced height and less
Forest Ecology
313
3
Plot scores axis 1
2
1
0
-1
0
10
20
30
40
Cover1000
Fig. 6 Relationship between the plot scores on axis 1 of the
CA
performed
on
the
canopy
layer
matrix
(7 \ height \ 15 m) and the percentage of wood cover in a
radii of 1000 m around the forest fragment containing the plot.
The line was obtained with a quadratic smoothing method
extensive root systems of the field layer could make
them more sensitive to local environmental differences (finer grain) that may not influence trees
(Sagers and Lyon 1997; Burnett et al. 1998; de Blois
et al. 2001). Moreover, such differences between
forest layer responses to environmental conditions
could be linked to the vertical changes of physical
parameters from field to canopy level, such as
humidity or temperature (Cantlon 1953).
However, given the importance of human activities in these forests (De Warnaffe et al. 2006), the
canopy layer may also reflect the influence of past or
current management practices that can disrupt the
correspondence between canopy composition and
environmental conditions (Decocq 2000; de Blois
et al. 2001). For the canopy layer, the composition
described by plot scores on axis 1 (CA of the canopy
layer) differs from those found for the three understory layers. Two species, A. campestre and
P. tremula, considerably contribute to the first axis.
These two species are present in stands characterized
by a low basal area (correlation between plot scores
on axis 1 and stand basal area: rs = -0.413,
P \ 0.001). The low basal area of these stands may
indicate that they are young stands (from a successional point of view). By investigating their history,
we effectively found that these stands were temporarily cleared for agriculture between 1860 and 1998
(Arrignon 2003). The trees (Acer and Populus) now
present in the canopy had therefore established
themselves in a previously unforested open area.
These conditions no longer exist, which could explain
why we did not find the same floristic gradient on
axis 1 for the understory layers. Decocq (2000) also
observed that pioneer species (such as P. tremula)
were present in the arborescent layer but not in the
understory, probably due to the young age of the
forests he studied. He hypothesized that since the
canopy was now closed, the shaded environment in
the understory prevented the regeneration of these
pioneer species. We observed the same pattern for
pioneer light-demanding species like oaks in our
stands with little regeneration and the contrary for
shade-demanding species like Carpinus or Fagus
more represented as young stages than as adults.
Orwig and Abrams (1994) have shown that the
majority of the adult oaks present in the stands they
studied had established themselves from the mid of
Table 5 Spearman correlations between the plots scores on the two first axes obtained from the CA performed on the four forest
layers matrix (N = 59)
Canopy
Axis 1
Coppice layer
Shrub layer
Field layer
Coppice layer
Axis 2
Axis 1
Shrub layer
Axis 2
Axis 1
0.482**
Axis 2
0.127
Axis 1
20.284*
Axis 2
20.417**
0.206
20.426**
-0.099
Axis 1
20.360**
0.458**
20.520**
0.077
Axis 2
0.181
20.231**
0.276**
0.157
20.677**
Axis 1
Axis 2
1.000
0.113
-0.070
1.000
0.580**
20.679**
0.419**
Significant relationships are indicated in bold characters with * P \ 0.05 and ** P \ 0.01
1.000
0.204
1.000
0.506**
0.278*
-0.190
0.013
314
the 1800 to the start of the 1900, during a period of
repeated cutting. They also observed a low rate of
regeneration of these oaks in the understory layers
probably due to a reduction in the frequency of fires
and of cuttings in their study area. Several works
have shown that land-use history can have long-term
effects on woody species composition (de Blois et al.
2001; Bellemare et al. 2002; Onaindia et al. 2004). In
our study area, previous works (Guyon et al. 1996;
De Warnaffe et al. 2006) have shown a lengthening
of cutting regimes which could also explain the
distribution of the tree species in the upper and
understory layers according to their shade tolerance.
Spatial variables
Forest fragment area and distance from the forest
edge were significantly correlated with stand composition for all the forest layers investigated. One
limitation of our study is that the distance from the
forest edge and the forest fragment area are correlated
in our sampling scheme. The opposition in the
factorial map of the localization of C. betulus,
F. sylvatica, and Q. petraea corresponding to the
plots located far from the forest edge, and the
localisation of Q. pubescens and F. excelsior corresponding to the plots located near the forest edge, are
consistent with the observations made by Ranney
et al. (1981) and Matlack (1993), who found lightdemanding species at the edges, as opposed to shadedemanding species in forest interiors. The relationships were stronger for the dominant layers since the
composition of the field layer was less influenced by
the distance from the forest edge than by environmental conditions (soil pH and slope aspect), as
reported by McDonald and Urban (2006). De Blois
et al. (2001) found that the edge had a significant
effect on herbaceous and shrub compositions but not
on tree composition. De Blois et al. (2001) hypothesized that the absence of relationship for the tree
layer was linked to the past disturbances and that
trees had a lower response rate to edge effect, leading
more to changes in density than in composition for
this forest layer (Murcia 1995). In our forest
fragments, forest edges are more frequently logged
than forest interiors (De Warnaffe et al. 2006). This
may explain the discrepancies between our results
and those of de Blois et al. (2001). It was shown in
the study area that the repeated coppicing had an
A.G. Van der Valk (ed.)
adverse effect on F. sylvatica and Q. petraea, to the
advantage of more pioneer-type species such as
Q. pubescens and Q. robur, for example (Gonin
1993). This has also been found in other forests
(Orwig and Abrams 1994; Decocq et al. 2004;
Onaindia et al. 2004). Thus, even if forest owners,
mainly farmers, do not plant trees (De Warnaffe et al.
2006), they can still modify woody species composition through their management practices and
particularly through cutting frequency and spatial
distribution of cuttings relative to the distance from
the edge. The variation of species composition
according to the distance from the edge for the four
layers in our stands may therefore be related to both a
microclimatic (e.g., temperature and/or humidity) or
a biotic gradient (e.g., seedling predation) linked to
edge effect (Matlack 1993; Gehlhausen et al. 2000;
Harper et al. 2005) and/or to the differences in
disturbance frequency between edge and forest
interior (Palik and Murphy 1990; Kupfer and Runkle
2003). Further work is needed to evaluate these nonexclusive hypotheses.
Landscape context (i.e., wood cover) accounted in
large part for the composition of the dominant layers
and particularly for the canopy layer (plot scores on
axis 1). Petersen (2002) also found a correlation
between tree composition and distance to the nearest
forest source. In our study, the more significant
correlation was found with the wood cover in a buffer
zone with a radius of 1000 m. This distance corresponds to the limiting dispersal distance of 1000 m
found by Greene and Johnson (1995) for Acer negundo, a wind-dispersed species, like A. campestre
and P. tremula, the two main species contributing to
the first axis of the canopy CA. The minor role played
by landscape context for the understory layers is
consistent with the results of Guirado et al. (2007),
who found that woodlot size and connectivity played
a minor role on forest species richness and composition for understory layers, compared to plot level
factors.
Important relationships between the forest layer
compositions
Overstory and understory compositions were strongly
correlated when investigated by using plot scores in
the CA, as also found by Sagers and Lyon (1997) for
the plots scores of the first axis for the three strata
Forest Ecology
315
they investigated (including the tree strata). However,
in our study we found that the principal floristic
gradient of the canopy layer diverged from the
gradient of the other forest layers.
The relationships found between the floristic gradients obtained for each forest layer suggest that they
are depending on some common underlying factors.
For the canopy (axis 2), the coppice and the shrub and
to a lesser extent for the field layer, they probably
have, in common, their response to the distance from
the forest edge. The field and coppice layers seem to
respond also concomitantly to the soil pH gradient.
And finally, field and shrub layers probably have in
common their differences in composition between the
two main slope aspects. However, these relationships
do not mean that composition of the understory
changes across environmental gradients at the same
rate of the composition of overstory, as previously
underlined by several authors (Bratton 1975; McCune
and Antos 1981).
Woody species sub-communities corresponding to
the different forest layers are different compartments
of the total woody plant community that do not
respond to the same ecological factors, or not to the
same extent. The lower forest layers were more
related to the environmental variables than overstory
strata. For the canopy layer, besides the influence of
spatial variables, which explained a little part of
variation of the composition, we suspect a potential
masking effect exerted by logging activities and
forest history that could explain why the plots scores
on the first axis of the CA do not relate to
environmental variables. This work underline the
interest of considering separately the different forest
layers when studying woody species diversity and its
driving factors in forests, as previously suggested by
other authors (Bratton 1975; Lyon and Gross 2005).
Further work with variables accounting for logging
activities and forest history is needed to improve our
understanding of the driving factors of forest layers
composition in these managed fragmented forests.
Acknowledgements The authors would like to thank L.
Raison, for his help with the field inventories, S. Ladet, for her
work on GIS for spatial variables, and G. Wagman, for revising
the English. They are also very grateful to M.R. Bakker, L.
Augusto, and A. Cabanettes for their valuable comments on an
earlier version of the manuscript. They would also like to thank
O. Honnay and two anonymous reviewers for useful comments
that helped to improve the manuscript. This work received
financial support from the Midi-Pyrénées Region (CCRRDT
Research and Technology Programme) and from the CNRS
(Zone Atelier national programme). M. Gonzalez was funded
by the French Ministry for Research and Higher Education.
Appendix
Table 6 List of the species inventoried and their frequency in the four forest layers investigated: (field (H \ 1.30 m), shrub
(1.30 \ H\7 m), coppice (7 \ H\ 15 m) and canopy (H [ 15 m)) (N = 59 plots)
Species
Acronym
Life history traits
Shade tolerance
Forest layer
Dispersal
Field
Shrub
Coppice
Canopy
Acer campestre L.
ACECAM
Light tolerant
Wind
19
20
17
6
Acer platanoides L.
ACEPLA
Shade tolerant
Wind
2
2
1
3
Acer pseudoplatanus L.
ACEPSE
Shade tolerant
Wind
0
1
0
0
Betula pendula Roth
BETPEN
Light demanding
Wind
0
0
0
1
Carpinus betulus L.
CARBET
Shade demanding
Wind
26
34
34
17
Castanea sativa Mill.
CASSAT
Light tolerant
Animals
8
10
9
10
Corylus avellana L.
CORAVE
Shade demanding
Animals
24
27
1
0
Cornus sanguinea L.
CORSAN
Light tolerant
Birds
40
34
0
0
Crataegus laevigata D.C.
CRALAE
Light tolerant
Birds
9
17
0
0
Crataegus monogyna Jacq.
CRAMON
Light tolerant
Birds
51
52
3
0
Cytisus scoparius L.
CYTSCO
Light demanding
Animals
4
4
0
0
Erica vagans L.
ERIVAG
Light tolerant
Wind
1
0
0
0
Fagus sylvatica L.
FAGSYL
Shade demanding
Animals
4
7
3
5
316
A.G. Van der Valk (ed.)
Table 6 continued
Species
Acronym
Life history traits
Shade tolerance
Forest layer
Dispersal
Field
Shrub
Coppice
Canopy
Frangula dodonei Ard.
FRADOD
Light tolerant
Birds
2
2
0
0
Fraxinus excelsior L.
FRAEXC
Light tolerant
Wind
13
12
5
4
Ilex aquifolium L.
ILEAQU
Shade demanding
Birds
7
7
0
0
Juniperus communis L.
JUNCOM
Light demanding
Birds
10
9
0
0
Lonicera xylosteum L.
LONXYL
Light tolerant
Birds
30
26
0
0
Malus sylvestris Mill.
MALSYL
Light tolerant
Animals
1
11
0
0
Mespilus germanica L.
MESGER
Light tolerant
Animals
4
4
0
0
Populus nigra L.
POPNIG
Light demanding
Wind
0
0
1
0
Populus tremula L.
POPTRE
Light demanding
Wind
4
2
4
6
Prunus avium L.
PRUAVI
Shade demanding
Animals
47
42
22
21
Prunus spinosa L.
PRUSPI
Light tolerant
Birds
37
27
0
0
Pyrus pyraster L.
PYRPYR
Light tolerant
Animals
0
1
0
0
Quercus petraea Liebl.
QUEPET
Shade tolerant
Animals
8
10
15
30
Quercus pubescens Willd.
QUEPUB
Light demanding
Animals
4
12
22
41
Quercus robur L.
Rosa canina L.
QUEROB
ROSCAN
Light demanding
Light demanding
Animals
Birds
7
1
9
7
13
0
45
0
Sorbus domestica L.
SORDOM
Light tolerant
Animals
Sorbus torminalis (L.) Crantz
SORTOR
Light demanding
Birds
1
1
0
0
42
54
37
7
Ulex europaeus L.
ULEEUR
Light demanding
Animals
Ulmus minor Mill.
ULMMIN
Light demanding
Wind
1
1
0
0
13
10
3
Viburnum lantana L.
VIBLAN
Light tolerant
Birds
15
1
6
0
0
After each species name, the acronym used in the factorial map and two life history traits, shade tolerance and dispersal mode, are
given(according to Rameau et al. 1989)
References
Arrignon F (2003) Importance de l’histoire et de la topographie
sur la répartition spatiale de la flore dans un bois fragmenté du Sud-Ouest de la France. Dissertation, University
of Toulouse
Augusto L, Dupouey JL, Ranger J (2003) Effects of tree species on understory vegetation and environmental
conditions in temperate forests. Ann Sci 60:823–831. doi:
10.1051/forest:2003077
Balent G, Courtiade B (1992) Modelling bird communities/
landscape patterns relationships in a rural area of southwestern France. Landscape Ecol 6:195–211. doi:10.1007/
BF00130031
Bellemare J, Motzkin G, Foster DR (2002) Legacies of the
agricultural past in the forested present: an assessment of
historical land-use effects on rich mesic forests. J Biogeogr
29:1401–1420. doi:10.1046/j.1365-2699.2002.00762.x
Bennie J, Hill MO, Baxter R et al (2006) Influence of slope and
aspect on long-term vegetation change in British chalk
grasslands. J Ecol 94:355–368. doi:10.1111/j.1365-2745.
2006.01104.x
Benzécri JP (1973) L’analyse des données. II. L’analyse des
Correspondances. Dunod, Paris
Bestelmeyer BT, Miller JR, Wiens JA (2003) Applying species
diversity theory to land management. Ecol Appl 13:
1750–1761. doi:10.1890/02-5089
Borchsenius F, Nielsen PK, Lawesson JE (2004) Vegetation
structure and diversity of an ancient temperate deciduous
forest in SW Denmark. Plant Ecol 175:121–135. doi:
10.1023/B:VEGE.0000048095.29961.c5
Bratton SP (1975) A comparison of the beta diversity functions
of the overstory and herbaceous understory of a deciduous
forest. Bull Torrey Bot Club 102:55–60. doi:10.2307/248
4413
Brunet J, Falkengren-Grerup U, Rühling A et al (1997)
Regional differences in floristic change in south Swedish
oak forests as related to soil chemistry and land use. J Veg
Sci 8:329–336. doi:10.2307/3237321
Burnett MR, August PV, Brown JH et al (1998) The influence
of geomorphological heterogeneity on biodiversity. I. A
patch-scale perspective. Conserv Biol 12:363–370. doi:
10.1046/j.1523-1739.1998.96238.x
Butaye J, Jacquemyn H, Honnay O et al (2001) The species
pool concept applied to forests in a fragmented landscape:
dispersal limitation versus habitat limitation. J Veg Sci
13:27–34. doi:10.1658/1100-9233(2002)013[0027:TSPC
AT]2.0.CO;2
Forest Ecology
Cantlon JE (1953) Vegetation and microclimates on north and
south slopes of Cushetunk Mountain, New Jersey. Ecol
Monogr 23:241–270. doi:10.2307/1943593
Collins RJ, Carson WP (2004) The effects of environment and
life stage on Quercus abundance in the eastern deciduous
forest, USA: are sapling densities most responsive to
environmental gradients? For Ecol Manage 201:241–258.
doi:10.1016/j.foreco.2004.06.023
de Blois S, Domon G, Bouchard A (2001) Environmental,
historical, and contextual determinants of vegetation
cover: a landscape perspective. Landscape Ecol 16:421–
436. doi:10.1023/A:1017548003345
De Warnaffe GD, Deconchat M, Ladet S et al (2006) Variability of cutting regimes in small private woodlots of
south-western France. Ann Sci 63:915–927. doi:10.1051/
forest:2006075
Decocq G (2000) The masking effect of silviculture on substrate-induced plant diversity in oak-hornbeam forests
from northern France. Biodivers Conserv 9:1467–1491.
doi:10.1023/A:1008919027221
Decocq G (2002) Patterns of plant species and community
diversity at different organizational levels in a forested
riparian landscape. J Veg Sci 13:91–106. doi:10.1658/
1100-9233(2002)013[0091:POPSAC]2.0.CO;2
Decocq G, Aubert M, Dupont F et al (2004) Plant diversity in a
managed temperate deciduous forest: understorey
response to two silvicultural systems. J Appl Ecol 41:
1065–1079. doi:10.1111/j.0021-8901.2004.00960.x
Deconchat M, Balent G (2001) Effets des perturbations du sol
et de la mise en lumière occasionnées par l’exploitation
forestière sur la flore à une échelle fine. Ann Sci 58:
315–328. doi:10.1051/forest:2001129
Diekmann M, Eilertsen O, Fremstad E et al (1999) Beech
forest communities in the Nordic countries—a multivariate analysis. Plant Ecol 140:203–220. doi:10.1023/A:1009
768225798
Duchaufour P (1983) Pédologie. Tome I: Pédogénèse et classification, 2nd edn. Masson, Paris
Duchaufour P, Bonneau M (1959) Une nouvelle méthode de
dosage du phosphore assimilable dans les sols forestiers.
Bull. Assoc. Fr. Etude Sol 4:193–198
Dzwonko Z, Loster S (1988) Species richness of small
woodlands on the western Carpathian foothills. Vegetatio
76:15–27
Elkie P, Rempel R, Carr A (1999) Patch Analyst User’s
Manual. Ont. Min. Natur. Resour. Northwest Sci. &
Technol. Thunder Bay, Ont. http://flash.lakeheadu.ca/*
rrempel/patch/
ESRI (1999) Arcview GIS (version for Windows 3.2a). Environmental Systems Research Institute, Inc., Redlands, CA,
USA http://www.esrifrance.fr/
Frazer GW, Canham CD, Lertzman KP (1999) Gap Light
Analyser GLA Version 2.0: imaging software to extract
canopy structure and gap light transmission indices from
true-colour fisheye photographs: users manual and program documentation. Simon Frazer University and the
Institute of Ecosystems Studies, Burnaby, British
Columbia and Millbrook, New York
Gehlhausen SM, Schwartz MW, Augspurger CK (2000) Vegetation and microclimatic edge effects in two mixed-
317
mesophytic forest fragments. Plant Ecol 147:21–35. doi:
10.1023/A:1009846507652
Gonin P (1993) Les stations à intérêt forestier sur les coteaux et
vallées de Midi-Pyrénées situés à l’est de la Garonne. ICatalogue; II- Méthodologie. CETEF Garonnais, Catalogue station, Toulouse, France
Gracia M, Montané F, Piqué J et al (2007) Overstory
structure and topographic gradients determining diversity and abundance of understory shrub species in
temperate forests in central Pyrenees (NE Spain). For
Ecol Manage 242:391–397. doi:10.1016/j.foreco.2007.
01.056
Grashof-Bokdam C (1997) Forest species in an agricultural
landscape in The Netherlands: effects of habitat fragmentation. J Veg Sci 8:21–28. doi:10.2307/3237238
Greene DF, Johnson EA (1995) Long-distance wind dispersal
of tree seeds. Can J For Res 73:1036–1045
Guirado M, Pino J, Roda F (2007) Comparing the role of site
disturbance and landscape properties on understory species richness in fragmented periurban Mediterranean
forests. Landscape Ecol 22:117–129. doi:10.1007/
s10980-006-9009-y
Guyon JP, Bessières F, Faingnaert O (1996) Evolution des
formations boisées sur le territoire de deux communes du
canton d’Aurignac (Haute-Garonne) de 1942 à 1992. In:
Balent G et al (eds) La forêt paysanne dans l’espace rural.
Biodiversité, paysages, produits. INRA, Paris
Harcombe PA, Bill CJ, Fulton M et al (2002) Stand dynamics
over 18 years in a southern mixed hardwood forest, Texas,
USA. J Ecol 90:947–957. doi:10.1046/j.1365-2745.2002.
00735.x
Harper KA, MacDonald SE, Burton PJ et al (2005) Edge influence on forest structure and composition in fragmented
landscapes. Conserv Biol 19:768–782. doi:10.1111/
j.1523-1739.2005.00045.x
Honnay O, Hermy M, Coppin P (1999) Effects of area, age and
diversity of forest patches in Belgium on plant species
richness, and implications for conservation and reforestation. Biol Conserv 87:73–84. doi:10.1016/S0006-3207
(98)00038-X
Ihaka R, Gentleman R (1996) R: a language for data analysis
and graphics. J Comput Graph Statist 5:299–314. doi:
10.2307/1390807
Jacquemyn H, Butaye J, Hermy M (2003) Influence of environmental and spatial variables on regional distribution of
forest plant species in a fragmented and changing landscape. Ecography 26:768–776. doi:10.1111/j.0906-7590.
2003.03620.x
Kupfer JA, Runkle JR (2003) Edge-mediated effects on stand
dynamic processes in forest interiors: a coupled field and
simulation approach. Oikos 101:135–146. doi:10.1034/
j.1600-0706.2003.12119.x
Legare S, Bergeron Y, Pare D (2002) Influence of forest
composition on understory cover in boreal mixedwood
forests of western Quebec. Silva Fenn 36:353–366
Lenière A, Houle G (2006) Response of herbaceous plant
diversity to reduced structural diversity in maple-dominated (Acer saccharum Marsh.) forests managed for sap
extraction. For Ecol Manage 231:94–104. doi:10.1016/
j.foreco.2006.05.024
318
Leroyer C (2002) Structure spatiale de la végétation forestière
à une échelle fine. Dissertation, National School of
Agronomy of Toulouse
Lyon J, Gross NM (2005) Patterns of plant diversity and plant–
environmental relationships across three riparian corridors. For Ecol Manage 204:267–278. doi:10.1016/
j.foreco.2004.09.019
Magurran AE (1988) Ecological diversity and its measurement.
Princeton University Press, Princeton
Matlack GR (1993) Microenvironment variation within
and among forest edge sites in the eastern United States. Biol
Conserv 66:185–194. doi:10.1016/0006-3207(93)90004-K
McCune B, Antos JA (1981) Correlations between forest layers
in the Swan Valley, Montana. Ecology 62:1196–1204.
doi:10.2307/1937284
McDonald RI, Urban DL (2006) Edge effect on species composition and exotic species abundance in the North
Carolina Piedmont. Biol Inv 8:1049–1060. doi:10.1007/
s10530-005-5227-5
McEuen AB, Curran LM (2004) Seed dispersal and recruitment limitation across spatial scales in temperate forest
fragments. Ecology 85:507–518. doi:10.1890/03-4006
Murcia C (1995) Edge effects in fragmented forests: implications for conservation. Trends Ecol Evol 10:58–62. doi:
10.1016/S0169-5347(00)88977-6
Okland T, Rydgren K, Okland RH et al (2003) Variation in
environmental conditions, understorey species number,
abundance and composition among natural and managed
Picea-abies forest stands. For Ecol Manage 177:17–37.
doi:10.1016/S0378-1127(02)00331-6
Onaindia M, Dominguez I, Albizu I et al (2004) Vegetation
diversity and vertical structure as indicators of forest
disturbance. For Ecol Manage 195:341–354. doi:10.1016/
j.foreco.2004.02.059
Orwig DA, Abrams MD (1994) Land-use history (1720–1992),
composition, and dynamics of oak-pine forests within the
Piedmont and Coastal Plain of northern Virginia. Can J
For Res 24:1216–1225. doi:10.1139/x94-160
Palik BJ, Murphy PG (1990) Disturbance versus edge effects in
Sugar-Mapple/Beech forests fragments. For Ecol Manage
32:187–202. doi:10.1016/0378-1127(90)90170-G
Petersen PM (2002) Importance of site conditions and time
since abandonment for coppice vegetation on Langeland,
Denmark. Nord J Bot 22:463–482. doi:10.1111/j.17561051.2002.tb01400.x
Rameau J-C, Mansion D, Dumé G (1989) Flore Forestière
Française. Guide écologique illustré. Plaines et collines.
Institut Développement Forestier, Dijon
A.G. Van der Valk (ed.)
Ranney JW, Bruner MC, Levenson JB (1981) The importance
of edge in the structure and dynamics of forest islands. In:
Burgess RL, Sharpe DM (eds) Forest Island dynamics in
man-dominated landscapes. Springer, New York
Ries L, Fletcher RJJ, Battin J et al (2004) Ecological responses
to habitat edges: mechanisms, models, and variability
explained. Annu Rev Ecol Evol Syst 35:491–522. doi:
10.1146/annurev.ecolsys.35.112202.130148
Sagers CL, Lyon J (1997) Gradient analysis in a riparian
landscape: contrasts among forest layers. For Ecol Manage 96:13–26. doi:10.1016/S0378-1127(97)00050-9
Schuster B, Diekmann M (2005) Species richness and environmental correlates in deciduous forests of Northwest
Germany. For Ecol Manage 206:197–205. doi:10.1016/
j.foreco.2004.10.063
Small CJ, MacCarthy BC (2002) Spatial and temporal variability
of herbaceous vegetation in an eastern deciduous forest.
Plant Ecol 164:37–48. doi:10.1023/A:1021209528643
Stapanian MA, Cassel DL, Cline SP (1997) Regional patterns
of local diversity of trees: associations with anthropogenic
disturbance. For Ecol Manage 93:33–44. doi:10.1016/
S0378-1127(96)03944-8
Stohlgren TJ, Bachand RR, Onami Y et al (1998) Species–
environment relationships and vegetation patterns: effects
of spatial scale and tree life-stage. Plant Ecol 135:
215–228. doi:10.1023/A:1009788326991
Svenning JC, Skov F (2002) Mesoscale distribution of undersorey plants in temperate forest (Kalo, Denmark): the
importance of environment and dispersal. Plant Ecol
160:169–185. doi:10.1023/A:1015885321263
SYSTAT (1999) Systat (R) 9.0 for Windows (R). Statistics I.
SPSS, Chicago
Thioulouse J, Chessel D, Dolédec S et al (1997) ADE-4: a
multivariate analysis and graphical display software. Stat
Comput 7:75–83. doi:10.1023/A:1018513530268
Tutin TG, Heywood VH, Burges NA et al (1983) The Flora
Europaea. Cambridge University Press, Cambridge
Weiher E, Howe A (2003) Scale-dependence of environmental effects on species richness in oak savannas. J Veg Sci
14:917–920.
doi:10.1658/1100-9233(2003)014[0917:
SOEEOS]2.0.CO;2
Whittaker RJ, Willis KJ, Field R (2001) Scale and species
richness: towards a general, hierarchical theory of species
diversity. J Biogeogr 28:453–470. doi:10.1046/j.13652699.2001.00563.x
The importance of clonal growth to the recovery
of Gaultheria procumbens L. (Ericaceae) after
forest disturbance
F. M. Moola Æ L. Vasseur
Originally published in the journal Plant Ecology, Volume 201, No. 1, 319–337.
DOI: 10.1007/s11258-008-9496-9 Springer Science+Business Media B.V. 2008
Abstract We investigated the importance of clonal
growth to the recovery of a common eastern North
American sub-shrub, Gaultheria procumbens L.
(Ericacea), after clearcut logging. Changes in vegetative growth and development of G. procumbens clones
and clonal populations were examined in a chronosequence of logged stands representing different
stages of successional development after clearcutting
(open habitat, young regenerating forest, closed
regenerating forest) and in neighboring undisturbed
late-successional forests representative of presettlement conditions. We specifically quantified seedling
presence and above-ground ramet production, demographic condition (e.g., sexual vs. vegetative stems),
belowground rhizome growth and spread, and assessed
F. M. Moola L. Vasseur
Department of Biology, Dalhousie University, Halifax,
NS, Canada B3H 4J1
F. M. Moola L. Vasseur
Department of Biology, St. Mary’s University, Halifax,
NS, Canada B3H 3C3
Present Address:
F. M. Moola (&)
David Suzuki Foundation, 2211 West 4th Avenue,
Vancouver, BC, Canada V6K 4S2
e-mail: fmoola@davidsuzuki.org
Present Address:
L. Vasseur
Laurentian University, 935 Ramsey Lake Road, Sudbury,
ON, Canada P3E 2C6
the degree of intraspecific variation in clonal morphology and biomass allocation in stands differing in
their disturbance history and degree of successional
development. Recovery in G. procumbens was largely
driven by the ‘‘release growth’’ of pre-existing clonal
bud-banks in response to canopy removal. Release
growth was expressed as greater ramet initiation,
rhizome branching and clonal spread. Conversely, we
found no evidence of sexual establishment in the
species, although production of reproductive biomass
(e.g., inflorescence mass, number of flowering shoots)
was significant. These findings support a deterministic
model of vascular resistance and resilience to catastrophic disturbance, in which recovery of forest plant
communities derives from the life-history characteristics of constituent species.
Keywords Clonal growth Disturbance
Ericacea Old growth Phalanx Resilience
Rhizome Understory
Introduction
The understory flora of temperate forests exhibit a
variety of life strategies, but many are functionally
clonal (Eriksson 1989; Peterson and Jones 1997; Miller
et al. 2002). Although recruitment from seeds is
important for many species and may be favored under
certain environmental conditions (Bierzychudek 1982;
Kanno and Seiwa 2004), many forest plants are
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_24
319
320
maintained in the understory by the asexual propagation of rhizomes, stolons, layers and other perennating
structures (Klimes et al. 1997; Peterson and Jones
1997; Lezberg et al. 1999). For example, long-lived
clonal perennials predominate the understories of
coniferous forests of the Pacific Northwest (Halpern
1989; Lezberg et al. 1999) as well as mixed broadleaved forests in eastern Canada (Sobey and Barkhouse
1977; Roberts and Ramovs 2005) and western Europe
(Peterken and Game 1984; Graae and Sunde 2000).
Although, persistence in a vegetative form has been
shown to be advantageous in all stages of forest
succession (Hughes et al. 1988; O’Dea et al. 1995;
Lezberg et al. 1999), it is particularly important
immediately after forest disturbance, with the only
exception being plant recovery after very severe
disturbance events (e.g., landslides, erosion, land
cultivation, Lee 2004; Roberts 2004). Such truly
disruptive disturbances destroy all pre-existing plants
and propagules due to the severity of damage they incur
to the forest floor and soil (Hughes and Fahey 1991;
Matlack 1994; Ramovs and Roberts 2003). Conversely, in less intensively disturbed habitats (e.g.,
areas impacted by wildfire, windthrow or logging),
clonal growth can allow for the continuity of successful
genotypes after forest disturbance (Matlack et al.
1993b; Lezberg et al. 2001; Roberts 2004). This is
because vegetative reproduction allows most understory herbs and shrubs to re-establish quickly in situ
through the vegetative re-growth of pre-existing clones
as opposed to the recruitment of new individuals from
buried or dispersed seed (Halpern 1989; Hughes and
Fahey 1991; Cirne and Scarano 2001).
The above-ground stems (i.e., ramets) of clonal
plants are often interconnected for long periods of
time, and thus long-lived clonal forest flora are likely
to experience heterogeneity in environmental and
resource conditions as succession proceeds after
disturbance (e.g., changes in light intensity and
quality concomitant with canopy recovery, Antos
and Zobel 1984; Marino et al. 1997; Cirne and
Scarano 2001). Both experimental and field studies
have shown that clonal plants may respond to such
environmental and resource heterogeneity by changing their morphology (e.g., shortening spacer lengths
under favorable conditions, de Kroon et al. 1994; de
Kroon and Hutchings 1995), reducing expenditures
(e.g., decreasing ramet production under unfavorable
conditions, Lezberg et al. 2001) or shifting the
A.G. Van der Valk (ed.)
allocation of biomass (e.g., between vegetative and
sexual structures) under different conditions of
resource supply (Pitelka et al. 1980; Messier et al.
1989; de Kroon and Hutchings 1995; Sun et al.
2002). However, few studies have investigated such
intraspecific variation in clonal traits in relation to
succession after large-scale anthropogenic disturbances, such as clearcut logging (but see Huffman
et al. 1994; O’Dea et al. 1995; Moola and Mallik
1998). Greater autecological knowledge is necessary
to determine the mechanisms by which clonal growth
may confer species’ resistance to logging and thereby
facilitate the recovery and long-term persistence of
understory flora in managed forests.
The objective of this study was to examine the
importance of clonal growth as compared to sexual
reproduction in the recovery of a common eastern
North American sub-shrub, Gaultheria procumbens L.
(eastern teaberry), after catastrophic forest disturbance
(clearcut logging). Earlier work has documented the
response of the species to other land use impacts and
natural disturbance (Matlack et al. 1993a; Donohue
et al. 2000), but detailed studies of long-term recovery
with succession have not yet been made. In this study
we report and discuss a large set of field data on the
autecology of the species in a chronosequence of
logged and undisturbed forest sites. We specifically
quantified seedling presence and above-ground ramet
production, demographic condition (e.g., sexual vs.
vegetative stems), belowground rhizome growth and
spread across the chronosequence. In addition, we
present new information on the degree of intraspecific
variation in clonal morphology and biomass allocation
exhibited by the species during forest succession after
disturbance.
Methods
Study species
Gaultheria procumbens L. (Ericaceae) is an evergreen sub-shrub (\20 cm in height) common to the
understories of closed-canopy coniferous (Pinus spp.,
Picea spp., Abies balamea L. Mill), mixedwood
(Picea spp.–Acer rubrum L.–Abies balsamea L. Mill)
and tolerant hardwood (Acer spp.–Quercus spp.)
forests in central and eastern North America (Donohue et al. 2000; Roberts and Zhu 2002; Moola and
Forest Ecology
Vasseur 2004). It is also found in the understories of
open-canopy forested heathlands (e.g., New Jersey
Pine Barrens, Matlack et al. 1993a), scrub forest
(e.g., pitch pine (Pinus rigida Mill.), scrub oak
(Quercus ilicifolia/Q. prinoides), Motzkin et al.
1999; Donohue et al. 2000), treeless barrens (Strang
1970, 1972), sphagnum bogs (Mirick and Quinn
1981) and cutover land (Roberts and Zhu 2002;
Moola and Vasseur 2004). It has an affinity for acidic
soil, and grows well on many soil types of low
nutrient status, including peat, sand, and sandy loam.
The perennating clonal structures of G. procumbens
are restricted to the upper sections of the Ao soil
horizon (2–3 cm in depth, Coladonato 1994), thereby
making the species vulnerable to severe disturbances
that impact organic soil layers, such as hot forest
fires, which can burn-up upper soil horizons, or
tillage, which destroys buried rhizomes and roots
from mechanical soil turnover (Matlack et al. 1993a;
Donohue et al. 2000). Due to the vulnerability of its
clonal structures to damage or destruction, some
earlier studies have suggested that the species may be
restricted to habitats that have remained continuously
wooded or have otherwise escaped intensive landuse
(Whitney and Foster 1988; Motzkin et al. 1999).
Vegetative reproduction is by additional branching
on existing old aerial ramets (re-growth) or the
initiation of new ramets from buds on belowground
rhizomes (new growth) (Mirick and Quinn 1981).
New ramets that are initiated as either re-growth or
new growth can produce flowers and fruit in their
year of formation and thus presumably contribute to
sexual reproduction (sexual stems) (Donohue et al.
2000). Alternatively, they may remain as purely
vegetative stems. Ramet life span is unknown, but is
expected to be far less than the duration of belowground clonal components (Moola and Mallik 1998).
This is because, unlike belowground components,
aerial ramets are far more susceptible to processes
that can result in stem mortality, such as selfthinning, herbivory, and disturbance (e.g., fire;
Peterson and Jones 1997; Sun et al. 2002). Annual
elongation of rhizomes through additional lateral
growth (spacer length) moves the clone along the
forest floor approximately 10–43 cm year-1 (Sobey
and Barkhouse 1977; Donohue et al. 2000). Nodes
(points of root initiation) are located at least several
centimeters apart along the length of the spreading
rhizomes (Matlack et al. 1993a).
321
Study area
We studied the clonal recovery and development
of G. procumbens in a chronosequence of logged
stands representing different stages of successional
development after clearcutting and in neighboring
undisturbed late-successional forests, which are representative of presettlement conditions in the Acadian
Forest Region (AFR) of North America (Mosseler
et al. 2003) (Table 1). The chronosequence was
located in the vicinity of Liverpool, Nova Scotia
(lat. 44050 N, long. 64460 W). This area has been
classified by Rowe (1972) as belonging to the Atlantic
Uplands Zone of the Acadian Forest Region. Historically, the area was dominated by late-successional
stands of red spruce (Picea rubens (Sarg.)) in mixture
with lesser amounts of eastern hemlock (Tsuga
canadensis L. Carr.), white pine (Pinus strobus L.),
red pine (Pinus resinosa (Ait.)), balsam fir (Abies
balsamea L. Mill), black spruce (Picea mariana
(Mill) B.S.P.), red maple (Acer rubrum L.), and white
birch (Betula papyrifera (Marsh.)) (Rowe 1972).
However, a long history of forest clearance from
agriculture and timber harvesting has had a significant
impact on both the age structure and species composition of forests in the study area (Basquill et al.
2001). The climate of the study area is humid
temperate maritime and is the mildest in Atlantic
Canada. The mean annual temperature is 7.3C. Mean
annual precipitation is 1,601.0 mm, nearly 90% of
which occurs as rainfall. The abundant precipitation is
evenly distributed throughout the year (Canadian
Climate Normals Atlantic Provinces (1961–1990)
(1993). Moola and Vasseur (2004) provide a more
comprehensive description of the study area.
Stand selection
To evaluate the variation in environmental conditions
and patterns of clonal re-establishment, stands were
replicated in a completely randomized design: 2-year
clearcuts (n = 3); 8-year clearcuts (n = 3); 56-year
clearcuts (n = 2); and late-successional forests
(n = 3). The youngest stands (2-year-old clearcuts)
correspond to the open and exposed conditions found
immediately after clearcutting (open habitat); the
8-year-old stands correspond to the early stages of
canopy re-establishment (young regenerating forest),
the 56-year-old stands correspond to the dense
322
A.G. Van der Valk (ed.)
Table 1 Characteristics of chronosequence stands used for individual clone and clone population studies. Mean values with standard
errors (in brackets) are presented
8-Year
clearcut
56-Year
clearcut
Late-successional
forest
2-Year
clearcut
Number of stands (n)
3
3
3
2
Age (years)
128.22 (8.17)
1 (0)
5 (0)
55 (0)
17.36 (0.97)
–
–
13.55 (0.26)
Tree basal area (m /ha)
42.63 (2.27)
–
–
47.68 (3.54)
Tree density (no./ha)b
1253.00 (124.00)
–
–
2336.00 (179.00)
Spring canopy cover (%)a
82.17 (0.60)
0 (0)
0.96 (0.40)
89.97 (0.70)
Summer canopy cover (%)a
95.40 (0.30)
0 (0)
5.09 (1.40)
95.94 (0.30)
% PPFD at 0.1 m heightc
5.80 (0.85)
77.40 (21.66)
56.64 (2.20)
6.78 (1.37)
% PPFD at 0.5 m heightc
6.34 (0.50)
100.00 (0)
74.36 (7.53)
7.11 (1.17)
6.71 (0.15)
100.00 (0)
85.62 (13.79)
7.83 (1.39)
36
36
36
0d
Tree DBH (cm)b
2
% PPFD at 1.0 height
b
c
Total number of clonal fragments sampledd
a
From Moola and Vasseur (2004)
b
From Moola and Vasseur (2006)
c
From Moola (2005)
d
No clonal fragments were sampled from the 56-year-old clearcuts due to a prohibition on destructive sampling in Kejimkujik
National Park (see Methods)
closed-canopy conditions of overstory recovery
(closed regenerating forest), and the late-successional
stands (age 100–165 years) are representative of presettlement conditions (mature–old growth climax
forest) (Mosseler et al. 2003).
Replicate stands within each age-class were randomly chosen from a pool of candidate sites identified
in GIS queries of available forest inventory databases
maintained by Kejimikujik National Park, the Nova
Scotia Department of Natural Resources, and Bowater
Pulp and Paper Company (see Moola and Vasseur
2004). In addition, we ground-truthed all candidate
stands prior to sampling, in an effort to ensure that they
were similar in site type, slope, aspect, soil pH, and
drainage. All of the sites were found on Danesville
sandy loam, Halifax sandy loam, or Bridgewater sandy
loam soils that are stony and shallow, with cobbles and
boulders present in the parent material and on the
surface. Soils are derived from olive colored sandy
loam till with which quartzite is the dominant rock.
The topography is undulating to knobby and soils are
well-drained (Cann and Hilchey 1959). Company
records provided information about the forest cover
prior to logging and silvicultural history (P. Jones,
personal communication). All of the post-clearcut
stands had been mature mixed-coniferous stands
dominated by red spruce, prior to being logged.
Clearcuts were tree-length harvests in which only
trunks of delimbed trees were removed. However,
younger and older clearcuts did differ in the method of
skidding. The pre-disturbance canopies of the oldest
clearcuts (56 years) had been hand-felled, with
trimmed logs horse-skidded to secondary roadways.
Conversely, the younger post-clearcut stands (2–
8 years) originated from mechanical harvests with
logs moved from stumps to roadside using wheeled
skidders. Because of the differences in harvesting
system used to log young and old clearcuts, disturbance intensity and recovery time are the confounding
effects in this study. None of the stands have received
any post-harvest silvicultural treatment, such as
planting, vegetation management (e.g., herbicide
treatment) or pre-commercial thinning.
Sampling techniques
Individual clones
Morphological and biomass allocation measurements
were made on excavated fragments of G. procumbens
clones (polycormons). Polycormons consist of the
portions of a clonal genet that are still attached
Forest Ecology
through living tissue (Kull 1995). In G. procumbens,
this includes interconnected aboveground (stems,
leaves, flowers, fruits) and belowground (rhizomes,
roots) biomass (Donohue et al. 2000). Representative
polycormons were sampled in all stands, except for
the oldest clearcut sites in the chronosequence (56year clearcuts). This was because these sites occur on
federally protected land in Kejimkujik National Park,
where we were not given permission to destructively
sample G. procumbens populations.
In sites where we had permission to destructively
sample G. procumbens clones, we randomly located
three 0.04 ha destructive sample plots (DSP) in each
stand. We excavated four polycormons in each
0.04 ha DSP, for a total of 12 clonal fragment
samples per replicate stand (108 clonal fragments in
total). Sampling was done in mid October 1999. Each
polycormon was excavated from a randomly chosen
undamaged aerial ramet to include all attached
rhizomes, roots, and associated aboveground biomass. Excavations began at the initially selected
ramet and proceeded along the length of attached
rhizomes until the decayed end or the distil tip of a
new rhizome module was reached (Lezberg et al.
2001). On occasion, a selected polycormon extended
beyond the perimeter of the 0.04 ha DSP. In such
cases, we excavated the entire clonal fragment but
excluded the proportion of plant biomass located
outside the DSP from subsequent data analysis.
Excavations were made by carefully removing the
attached clonal components from organic soil layers
with fingers and the assistance of a small trowel.
Following excavation, several morphological parameters were measured on-site (see Morphological
measurements). Afterwards, the clonal fragment was
labeled, bagged, and then placed in a cooler for
transport to the laboratory. In the laboratory, the
clonal fragments were carefully washed of remaining
dirt and debris, air-dried, and frozen for later biomass
allocation analysis (Lezberg et al. 2001).
As part of the excavation of polycormons we also
searched for any G. procumbens seedlings that may
have been present among established clones.
323
new rhizomes (spacer length, cm), measured from the
distil growing tip to the proximal point of branching
or linear extension from older rhizome modules
(Huffman et al. 1994). Spacer length distance is
equivalent to the annual extension of rhizomes, and
determines the speed of vegetative mobility (Kull
1995); (b) the summed length of all new rhizome
growth (total new rhizome length, cm); (c) the
summed length of all older rhizome growth (total
old rhizome length, cm); (d) the summed length of all
rhizome growth combined (total rhizome length, cm);
(e) the number of rhizome apices (new tips); (f)
number of rhizome branches (branching intensity);
(g) number of aerial ramets; and (h) the distance
between consecutive aerial ramets (inter-ramet distance, cm). We also estimated, (i) the current relative
growth rate in each excavated clonal fragment
(annual growth percent, %) by dividing the total
length of new rhizome growth by the total length of
older rhizome growth (Huffman et al. 1994). New
(i.e., current season) rhizomes could be distinguished
from older rhizomes by the unsuberized, fleshy, and
pink colored condition of their tissue. In contrast,
older rhizomes were suberized, woody, and brown or
reddish-brown in color (Donohue et al. 2000).
Biomass measurements
Frozen polycormons were thawed at room temperature and subsequently separated into four biomass
components: (a) rhizomes and roots, (b) aerial stems,
(c) leaves, and (d) sexual organs (flowers and fruit).
All plant tissues were oven-dried to constant mass at
70C for 48 h. and then weighed to 0.001 g. Proportions of component biomass (i.e., % of total) were
calculated by dividing different component weights
by the weight of the entire excavated polycormon
(Sun et al. 2002). The ratio of above- to belowground
biomass was calculated for each clonal fragment by
dividing the dry weight of aboveground structures
(stems, leaves, flowers, and fruit) by belowground
structures (rhizomes and roots).
Morphological measurements
Clonal populations
The following morphological parameters were measured on each clonal fragment immediately following
its excavation in the field: (a) the length of individual
Gaultheria procumbens populations were described
in four 1 m2 quadrats adjoining the area where the
clonal fragments were excavated. Sampling was
324
conducted during a 2-week period in early July when
clones were flowering, but had not yet produced fruit.
The population quadrats were located on the outside
four corners of each 0.04 ha DSP plot, for a total of
12 quadrats per replicate stand (132 quadrats in total).
Within each 1 m2 quadrat we assessed total ramet
production (standing stem density) and reproductive
activity by unit area by counting the number of
flowering (i.e., sexual) and non-flowering (i.e., vegetative) aerial stems as well as the overall number of
flowers per m2. A sub-sample of 15 randomly chosen
stems in each quadrat was also measured for height,
depth of origin, and the number of leaves and flowers
(reproductive activity per stem) on each stem. On
occasion, a quadrat contained less than 15 suitable
stems that could be sampled in this way. In such
cases, we sampled additional aerial stems immediately outside the 1 m2 quadrat until a sample of 15
aerial stems was obtained. Atypical large or small
stems were excluded from the sample (Lezberg et al.
1999). Each of the selected aerial stems in the sample
was also classified according to its current demographic condition in four classes: (a) old nonreproductive (i.e., an old vegetative stem without
current growth), (b) old re-growth (i.e., an old
vegetative stem with additional clonal branching),
(c) new growth (i.e., a new vegetative stem), and (d)
seedling (i.e., an independent stem possessing intact
primary roots) (Donohue et al. 2000). New clonal
growth was easily distinguished, as new aerial stems
and branches were unsuberized, fleshy, and whitegreen in color. Conversely, older aerial stems were
brown to black in color and well suberized (Donohue
et al. 2000). The relative proportion of ramets in each
demographic class was determined at the stand scale
(i.e., among 180 stems) and compared across the
chronosequence (n = 2–3 replicate stands per chronosequence age-class). In total, 1,980 aerial stems
were examined across the chronosequence. All were
of vegetative origin as no seedlings were observed
among any of the quadrats.
Statistical analysis
Individual clones: univariate analysis
One-way analysis of variances (ANOVA) was used to
test whether the measured morphological and biomass parameters differed significantly across the
A.G. Van der Valk (ed.)
chronosequence. Aggregated stand means of each
parameter were tested (i.e., means of 12 clonal
fragments per stand were tested). Where a significant
effect was found, a Tukey HSD test was used to
compare stand means among individual age-classes
(Zar 1996). Log transformations were used on most
variables in order to improve the homogeneity of
variances and normality of data prior to statistical
analysis. The distribution of spacer lengths and interramet distances within harvested polycormons was
also compared between age-classes with a Kolmogorov
–Smirnov test (Zar 1996; Sun et al. 2002). Significance in means for all tests was determined at the
P = 0.05 level. Analyses were conducted in SPSS
ver. 6.1.3. (SPSS 1995).
Individual clones: multivariate analysis
In addition to comparing how individual morphological traits in G. procumbens differed across the
chronosequence, we were also interested in whether
polycormons harvested from different age classes
could be discriminated on the basis of overall
morphology (i.e., multiple morphological parameters
tested together). For this purpose, we employed a
multivariate linear discriminant analysis (LDA)
(Jongman et al. 1995) on harvested clonal fragments,
with measured morphological parameters employed as
explanatory factors and chronosequence age-classes
treated as a priori defined groups (Leps and Smilauer
2003). The LDA was initially run using all 14
morphological measurements taken from the clonal
fragments harvested in the late-successional, 2-year,
and 8-year clearcuts. Polycormons from the 56-year
clearcut stands had to be excluded from this analysis,
as we did not have a full suite of morphological data
(see Methods—Sampling techniques). All the clonal
measurements were log transformed prior to analysis,
since most did not meet the criterion of homogeneity
of variance (Leps and Smilauer 2003). The original
morphological parameters employed in the LDA
included: ramet number per clonal fragment, ramet
number per centimeter of rhizome, mean inter-ramet
distance (cm), branching intensity, number of rhizome
tips, mean spacer length (cm), total new rhizome
length (cm), total old rhizome length (cm), total
rhizome extension (cm), relative clonal growth rate
(%), absolute rhizome weight (g), absolute stem
weight (g), absolute leaf weight (g), and the ratio of
Forest Ecology
above- to belowground biomass. From this full set of
morphological measurements, we subsequently
excluded those parameters that were either highly
correlated or of low explanatory power, since such
measurements were more or less redundant or superfluous in the LDA (Norusis 1992). The total rhizome
length, total old rhizome length, and stem weight
variables were removed since they were all highly
correlated with other measured morphological parameters (e.g., with rhizome weight and ramet number per
clonal fragment, r [ 0.70 in a pooled within-groups
correlation matrix) (Norusis 1992). We employed a
‘‘Forward Selection’’ of remaining morphological
parameters to determine the minimum number
required to discriminate polycormons, approximately
as well as the full suite of clonal measurements (Leps
and Smilauer 2003). Based on this analysis, we
decided to exclude the mean spacer length, total new
rhizome length, and the ratio of above- to belowground biomass measurements from the final LDA, as
they were insignificant in a non-parametric Monte
Carlo permutation test (P C 0.05, 999 random runs).
Thus, the final LDA model was based on eight
morphological (explanatory) parameters, following
the exclusion of 6 of the original 14 clonal measurements. This sub-set of morphological parameters
explained 89.9% (1.172 of 1.303) of the variation that
could be explained by all original 14 clonal characteristics together. Pooled within-group correlations
between the selected morphological measurements
and canonical discriminant functions were obtained
from the LDA to assess the contribution of individual
clonal traits to the discrimination of polycormons
along multivariate axes (Norusis 1992). Forward
Selection was conducted in CANOCO ver. 4.0.2 (ter
Braak and Smilauer 1999) and the LDA was run in
SPSS ver. 6.1.3. (SPSS 1995).
Clonal populations: univariate analysis
Differences in the standing-stem density and the
relative proportion of vegetative and reproductive
ramets summarized at the stand level were tested
across the chronosequence in a one-way analysis of
variance (ANOVA), followed by a Tukey HSD test to
compare the means of individual age-classes (n = 2–
3 replicate stands/age-class). The proportions of
aerial ramets in each demographic condition class
and the number of leaves, number of flowers, height,
325
and sprouting depth of clonal ramets were treated
statistically in the same way. Ramet proportion data
was transformed (arcsine) to ensure normal distribution. The other variables were analyzed after log
transformation in order to homogenize the variation.
Significance in means summarized at the stand scale
(n = 2–3 replicates) was determined at the P = 0.05
level. Analyses were conducted in SPSS ver. 6.1.3.
(SPSS 1995).
Results
Clonal populations: general response
to clearcutting
Gaultheria procumbens populations exhibited a
delayed positive response to clearcutting. Total
standing stem density did not significantly change
immediately following canopy removal (Tukey HSD
of late-successional versus 2-year clearcut sites;
P = 0.981) but did so shortly thereafter (Table 2).
In particular, the mean number of ramets per unit area
(m2) increased by five times in the eighth growing
season after logging, relative to pre-disturbance
conditions (Tukey HSD of late-successional versus
8-year clearcut sites; P = 0.008). This dramatic
increase in population numbers after clearcutting
was short-lived, as ramet density returned to predisturbance levels by the fifth decade of secondary
succession after logging (Tukey HSD of late-successional versus 56-year clearcut sites; P = 0.239).
Changes in the reproductive composition of
G. procumbens populations exhibited a similar successional response to logging. Both the absolute and
relative number of vegetative and sexual stems
changed several years after clearcutting, only to
return to pre-disturbance conditions later on in
secondary succession (ANOVA; P B 0.05) (Table 2).
Populations were dominated by vegetative ramets in
all chronosequence age-classes, although the proportion of vegetative stems was significantly greater in
closed-canopy stands (late-successional and 56-year
clearcuts, 94.8 and 98.1% of all stems, respectively),
than in early seral open habitats (2-year and 8-year
clearcuts, 76.2 and 69.8%, respectively) (Table 2;
Tukey HSD; P B 0.005).
Sexual reproductive activity per unit area (m2) and
per stem increased significantly with clearcutting at
0.002
Mean values with standard errors (in brackets) are presented. Like letters within columns denote statistically similar means (Tukey HSD) at P B 0.05 following a significant 1Way ANOVA
0.001a (0.001)
0.01
0.001
0.024
0.022
0.001
1.44b (0.16)
P-value in ANOVA
0.003
0.33a (0.00)
98.1%a (0.93)
1.85%a (0.93)
13.25a (2.58)
0.42a (0.25)
56 Year Clearcut (n = 2)
12.8a (2.33)
1.04b 0.36)
22.28b (8.29)
105.28b (36.56)
69.8%b (5.79)
76.2%b (6.15)
23.76%b (6.15)
30.23%b (5.79)
143.11b (35.32)
26.50a (6.01)
19.9a (5.9)
8-Year clearcut (n = 3)
101.4b (23.10)
6.56a (1.33)
41.72b (13.64)
2-Year clearcut (n = 3)
0.08a (0.07)
2.22a (1.90)
94.8%a (4.27)
5.24%a (4.27)
29.06a (4.89)
Total
Non-flowering
27.7a (5.42)
1.39a (1.23)
Late-successional forest (n = 3)
Flowering
Flowering
Non-flowering
Relative density of aerial ramets (%/m2)
Absolute density of aerial ramets (no./m2)
Flowers
(no./m2)
Flowers
(no./per stem)
A.G. Van der Valk (ed.)
Age class
Table 2 Standing stem density and reproductive activity per unit area (m2) and per stem in G. procumbens populations sampled in a chronosequence of late-successional and
clearcut stands
326
first (2–8 years after logging), but dropped to
extremely low levels later on in secondary succession, concomitant with canopy closure (Table 2;
ANOVA; P B 0.05). For example, a much higher
proportion of aerial ramets consisted of flowering
stems in open (2-year clearcut; 23.8%) and young
regenerating forests (8-year clearcut; 30.2%), than in
closed-canopy secondary (56-year clearcut; 1.9%) or
late-successional stands (5.2%) (Tukey HSD;
P B 0.05). Similarly, the number of flowers per unit
area (m2) and per stem was far greater in recently
logged sites, than in older recovering forests or
undisturbed late-successional stands (e.g., ANOVA
on number of flowers per m2; F[3,7] = 18.06,
P = 0.001).
Despite the major increase in sexual reproductive
activity with clearcutting, we found no evidence of
actual recent sexual establishment in G. procumbens
populations. For example, of the 1,980 stems that we
examined in detail in the chronosequence, all were of
vegetative origin (Fig. 1). Furthermore, we failed to
locate a single seedling in any of the population
surveys (132 quadrats) or among the destructively
harvested polycormons (108 clonal fragments) in the
chronosequence, despite the wide range in resources
(e.g., light), stand conditions (e.g., canopy cover), and
substrates (e.g., decayed wood, exposed mineral soil)
investigated (Table 1).
The demographic structure (i.e., ramet condition
classes) of G. procumbens populations was affected
by clearcutting (Fig. 1). Populations in the 2-year
clearcuts were comprised primarily of new vegetative
ramets (57.6% of all stems examined), that originated
from buds on buried rhizomes (new growth ramets).
The proportion of new growth ramets decreased with
stand recovery after logging and was statistically
lowest in the 56-year-old clearcut sites (11.9%)
(ANOVA on proportion of new ramets, F[3,7] =
36.27, P = 0.0001). The predominant form of ramet
recruitment in older clearcuts (8–56 years) and latesuccessional forests was by additional branching of
pre-existing stems (old re-growth ramets), the proportion of which remained constant across the
chronosequence (ANOVA on proportion of re-growth
ramets, F[3,7] = 3.74, P = 0.078). Conversely, the
proportion of old ramets with no growth (old nonreproductive ramets) varied significantly among chronosequence age-classes, indicating a significant effect
of clearcutting (ANOVA on proportion of old ramets,
Forest Ecology
327
Fig. 1 Mean relative
number of (a) new ramets,
(b) old ramets with new
growth, and (c) old ramets
in Gaultheria procumbens
populations in a
chronosequence of latesuccessional and clearcut
stands. Standard error bars
are shown. Means sharing
the same letter within a
ramet stage class (i.e.,
above like bars) are not
significantly different
(Turkey HSD) at P B 0.05,
following a significant
ANOVA. Demographic
stage structure was assessed
from 180 randomly chosen
ramets in each replicate
stand of the chronosequence
(n = 2–3 replicates per
chronosequence age-class)
F[3,7] = 21.84, P = 0.001). In particular, a higher
proportion of ramets were non-reproductive in the
understories of closed-canopy forests (56-year clearcut
and late-successional sites, 23.5–30.3% respectively)
than in early-seral open or regenerating habitats
(2-year and 8-year cleracuts, 7.0–7.6% respectively)
(P B 0.005 in Tukey HSD). None of the stems
surveyed were of seedling origin.
Individual clones: above- and belowground
morphology
Logging increased the growth and vigor of G. procumbens clones, although this positive response was
not apparent in measured polycormons (i.e., clonal
fragments) until the eighth growing season after
disturbance (Table 3). For example, harvested polycormons in 8-year clearcuts supported a far greater
number of aerial ramets (on the whole and per length
of rhizome), had larger belowground rhizome systems (e.g., greater in current season and overall
rhizome growth), and produced more new rhizomes
(rhizome tips) and branches (branching intensity) per
polycormon than in other age-classes (Tukey HSD;
P B 0.05). Polycormons growing in open and regenerating clearcuts also produced more leaves per ramet
(ANOVA on leaf number per ramet; F[2,5] = 14.39,
P = 0.002), but were significantly shorter in height
(ANOVA on ramet height; F[3,7] = 8.42, P = 0.01).
Although polycormons in 8-year clearcuts comprised significantly larger rhizome systems in
absolute terms, the rhizome relative growth rate
(annual growth percent) was greatest in the 2-year
clearcuts. Polycormons in these sites expanded by
43.38% annually, compared to 13.64% in the 8-year
clearcut sites and 8.08% in the late-successional
stands (Table 3). Even though the difference in
annual rhizome growth percent was large between
clearcut age-classes, it was not statistically significant
because of high variation among measured clonal
fragments (Tukey HSD of 2-year clearcut versus 8year clearcut sites; P = 0.183).
None of the harvested polycormons possessed intact
primary roots, which would have indicated that they
328
Table 3 Morphological characteristics of above- and belowground plant components of sampled G. procumbens clonal fragments in a chronosequence of late-successional and
clearcut stands
Late-successional Forest (n = 3)
2-Year clearcut (n = 3)
8-Year clearcut (n = 3)
56-Year clearcut (n = 2)
Mean
Mean
Mean
S.E.
Mean
P-value
in ANOVA
6.83a
4.64a
0.42
0.02
0.01
0.01
S.E.
S.E.
SE
Aboveground characteristics
Ramet height (cm)a
Number of leaves per rameta
5.08a
4.11a
0.46
0.29
4.27b
5.74b
0.22
0.26
4.75b
5.89b
0.11
0.07
29.72a
11.02
16.58a
2.16
114.71b
42.29
0.08a
0.01
0.09ab
0.00
0.11b
0.00
0.03
12.61a
0.38
6.32b
0.38
5.49b
0.40
0.00
Rhizome depth (cm)a
2.34a
0.24
3.18a
0.24
3.18a
0.19
Number of rhizome branches per clonal
fragment (branching intensity) §
5.56a
3.06
7.69a
0.43
52.75b
26.75
Number of rhizome tips per clonal fragment §
4.33a
1.92
5.78a
0.31
23.17b
12.67
0.05
Mean spacer length (cm) §
7.14a
0.77
8.98a
0.86
8.68a
0.81
0.26
Number of ramets per clonal fragment §
Number of ramets per centimeter of rhizome §
Inter-ramet distance (cm) §
0.02
Belowground characteristics
2.92a
0.04
0.12
0.03
Total new rhizome length (cm)
31.89a
16.39
51.48a
3.64
156.85b
83.40
0.05
Total old rhizome length (cm)
Total rhizome extension (cm)
382.12ab
414.01ab
113.94
130.16
256.03b
307.52b
24.46
130.16
1136.79a
1293.65a
434.47
517.87
0.03
0.04
Aboveground:belowground biomass
0.92a
0.03
1.43a
0.18
1.64a
0.49
0.14
Rhizome relative growth rate (%) §
8.08a
1.92
43.38b
1.92
13.64ab
3.86
0.04
* Mean of 180 randomly selected ramets in each stand of the chronosequence (2–3 replicate stands per chronosequence age-class)
A.G. Van der Valk (ed.)
Mean values with standard errors (S.E.) are presented in rows. Like letters within rows denote statistically similar means (Tukey HSD) at P B 0.05, following a significant 1-Way
ANOVA. Data on most plant components were unavailable for clonal fragments in the 56-year-old clearcut stands due to a prohibition on destructive sampling (see Methods).
Morphological parameters employed in the Linear Discriminant Analysis of clonal fragments (see Fig. 5) are identified with the symbol §
Forest Ecology
had started from seed somewhat recently (\10 years
previously, Lezberg et al. 2001). Rather, the presence
of well-suberized, thickened, and decaying older
rhizome clumps from which newer rhizomes had
clearly originated in the past suggests that most
polycormons had been clonal for many years if not
decades. However, we were unable to confirm this by
aging the polycormons. This was due to the extent of
decay among older rhizomes, which made counting
annual rhizome growth rings impossible.
Clearcutting had no effect on the mean length of
new rhizomes (spacer length) produced annually in
G. procumbens clones (Table 3; ANOVA on spacer
length; F[5,7] = 1.76, P = 0.264). Nevertheless, the
distribution of spacer lengths was significantly different in clones growing in late-successional stands
compared to clearcut forests (e.g., Kolmogorov–
Smirnov test on polycormons in late-successional
versus 2-year clearcuts; P = 0.002) (Fig. 2). Spacer
lengths in early-seral habitats tended to be greater
than in late-successional habitats. For example,
approximately 19% of spacer lengths produced by
polycormons growing in open (2-year clearcuts) were
longer than 10 cm. Conversely, only 4% of spacer
lengths in late-successional habitats were [10 cm in
length (Fig. 2).
As with spacer length, the distribution of clonal
inter-ramet distances was significantly different in
late-successional stands compared to clearcut forests
(e.g., Kolmogorov–Smirnov test on polycormons in
late-successional versus 2-year clearcuts; P = 0.0001)
(Fig. 3). Generally, the distance between ramets was
much shorter on polycormons growing in clearcut
habitats than in late-successional stands (Table 3;
ANOVA on inter-ramet distance; F[5,7] = 78.10,
P = 0.0001). This was because polycormons in clearcut habitats produced a greater density of ramets per
length of rhizome (Table 3), thereby forming coalesced clumps of stems in the understory. For example,
although inter-ramet distance ranged from 0.5 cm to as
much as 48 cm in clearcut polycormons, consecutive
ramets were most often produced less than 2 cm apart
(Fig. 3). Conversely, ramets were more equitably
initiated along the length of rhizomes growing in
late-successional stands and in many cases were
separated by considerable distances ([20 cm). The
consequence of this, was that ramets in some latesuccessional polycormons were formed in significant
isolation from one another on the forest floor.
329
Fig. 2 Distribution of annual extension of rhizome (spacer
length, cm) in Gaultheria procumbens polycormons harvested
from (A) late-sucessional forest, (B) 2-year clearcut, and (C) 8year clearcut stands of the chronosequence. Frequencies
represent the average of 12 clonal fragments per stand and
three replicate stands per chronosequence age-class. Spacer
length data was unavailable for Gaultheria clonal fragments
present in the 56-year clearcut stands of the chronosequence
(see Methods). Different letters above the histograms indicate
that the distribution of spacer lengths was significantly
different among chronosequence age-class (P B 0.05, Kolmogorov–Smirnov test)
Individual clones: biomass allocation
Gaultheria procumbens did not allocate biomass
differently to various vegetative components in
response to disturbance (Fig. 4). Biomass allocation
to leaves, stems, rhizomes, and roots remained
constant across the chronosequence (ANOVA;
P C 0.05). The exception to this was allocation to
sexual components (inflorescences, fruit). A greater
proportion of overall biomass consisted of sexual
biomass in open and regenerating clearcuts than in
late-successional forests in the chronosequence
330
A.G. Van der Valk (ed.)
Fig. 3 Distribution of
inter-ramet distances (cm)
in G. procumbens
polycormons harvested
from (A) late-successional
forest, (B) 2-year clearcut,
and (C) 8-year clearcut
stands of the
chronosequence. Interramet data was unavailable
for Gaultheria clonal
fragments present in the 56year clearcut stands of the
chronosequence (see
Methods). Different letters
above the histograms
indicate that the distribution
of inter-ramet distances was
significantly different
among chronosequence
age-class (P B 0.05,
Kolmogorov–Smirnov test)
(ANOVA on relative proportion of reproductive
biomass; F[5,7] = 6.766, P = 0.038).
Individual clones: intraspecific variation
in overall morphology
Linear discriminant analysis confirmed the separation of the harvested polycormons according to
seral origin (open habitat, young regenerating
forest, late-successional forest) with 92.2% accuracy. The first discriminant function accounted for
71.6% of the variation among harvested clonal
fragments (Table 4). Function 1 was most strongly
correlated with inter-ramet distance, branching
intensity, and the number of new rhizomes, indicating that these morphological parameters best
Forest Ecology
331
Fig. 4 Mean biomass and proportional biomass (i.e., % of
total) of component plant parts of Gaultheria procumbens
polycormons harvested from a chronosequence of late-successional and clearcut stands. (A) Mean dry mass (g) of rhizomes
and roots, stems, leaves, and flowers and fruit. (B) Mean
proportion of total dry biomass (%) for component plant parts.
Component plant parts that are significantly different in either
biomass or proportional biomass are identified in the figure
with the symbol * (1-way ANOVA at P B 0.05). Standard
errors and significant pair wise differences among age-classes
are described in the text (see Results). Biomass data were
unavailable for Gaultheria procumbens polycormons present in
the 56-year clearcut stands of the chronosequence (see
Methods)
distinguish polycormons along the first multivariate axis. This axis is indicative of a gradient in
growth form that is associated with successional
development. Function 2 accounted for 28.4% of
the variance and was best correlated with ramet
density, dry mass of rhizomes, and the relative
332
A.G. Van der Valk (ed.)
Table 4 Summary of a Linear Discriminant Analysis (LDA) of G. procumbens clonal fragments harvested from a chronosequence
of late-successional and clearcut stands
Function
Eigenvalue
% of variance
1
2.477
71.6
2
0.981
28.4
Cumulative (%)
Canonical correlation
df
Significance
71.6
0.844
20
0.00
100.0
0.704
9
0.00
Fig. 5 Ordination plot of
Gaultheria procumbens
polycormons based of a
Linear Discriminant
Analysis (LDA) of eight
significant morphological
parameters. Symbols
represent individual clonal
fragments harvested from a
chronosequence of latesuccessional and clearcut
stands. Morphological data
was unavailable for
G. procumbens
polycormons present in the
56-year clearcut stands (see
Methods). Function axes 1
and 2 are shown
clonal growth rate. It distinguishes polycormons in
open habitats from those in young regenerating
forests.
As shown in the ordination diagram (Fig. 5),
polycormons growing under late-successional conditions are distinct from polycormons in seral habitats
in terms of overall morphology and architecture.
Clones occur as large coalesced clumps of stems in
open habitats (phalanx growth form) and spreading
isolated chains of connected stems in late-successional stands (guerrilla growth form). The differences
in these two archetype growth forms are largely
related to the intensity of clonal branching, interramet distance, the initiation of new rhizomes, and
the overall biomass of clonal components (primarily
leaves) (Table 5).
Discussion
Importance of clonal growth to recovery
and persistence after disturbance
The results of this study show that G. procumbens is
tolerant of clearcut logging due to its ability to respond
positively to canopy removal with the recruitment of
new stems aboveground and vigorous vegetative
expansion belowground (e.g., increased relative rhizome growth rate). Ramet initiation in early-seral
habitats (e.g., open and young regenerating forest) is
dominated by the ‘‘release growth’’ of new vegetative
stems initiated from persistent rhizome bud-banks,
rather than additional branching of existing ramets
(Matlack et al. 1993a). Conversely, populations
Forest Ecology
333
Table 5 Pooled within-groups correlations between morphological parameters and standardized canonical discriminant functions in
a Linear Discriminant Analysis (LDA) of harvested G. procumbens clonal fragments
Morphological Parameters
Function 1
Inter-ramet distance (cm)
-0.475*
0.284
Branching intensity
0.465*
0.367
Number of rhizome tips
0.349*
0.260
-0.099*
0.048
Number of ramets
0.235
0.786*
Dry mass of leaves (g)
0.353
0.671*
Number of ramets per centimeter of rhizome
Function 2
Dry mass of rhizomes (g)
0.148
0.587*
Relative clonal growth rate (% year-1)
0.265
-0.487*
Morphological parameters are ordered by absolute size of correlation with each ordination axis. The largest absolute correlation
between each variable and any discriminant function is denoted by a *. Only morphological parameters with significant explanatory
value were included in the LDA (Forward Selection; P B 0.05 in a Monte Carlo Permutation test of the marginal effects of individual
parameters; see Methods). Age-class means of morphological parameters are presented in Table 3
remain largely non-reproductive in closed-canopy
stands, where light transmission to the understory is
extremely low due to interference from taller shrub and
tree strata. Dense canopy conditions have also been
shown to reduce ramet recruitment and other aspects of
clonal growth (e.g., rhizome branching) in other
understory species (Tappeiner and Alaback 1989;
Whitman et al. 1998; Lezberg et al. 2001). Decreased
production of new biomass in favor of greater persistence of existing stems and foliage may be
advantageous in stressful low resource environments
such as deeply shaded forest understories as it reduces
costly resource expenditures (Silva et al. 1982;
Lezberg et al. 2001). For example, Lezberg et al.
(2001) found that continual re-leafing of persistent
non-reproductive stems in Maianthemum dilatatum L.
was less costly than the vegetative initiation of new
ramets when resources such as light are in short supply
(e.g., after canopy closure, [20 years, de Kroon and
Hutchings 1995; Lezberg et al. 2001). This survival
strategy is in accordance with Grime’s (1979) conceptual model of stress-tolerance during succession
and may explain the long-term persistence in the
understory of G. procumbens and other understory
species in all stages of stand recovery after disturbance
(Halpern 1988; Moola and Vasseur 2004). Indeed, the
long-lived stems and evergreen habitat of G. procumbens
reduces the costly turnover of plant parts such as its
succulent leaves (Matlack et al. 1993a). Furthermore,
its persistent foliage is photosynthetically active for
much of the year, thereby allowing it to avoid
competition with co-occurring deciduous shrubs and
trees through phenological displacement, even though
these strata overtop it year-round (Matlack et al.
1993a).
Advantages of clonal growth over sexual
reproduction
Although a significant proportion of G. procumbens
aerial ramets are potentially sexually reproductive in
the first year of growth (e.g., high rates of flowering),
establishment from seed appears to be insignificant as
a means of recovery after logging disturbance.
Indeed, though Reyes (2002) found seeds of G.
procumbens to be abundant in the soil profile of the
open habitat and young recovering forest sites of our
chronosequence, we observed no seedlings in any of
the population survey plots nor did we find any
polycormons of recent sexual origin. Matlack and
Good (1990) similarly failed to observe any evidence
of seedling establishment in G. procumbens, despite
a much greater sampling effort (2,040 quadrats 1 m2)
and surveys over a much larger range of soil types
and disturbance conditions than represented in our
chronosequence. These findings indicate that, similar
to other understory species in northern temperate
forests (e.g., Vaccinium myrtilloides Michx., Rubus
spectabilis Pursh., Tappeiner et al. 1991; Moola and
Mallik 1998), recovery of G. procumbens after
disturbance is driven primarily by vegetative processes such as ramet initiation and persistence
334
(Hughes and Fahey 1991; Tappeiner et al. 1991;
Cirne and Scarano 2001).
The dominance of vegetative propagation over
sexual establishment has been described as paradoxical (Vander Kloet and Hill 1994), given the
significant allocation of resources that understory
species, like G. procumbens, invest in the development of reproductive biomass, with often limited
benefits in terms of regeneration success. For example, though we found no change in biomass allocation
among vegetative components (leaves, rhizomes,
stems), allocation to sexual biomass increased immediately after logging in G. procumbens. Increased
allocation to reproductive biomass after canopy
removal may simply be due to the larger size of
G. procumbens polycormons in younger clearcuts
(Tappeiner et al. 1991; Cirne and Scarano 2001; Sun
et al. 2002). Alternatively, it may be related to the
greater density of stems in disturbed habitats relative
to late-successional stands. For example, Sun et al.
(2002) found that sexual reproduction is favored over
clonal propagation in Scirpus mariqueter L. when the
density of ramets becomes so high that inter-ramet
interference occurs (e.g., self-thinning).
Vegetative growth confers a number of advantages
that directly improve regeneration success in disturbed habitats, though these benefits have not yet
been studied in detail in G. procumbens. For example, stored carbohydrate reserves in the biomass of
pre-existing clonal structures (e.g., buried rhizomes)
may subsidize the growth of vegetative offspring
after disturbance (Peterson and Jones 1997; Price and
Marshall 1999). These subsidies have been shown to
increase growth and improve the survivorship of new
clonal ramets, relative to seedlings, in other understory species (Frost 1984; de Steven 1989; Peterson
and Jones 1997).
Environmental heterogeneity and intraspecific
variation in clonal growth
Forest logging creates habitat heterogeneity and
changes the availability of resources for understory
plants, like G. procumbens. For instance, the complete removal of the overstory with clearcut logging
immediately alters the microclimatic conditions
affecting pre-existing understory plants (e.g., ? in
solar radiation and temperature maxima,—in relative
soil moisture and humidity, Ramovs and Roberts
A.G. Van der Valk (ed.)
2003). In addition, the direct mechanical effects of
logging and its associated practices (e.g., site
preparation) can affect the spatial heterogeneity of
the forest floor (microtopography), by fragmenting
coarse woody debris, removing the litter layer, and
creating mounds and deep ruts in the ground
(Roberts and Zhu 2002; Ramovs and Roberts 2003;
Moola and Vasseur 2004). These processes create an
environmental and resource matrix that is spatially
and temporally variable at a scale relevant to clonal
plants during succession after disturbance (Matlack
et al. 1993a; Price and Marshall 1999; Sun et al.
2002). Foraging hypotheses predict that clonal plants
will respond to such heterogeneity by shifting their
morphology and architecture so as to preferentially
exploit higher resource patches (de Kroon and
Hutchings 1995; Price and Marshall 1999). In many
plants, this plasticity in growth form is mediated by
greater branching of rhizomes, reduced spacer
length, and reductions in inter-ramet distance (i.e.,
internode length) (Peterson and Jones 1997). In this
study, we found that G. procumbens does show
considerable variation in morphology in response to
disturbance and subsequent stand recovery. The
results of the multivariate Linear Discriminant
Analysis show that G. procumbens clones exhibit
archetypal growth forms in resource-rich disturbed
habitats (e.g., open and regenerating young forests)
versus late-successional forests, that are characterized by low light conditions. Clones exist as large
coalesced clumps of stems in open habitats (phalanx
growth form) and spreading isolated chains that
cover large areas of the forest floor in late-successional stands (guerrilla growth form) (de Kroon and
Hutchings 1995; Klimes et al. 1997). The morphological traits that drive this intraspecific variation in
growth form are inter-ramet distance, branching
intensity, rhizome density, and the biomass of
vegetative components. Furthermore, the frequency
distribution of spacer lengths and inter-ramet distances reveals that polycormons tend to concentrate
growth under more favorable conditions through the
‘‘space-packing’’ of ramets, as opposed to reductions
in overall spacer length. Such a strategy may be
interpreted as an effort to consolidate growth in an
effort to exploit newly created favorable patches or
to rapidly re-establish aboveground photosynthetic
biomass after local disturbance (Matlack et al.
1993a; Cirne and Scarano 2001).
Forest Ecology
Acknowledgments Our appreciation is extended to
Kejimkujik National Park and Bowater Mersey Paper
Company for permission to collect data on lands owned or
managed under their authority. K. Brooks, S. Wilson, D.
Galway, A. Letourneau, C. Konoff, J. McLean, H. Verheul, and
H. Leblanc assisted with the collection of field data. Drs. A.
Mosseler, J. Major, R. Scheibling, B. Latta, M. Johnston, L.
Belanger, and M. O’Brien provided editorial comments.
Funding and material support was provided by the Canadian
National Science and Engineering Research Council (NSERC)
in the form of a PGSB scholarship to F. Moola and an NSERC/
Canadian Forest Service/Industrial (Bowater–Mersey Paper
Company) grant awarded to L. Vasseur. The Lett Fund of the
Department of Biology at Dalhousie University provided
additional financial support.
References
Antos JA, Zobel DB (1984) Ecological implications of
belowground morphology of nine coniferous forest herbs.
Bot Gaz 145(4):508–517. doi:10.1086/337486
Basquill SP, Woodley SJ, Pardy AB (2001) The history and
ecology of fire in Kejimkujik National Park. Parks Canada, Halifaxa
Bierzychudek P (1982) Life histories and demography of
shade—tolerant forest herbs: a review. New Phytol 90:
757–776. doi:10.1111/j.1469-8137.1982.tb03285.x
Canadian Climate Normals Atlantic Provinces (1961–1990)
(1993) Ottawa, Environment Canada
Cann DB, Hilchey JD (1959) Soil survey of Queens County,
Nova Scotia. Nova Scotia Soil Survey, Nova Scotia
Department of Agriculture and Marketing, Truro, Nova
Scotia
Cirne P, Scarano FR (2001) Resprouting and growth dynamics
after fire of the clonal shrub Andira legalis (Leguminosae)
in a sandy coastal plain in south-eastern Brazil. J Ecol
89:351–357. doi:10.1046/j.1365-2745.2001.00557.x
Coladonato M (1994) Gaultheria procumbens (L.). Fire Effects
Information System [Online]. F. S. U.S. Department of
Agriculture, Rocky Mountain Research Station, Fire Sciences Laboratory. 2005
de Kroon H, Hutchings MJ (1995) Morphological plasticity in
clonal plants: the foraging concept reconsidered. J Ecol
83:143–152. doi:10.2307/2261158
de Kroon H, Stuefer JE, Dong M et al (1994) On plastic and
non-plastic variation in clonal plant morphology and its
ecological significance. Folia Geobot Phytotax Praha
29:123–138. doi:10.1007/BF02803790
de Steven D (1989) Genet and ramet demography of Oenocarpus mapora ssp. mapora, a clonal palm of Panamanian
tropical moist forest. J Ecol 77:579–596. doi:10.2307/
2260771
Donohue K, Foster DR, Motzkin G (2000) Effects of the past
and the present on species distribution: land-use history
and demography of wintergreen. J Ecol 88:303–316. doi:
10.1046/j.1365-2745.2000.00441.x
Eriksson O (1989) Seedling dynamics and life histories in
clonal plants. Oikos 55:231–238. doi:10.2307/3565427
335
Frost PGH (1984) The responses and survival of organisms in
fire-prone environments. In: Booysen P de V, Tainton NM
(eds) Ecological effects of fires in South African ecosystems. Springer, New York, pp 273–310
Graae BJ, Sunde PB (2000) The impact of forest continuity and
management on forest floor vegetation evaluated by species traits. Ecography 23:720–731. doi:10.1034/j.16000587.2000.t01-1-230509.x
Halpern CB (1988) Early successional pathways and the
resistance and resilience of forest communities. Ecology
69:1703–1715. doi:10.2307/1941148
Halpern CB (1989) Early successional patterns of forest species: interactions of life history traits and disturbance.
Ecology 70:704–720. doi:10.2307/1940221
Huffman DW, Tappeiner JC, Zasada JC (1994) Regeneration
of salal (Gaultheria shallon) in the central Coast Range
forests of Oregon. Can J Bot 71:39–51. doi:10.1139/
b94-006
Hughes J, Fahey TJ (1991) Colonization dynamics of herbs and
shrubs in a disturbed northern hardwood forest. J Ecol
79:605–616. doi:10.2307/2260656
Hughes J, Fahey TJ, Bormann FH (1988) Population persistence
and reproductive ecology of a forest herb Aster acuminatus.
Am J Bot 75:1057–1064. doi:10.2307/2443773
Jongman RHG, ter Braak CJF, van Tongeren OFR (eds) (1995)
Data analysis in community and landscape ecology.
Cambridge University Press, Cambridge, UK
Kanno H, Seiwa K (2004) Sexual vs. vegetative reproduction
in relation to forest dynamics in the understory shrub,
Hydrangea paniculata (Saxifragaceae). Plant Ecol 170:
43–53. doi:10.1023/B:VEGE.0000019027.88318.54
Klimes L, Klimesova J, Henriks R, et al. (1997) Clonal plant
architecture: a comparative analysis of form and function.
de Kroon H, van Groenendael J (eds) The ecology and
evolution of clonal plants. Backhuys, Leiden, pp 1–29
Kull K (1995) Growth form parameters of clonal herbs. In:
Aaviksoo K, Kull K, Paal J, Trass H (eds) Consortium
Masingii: a Fetschrfit for Victor Masing, vol 9. Tartu
university, Tartu, pp 106–115
Lee P (2004) The impact of burn intensity from wildfires on
seed and vegetative banks, and emergent understory in
aspen-dominated boreal forests. Can J Bot 82:1468–1480.
doi:10.1139/b04-108
Leps J, Smilauer P (2003) Multivariate analysis of ecological data
using CANOCO. Cambridge University Press, Cambridge
Lezberg AL, Antos J, Halpern C (1999) Belowground traits of
herbaceous species in young coniferous forests of the
Olympic Peninsula, Washington. Can J Bot 77:936–943.
doi:10.1139/cjb-77-7-936
Lezberg AL, Halpern C, Antos J (2001) Clonal development of
Maianthemum dilatatum in forests of differing age and
structure. Can J Bot 79:1028–1038. doi:10.1139/cjb-799-1028
Marino PC, Eisenberg RM, Cornell HV (1997) Influence of
sunlight and soil nutrients on clonal growth and sexual
reproduction of the understory perennial herb Sanguinaria
canadensis L. J Torrey Bot Soc 124:219–227. doi:10.2307/
2996609
Matlack G (1994) Plant species migration in a mixed-history
forest landscape in eastern North America. Ecology
75:1491–1502. doi:10.2307/1937472
336
Matlack G, Gibson DJ, Good RE (1993a) Clonal propagation,
local disturbance, and the structure of vegetation: ericaceous shrubs in the pine barrens of New Jersey. Biol
Conserv 63:1–8. doi:10.1016/0006-3207(93)90067-B
Matlack G, Gibson DJ, Good RE (1993b) Regeneration of the
shrub Gaylussacia baccata and associated species after
low-intensity fire in an Atlantic coastal plain forest. Am J
Bot 80:119–126. doi:10.2307/2445029
Matlack G, Good RE (1990) Spatial heterogeneity on the soil
seed bank of a mature coastal plain forest. Bull Torrey Bot
Club 117:143–152. doi:10.2307/2997053
Messier C, Honer TW, Kimmins JP (1989) Photosynthetic
photon flux density, red:far-red ratio, and minimum light
requirement for survival of Gaultheria shallon in western
red cedar-western hemlock stands in coastal British
Columbia. Can J For Res 19:1470–1477. doi:10.1139/
x89-223
Miller TF, Mladenhoff DJ, Clayton MK (2002) Old growth
northern hardwood forests: spatial autocorrelation and
patterns of understory vegetation. Ecol Monogr 72:487–503
Mirick S, Quinn JA (1981) Some observations on the reproductive biology of Gaultheria procumbens (Ericaceae).
Am J Bot 68:1298–1305. doi:10.2307/2442728
Moola FM (2005) Recovery and persistence of late-successional vascular plants in managed coastal Acadian forests.
Department of Biology. Dalhousie University, Halifax
195 pp
Moola FM, Mallik AU (1998) Morphological plasticity and
regeneration strategies of velvet leaf blueberry following
canopy disturbance in boreal mixed wood forests
(Vaccinium myrtilloides Michx.). For Ecol Manage
111:35–50. doi:10.1016/S0378-1127(98)00306-5
Moola FM, Vasseur L (2004) Recovery of late-seral vascular
plants in a chronosequence of post-clearcut forest stands
in coastal Nova Scotia, Canada. Plant Ecol 172:183–197.
doi:10.1023/B:VEGE.0000026326.09137.06
Moola FM, Vasseur L (2006) Compositional stability following catastrophic disturbance in coastal Acadian forests:
recovery and relationships among woody strata and
environmental factors (unpublished data)
Mosseler A, Lynds JA, Major JE (2003) Old-growth forests of
the Acadian Forest Region. Environ Rev 11:47–77. doi:
10.1139/a03-015
Motzkin G, Patterson WAIII, Foster DR (1999) A historical
perspective on pitch pine—scrub oak communities in the
Connecticut Valley of Massachusetts. Ecosystems (NY,
Print) 2:255–273. doi:10.1007/s100219900073
Norusis MJ (1992) SPSS/PC ? Professional statistics version
5.0. Chicago
O’Dea ME, Zasada J, Tappeiner JCII (1995) Vine maple clone
growth and reproductive in managed and unmanaged
coastal Oregon Douglas-fir forests. Ecol Appl 5:63–73.
doi:10.2307/1942052
Peterken GF, Game M (1984) Historical factors affecting the
number and distribution of vascular plant species in the
woodlands of central Lincolnshire. J Ecol 72:155–182.
doi:10.2307/2260011
Peterson CJ, Jones RH (1997) Clonality in woody plants: a
review and comparison with clonal herbs. In: de Kroon H,
A.G. Van der Valk (ed.)
van Groenendael J (eds) The ecology and evolution of
clonal plants. Backhuys Press, Leiden, pp 263–289
Pitelka LF, Stanton DS, Peckenham MO (1980) Effects of light
and density on resource allocation in a forest herb, Aster
acuminatus (Compositae). Am J Bot 67:942–948. doi:
10.2307/2442435
Price EAC, Marshall C (1999) Clonal plants and environmental
heterogeneity. Plant Ecol 141:3–7. doi:10.1023/A:100983
8300691
Ramovs BV, Roberts MR (2003) Understory vegetation and
environment responses to tillage, forest harvesting, and
conifer plantation development. Ecol Appl 13:1682–1700.
doi:10.1890/02-5237
Reyes G (2002) Factors affecting vegetation re-establishment
patterns following clearcutting of coastal red spruce-balsam fir stands in southwestern Nova Scotia. Biology.
Dalhousie University, Halifax
Roberts MR (2004) Response of the herbaceous layer to natural
disturbance in North American forests. Can J Bot
82:1273–1283. doi:10.1139/b04-091
Roberts MR, Ramovs BV (2005) Response of plant functional
groups within plantations and naturally regenerated forests in southern New Brunswick, Canada. Can J For Res
35:1261–1276. doi:10.1139/x05-049
Roberts MR, Zhu L (2002) Early response of the herbaceous
layer to harvesting in a mixed coniferous-deciduous
forest in New Brunswick, Canada. For Ecol Manage
5594:1–15
Rowe JS (1972) Forest regions of Canada. Department of the
Environment, Canadian Forestry Service, Ottawa
Silva JF, Kana TM, Solbrig OT (1982) Shoot demography in
New England populations of Mainthemum canadense
Desf. Oecologia 52:181–186. doi:10.1007/BF00363834
Sobey DG, Barkhouse P (1977) The structure and rate of
growth and rhizomes of some forest herbs and dwarf
shrubs of the New Brunswick—Nova Scotia border
region. Can Field Nat 91:377–383
SPSS (1995) Version 6.1.3. SPSS Inc, Chicago
Strang RM (1970) The ecology of the rocky heathlands of
western Nova Scotia. Annual tall timbers fire ecology
conference. Tall Timbers Research Station, Tallahassee
Strang RM (1972) Ecology and land use of the barrens of
western Nova Scotia. Can J For Res 2:276–290. doi:
10.1139/x72-045
Sun S, Cai Y, An S (2002) Differences in morphology and
biomass allocation of Scirpus mariqueter between creekside and inland communities in the Changjiang estuary,
China. Wetlands 22:786–793. doi:10.1672/0277-5212
(2002)022[0786:DIMABA]2.0.CO;2
Tappeiner J, Zasada J, Ryan P et al (1991) Salmonberry clonal
and population structure: the basis for a persistent cover.
Ecology 72(2):609–618. doi:10.2307/2937201
Tappeiner JC, Alaback PB (1989) Early establishment and
vegetative growth of understory species in the western
hemlock–Sitka spruce forests of southeast Alaska. Can J
Bot 67:318–326. doi:10.1139/b89-046
ter Braak CJF, Smilauer P (1999) Canoco for windows version
4.02. Centre for Biometry Wageningen, CPRO-DLO,
Wageningen
Forest Ecology
Vander Kloet SP, Hill NM (1994) The paradox of berry production in temperate species of Vaccinium. Can J Bot
72:52–58. doi:10.1139/b94-007
Whitman AA, Hunter MLJ, Witham JW (1998) Age distribution of ramets of a forest herb, wild sarsaparilla, Aralia
nudicaulis (Araliaceae). Can Field Nat 112:37–44
337
Whitney GG, Foster DR (1988) Overstorey composition and
age as determinants of the understorey flota of woods of
central New England. J Ecol 76:867–876. doi:10.2307/
2260578
Zar JH (1996) Biostatistical analysis. Prentice Hall, Upper
Saddle River, NJ
Species richness and resilience of forest communities:
combined effects of short-term disturbance and long-term
pollution
Marina R. Trubina
Originally published in the journal Plant Ecology, Volume 201, No. 1, 339–350.
DOI: 10.1007/s11258-008-9558-z Springer Science+Business Media B.V. 2008
Abstract Recovery of the species richness of plant
communities after experimental disturbances of various severities were studied in spruce forests polluted
by atmospheric entry of SO2 and heavy metals from a
copper smelter. In the three toxic load zones (impact,
buffer, and background), 60 experimental ‘‘pit-andmound’’ complexes (sized 1 m 9 2 m, 20 complexes
in each zone) were created. Colonization of disturbed
areas by vascular plants was observed during a 6-year
period after the disturbance. The results showed that
the recovery processes were affected by disturbance
severity and that the recovery differed significantly
among the communities. In all of the zones, species
richness increased rapidly after mild disturbance. In
degraded communities, levelling of differences in the
rate of colonization after mild and severe disturbances was observed. The highest colonization rate
was found in the communities of background zone,
while the lowest was found in the heavily degraded
communities of impact zone. The disturbances
significantly increased the species diversity of communities in all zones and caused a certain reversion of
degraded communities to previous stage of anthropogenic succession. Mild disturbance promoted the
greatest increase in the diversity indices. The study
M. R. Trubina (&)
Institute of Plant and Animal Ecology, Ural Division,
Russian Academy of Science, 8th Street 202,
620144 Ekaterinburg, Russia
e-mail: mart@ipae.uran.ru
results indicate that recovery rate of species richness
of plant communities is determined by the duration of
negative effect of disturbances. Recovery also
depends significantly on the magnitude and endurance of positive effect of disturbances. The studied
communities differed significantly in these parameters. The study results also suggest that short-term
disturbances can significantly modify the process of
transformation of plant communities by atmospheric
pollution. On the other hand, long-term pollution can
considerably modify the response of forest communities to disturbances. The results also conclude that
the resilience of communities does not exclusively
depend on their species richness.
Keywords Copper smelter Spruce forests
Species diversity Recovery Succession
Vascular plants
Introduction
Short-term disturbances (windfalls, fires, eruptions,
etc.) constitute an inalienable part of natural ecosystems and they play an important role in their
structural organization and dynamics (Georgievsky
1992; Falinski 1978; Frelich and Reich 1999; Forest
et al. 1998; Kuuluvainen 1994; Peterson and Pickett
1995; Skvortsova et al. 1983; Turner et al. 1998;
Ulanova 2000). Currently, many ecosystems are
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_25
339
340
subject to long-term anthropogenic stress exposure,
which had led to significant decrease in biodiversity,
and to changes in their composition and structure.
Such changes can even modify ecosystem functions
(Hooper et al. 2005; Peterson et al. 1998), e.g., their
responses to natural disturbance. In this regard,
questions concerning the functional after-effects of
biodiversity decrease, and the changes in the composition and structure of the communities, are
becoming issues of current importance.
An important component of the general stability of
communities is resilience (recovery ability), in particular the time needed to return to the initial state
after some perturbation. The majority of community
resilience investigations have been carried out in
grasslands (Lavorel 1999; McNaughton 1977; Tilman
and Downing 1994; Whitford et al. 1999; see also
review Hooper et al. 2005) or in forest ecosystems
(Skvortsova et al. 1983; Cooper-Ellis et al. 1999;
Hautala et al. 2001, 2008; Jonsson and Essen 1998;
Mayer et al. 2004; Peterson and Campbell 1993;
Rydgren et al. 2004), which had not been affected by
atmospheric pollution. Meanwhile, environmental
pollution is one of the most wide-ranging types of
long-term anthropogenic stresses. It has caused
significant decreases in biodiversity and also changes
in the composition and structure of the forest
ecosystems (Vorobeichik et al. 1994; kompleksnaya
ecologicheskaya 1992; lesnye ekosystemy 1990;
Lukina and Nikonov 1993; Makhnev et al. 1990;
Smith 1985; Trubina and Makhnev 1997; Chernenkova 2002; Bobbink et al. 1998; Freedman and
Hutchinson 1980; Lee 1998; Salemaa et al. 2001).
Recovery rate of forest communities depends on
their diversity, composition, and structure at the time
of the disturbance (Forest et al. 1998; Jonsson and
Essen 1998; Hooper et al. 2005; Mayer et al. 2004;
Peterson and Campbell 1993; Rydgren et al. 2004;
Skvortsova et al. 1983; Turner et al. 1998). In a
degraded forest ecosystems, one can expect certain
changes in recovery, e.g., the decrease in colonization
rate and recovery ability of species richness on the
whole. On the other hand, under atmospheric pollution, short-term disturbances may also considerably
modify processes of ecosystem transformation, e.g.,
they may accelerate or slow down the decreasing
species richness.
These assumptions were examined experimentally
in the vicinity of a functioning copper smelter. Spruce
A.G. Van der Valk (ed.)
forests of the area have been affected by long-term
emissions of heavy metals and sulfur dioxide since
1940. In 1999, 60 pit-and-mound complexes, partly
imitating treefall disturbance, were created in the
area. Recovery of vegetation on the plots was
followed during 6 years. The objectives of this work
were: (1) to describe the recovery of vascular plant
species richness after different levels of disturbance
severity, and (2) to evaluate the degree of modifying
influences of the short-term disturbances on species
richness of forest plant communities under long-term
pollution.
Methods
The investigation was carried out in the vicinity of a
copper smelter located near the town of Revda,
50 km west of Ekaterinburg in the Middle Urals. The
area belongs to the southern taiga phytogeographical
subzone with a forest cover of about 60% consisting
mostly of secondary forests with mixed coniferous
and deciduous trees as well as birch and aspen stands
(Kolesnikov et al. 1973). The most common dominants of the overstorey include Pinus sylvestris L.,
Picea obovata (Ledeb.), Betula pendula (Roth.),
B. pubescens (Ehrh.), Abies sibirica (Ledeb.), and
Populus tremula L. Climate of the area is moderately
continental, annual precipitation level is 400–
600 mm on average, while depth of snow cover is
40–50 cm and more. Average annual temperature is
?1C, while during January and July temperatures
are between -16C to -17C and ?16C to ?18C,
respectively. The length of frost-free period is
90 days and the prevailing winds are westerly and
southwesterly (Prokaev 1976).
The copper smelter has been operating since 1940,
emitting a mass of particulate and gaseous pollutants
(in a ratio of 1:8), of which sulfur dioxide makes up
98.7% of the gaseous pollutants, while copper, zinc,
arsenic, and lead constitute 46.9%, 31.5%, 11.5%,
and 10.1% of the particulate pollutants, respectively
(Vorobeichik et al. 1994). The study sites were
established in coniferous forest stands in three zones,
which had been determined earlier according to the
degree of forest community transformation in previous investigations (Vorobeichik et al. 1994): the
impact—the zone of high load (distance from the
source of emission 1 km), the buffer—the zone of
Forest Ecology
intermediate load (6 km), and the background—the
zone of low load (30 km). The soils of the sites are
mountain forest brown soils. The tree layer has a
multilayered canopy and an uneven age structure.
The dominating tree species include Siberian spruce
(Picea obovata) and Siberian fir (Abies sibirica), of
which a few are up to 140 years old. In addition,
some deciduous trees occur, of which silver birch
(Betula pendula) is the most abundant. Detailed
information on the contents of toxicants, soil characteristics, and changes in forest communities in these
areas have been described earlier in the publications
by Gol’dberg (1997), Kaigorodova and Vorobeichik
(1996), Vorobeichik et al. (1994), and Vorobeichik
and Hantemirova (1994).
Ground vegetation of the background zone consists
of a mix of the following dominants and co-dominants: Oxalis acetosella, Aegopodium podagraria,
Gymnocarpium dryopteris, Dryopteris carthusiana,
Asarum europaeum, Majanthemum bifolium, Cerastium
pauciflorum, and Calamagrostis obtusata. Ground vegetation of the buffer zone consists of Oxalis acetosella,
Cerastium pauciflorum, Majanthemum bifolium, Carex
montana, Calamagrostis obtusata, Rubus saxatilis, and
Rubus idaeus and in the impact zone consists of
Equisetum sylvaticum, Agrostis tenuis, Calamagrostis
arundinacea, Calamagrostis langsdorffii, Chamerion angustifolium, and Majanthemum bifolium. The
total cover (sum of the projective cover of all species, ± SE) of vascular plants on the undisturbed sites
of the background, the buffer, and the impact zones was
152.4 ± 3.0%, 78.5 ± 5.6%, and 5.9 ± 0.8%, respectively, and the variation coefficient along the toxic
gradient increased from 16% to 126%.
In the beginning of August 1999, altogether 20
complexes of disturbed plots (sized 1 m 9 2 m) were
created in each load zone. The distance between the
complexes was 5–10 m. The experimental plots were
designed to imitate the pit-and-mound complexes that
tend to form after tree uprooting (Liechty et al. 1997;
Peterson et al. 1990). The 1 m2 pits were created by
removing all vegetation and the topsoil from the
depth of 20 cm. The disturbance was thought to
represent a severe one, as it did not only destroy the
vegetation, but also removed the diaspore bank and
baring of the mineral soil horizons. The excavated
topsoil was deposited near the pits in an area of 1 m2.
These plots imitated the mounds and represented
mild disturbance (partial death of plants, preservation
341
of soil diaspore bank, and favorable physical and
chemical substrate properties).
The species richness (which was measured by
counting the number of species) was recorded from
1 m 9 1 m plots during the 6-year post-disturbance
period (2000–2005). The species richness of the plots
that had no visible signs of natural disturbances (the
undisturbed plots) was measured twice, during 2000
and 2005. In the year 2000, species richness was
estimated at 140 undisturbed plots (sized 1 m 9 1 m).
The plots were located within a radius of 500–700 m
from the experimental complexes. In the year 2005, 60
similar undisturbed plots (20 plots per toxic load zone)
were studied within a radius of 50 m from the
experimental complexes. As species richness of the
undisturbed plots was similar during the two years,
their mean values were used in the further analysis.
One-way analysis of variance (ANOVA) was used
to test the differences in species richness between
disturbed and undisturbed plots in the each toxic load
zone. Two-way mixed-effects ANOVA was used to
test the importance of toxic load influence (random
factor), disturbance severity (fixed factor), and their
interactions during the different study years. Twoway repeated-measurements ANOVA was used to
estimate the influence of toxic load zone, time,
disturbance severity, and their interactions. The
multiple comparisons method (Scheffe’s test) was
used to test the differences between means in the
three toxic load zones. Before the variance analysis,
square-root transformation was applied to the data.
Results
Species richness of vascular plants in the undisturbed
plots of the background zone was 2 and 10 times
higher compared with the buffer and the impact
zones, respectively (Table 1). Besides, a significant
increase in the space variation of indices along the
toxic gradient was observed.
The colonization rate of disturbed plots differed
between the toxic load zones (F2,114 = 69.04;
P \ 0.001). During the study period, complexes of
the background zone had the highest average species
richness (Fig. 1), while the lowest values were found
in the impact zone. The colonization rate also significantly depended on the degree of disturbance severity
(F1,114 = 67.95; P \ 0.001). In all of the toxic load
342
A.G. Van der Valk (ed.)
Table 1 Mean number (±SE) of species and variation coefficient (CV) of index at the undisturbed forest floor patches in
the different toxic load zones
Indices
Toxic load zone
Background
Buffer
Impact
Number of species
per m2
12.2 ± 0.4a
6.2 ± 0.5b
1.1 ± 0.1c
CV (%)
27.9a
58.9b
75.7c
Values followed by the same letter are not significantly
different at an overall P \ 0.05
Fig. 1 Change in the number of species (mean values, m–2)
during the study period on the experimentally distributed
complexes sited in different toxic load zones. Symbols: (1)
disturbed and (2) undisturbed plots of background zone; (3)
disturbed and (4) undisturbed plots of buffer zone; (5)
disturbed and (6) undisturbed plots of impact zone
Fig. 2 Change in the number of species (mean values, m–2)
during the study period after a mild and b severe disturbances
in the different toxic load zones. Symbols: (1) disturbed and
(2) undisturbed plots of background zone; (3) disturbed and (4)
undisturbed plots of buffer zone; (5) disturbed and (6)
undisturbed plots of impact zone
zones, species richness increased more slowly after
severe disturbance (Fig. 2b) than after mild disturbance (Fig. 2a). The successional colonization
processes also differed significantly between the toxic
load zones of (zone 9 time F10,570 = 11.16;
P \ 0.001). Maximum colonization rate was recorded
in all toxic load zones within 1 year after the
disturbance. In the background zone, high colonization
rate was evident for three post-disturbance years,
while in the buffer zone—for 1 year, and in the impact
zone—for 2 years after mild disturbance and only
1 year after severe disturbance.
The toxic load zones modified significantly the
colonization processes after severe and mild disturbance (zone 9 type of disturbances 9 time F10,570 =
7.06; P \ 0.001). Levelling of the differences between
the colonization rates after severe and mild disturbance
was observed along the toxic gradient, especially
during the first post-disturbance year. The mean (±SE)
species richness after severe and mild disturbances was
3.95 ± 0.47 and 10.9 ± 0.74 in the background zone,
4.00 ± 0.54 and 6.7 ± 0.77 in the buffer zone,
2.80 ± 0.31, and 3.30 ± 0.33 species in the impact
zone. Six years after the disturbance, the greatest
differences between the indices of severely and mildly
disturbed plots were observed in the background zone,
and the lowest in the buffer zone.
One year after disturbance, species richness of the
mildly disturbed plots in the background zone already
did not differ from values of the undisturbed plots
(F1,83 = 2.12; P \ 0.149). In the subsequent years,
Forest Ecology
the species richness of mildly disturbed plots was
significantly higher than in the undisturbed ones
(P \ 0.001). Six years after the disturbance, species
richness was significantly lower in severely disturbed
plots than in undisturbed plots (F1,83 = 6.52;
P \ 0.013). On the whole, the species richness of
the complexes in the background zone did not differ
from the undisturbed areas by the third post-disturbance year (F1,103 = 1.02; P \ 0.314).
One year after disturbance, species richness of the
mildly disturbed plots in the buffer zone did not differ
from the undisturbed areas (F1,73 = 0.26;
P \ 0.610). In the subsequent years, the indices were
only slightly higher if compared with undisturbed
plots. By the fourth post-disturbance year, the
severely disturbed plots did not differ significantly
in their species richness from the undisturbed ones
(F1,73 = 1,66; P \ 0.202). In the buffer zone, 1 year
after disturbance, species richness of the complexes
did not differ from that in the undisturbed areas
(F1,93 = 1.44; P \ 0.233).
One year after disturbance, the species richness of
both mildly and severely disturbed plots in the impact
zone was already higher than in the undisturbed ones
(F1,98 = 63.48; P \ 0.001 and F1,98 = 47.92;
P \ 0.001, respectively). The same trend continued
through the subsequent years in spite of a certain
decrease after the first (severe disturbances) and the
second (mild disturbances) years of succession.
Six years after the disturbance, complexes of the
impact zone had significantly higher species richness
than in the undisturbed areas (F1,118 = 88.01;
P \ 0.001).
The pattern of species richness dynamics in the
total area of 20 m2 (Fig. 3) was very similar to that in
the area of 1 m2 (Fig. 2). An important observation
was that, from the second year after the disturbance
onwards, species richness of the mildly disturbed
buffer zone plots on the mesoscale (20 m2) was close
to the values of the undisturbed plots in the
background zone (Fig. 3a). However, on the microscale (1 m2), reversion to the previous succession
state was not clearly pronounced (Fig. 2a). Species
richness of the mildly disturbed impact zone plots
was close to the values of the undisturbed plots in the
buffer zone on both scales.
The total number of species found in the experimentally disturbed plots was considerably higher than in the
undisturbed areas in all toxic load zones (Table 2). The
343
Fig. 3 Change in the total number of species on the area
20 cm2 during the study period after a mild and b severe
disturbances in the different toxic load zones. Symbols:
(1) disturbed and (2) undisturbed plots of background zone:
(3) disturbed and (4) undisturbed plots of buffer zone; (5)
disturbed and (6) undisturbed plots of impact zone
highest number of species was registered in the mildly
disturbed areas. After severe disturbance, the total
number of species in the background, buffer, and impact
zones increased by 13%, 45%, and 72%, respectively,
and by 50%, 50%, and 85% after the mild one,
respectively. The proportion of the species registered in
the disturbed patches of the background, buffer, and
impact zones made up 51.3%, 57.4%, and 85.3% of the
total number of species in each zone, respectively. The
total number of species that were found from the treated
plots decreased along the toxic gradient and the species
composition was different (Table 3). The disturbed
plots in the background zone had mainly ruderal
(e.g., Plantago major, Taraxacum officinale, Tussilago
farfara, Urtica dioica), open (e.g., Agrostis tenuis,
Alchemilla sp., Coronaria flox-cuculi, Lathyrus
344
A.G. Van der Valk (ed.)
Table 2 Total number of species observed in experimentally
disturbed and undisturbed forest floor patches in the different
zones of pollution
Type of site
Toxic load zone
Background
Buffer
Impact
Undisturbed (20 m2)
39
29
5
Severe disturbances (20 m2)
45
53
18
Mild disturbances (20 m2)
78
58
33
80
68
34
2
Total (60 m )
pratensis, Potentilla erecta, Prunella vulgaris) or
wet (e.g., Calamagrostis langsdorfii, Crepis paludosa, Geum rivale, Filipendula ulmaria, Ranunculus
repens) habitat species, but also pioneer species of the
initial stages of secondary succession (e.g., Chamerion angustifolium, Chrysosplenium alternifolium,
Phegopteris connectilis) along with pine forest species
(e.g., Calamagrostis arundinaceae, Betonica officinalis,
Vicia sylvatica, Viola canina).
The disturbed plots of the degraded communities
had species from the aforementioned groups, and also
common spruce forest species, such as Athyrium filixfemina, Cerastium pauciflorum, Dryopteris carthusiana, Gymnocarpium dryopteris, Galium odoratum,
Luzula pilosa, etc. The proportion of these species in
the disturbed patches of buffer and impact zones made
up 28% and 40% of the total number of new species in
each toxic load zone, respectively.
Discussion
Peculiarities of species richness recovery after
disturbance
The results of this study suggest that species richness
of communities decreases under long-term atmospheric pollution by heavy metals and sulfur dioxide
and that this is accompanied by a significant decrease
in the colonization rate of the disturbed plots. The
result is quite expectable, especially if we take into
account the direct dependence of the colonization
processes on the initial state of communities before
disturbance (Forest et al. 1998; Hooper et al. 2005;
Jonsson and Essen 1998; Mayer et al. 2004; Peterson
and Campbell 1993; Rydgren et al. 2004; Skvortsova
et al. 1983; Turner et al. 1998).
The highest colonization rate in all of the toxic
load zones was observed during the first postdisturbance year, but the investigated communities
differed in the length of their intensive recovery
periods. The colonization rate in the background zone
decreased after 3 years of succession, which supports
the view of other researches on the succession rate in
unpolluted forests (Rydgren et al. 2004). The colonization rate in the degraded communities decreased
already during the first (or the second) year of
succession. The observed phenomenon cannot be
related to the lack of colonization space, as the
vegetation cover of the disturbed plots at the initial
stages of succession was rather low (Trubina 2003). It
is more likely caused by very fast exhaustion of the
available diaspore bank in the degraded communities.
The phenomenon is especially exhibited in the zone
of heaviest contamination (impact zone) and under
severe disturbance, i.e., when the diaspore bank (the
additional resource for colonization) is absent.
Species richness increased significantly more
slowly after severe disturbance than after mild
disturbance. The observed data agree perfectly with
results of other investigations concerning the influence of disturbance severity on colonization processes
(Hautala et al. 2001, 2008; Jonsson and Essen 1998;
Mayer et al. 2004; Peterson and Campbell 1993;
Rydgren et al. 2004; Skvortsova et al. 1983). This
phenomenon is most likely induced by the almost
complete destruction of the soil bank of vegetative
and generative diaspores of the soil bank after a severe
disturbance, as the importance of soil bank in the
processes of colonization is known to be very high
(Jonsson and Essen 1998; Mayer et al. 2004; Putz
1983; Rydgren et al. 2004; Turner et al. 1998). It is
also possible that the removal of upper layers of soil
(deterioration of the physical and chemical properties
of the substrate) might have reduced colonization and
survival of plants, but the importance of the latter
factor can hardly be assessed within this experiment.
Increasing levels of toxic loads and degree of
ecosystem degradation led to levelling of differences
in the colonization rate after severe and mild disturbance. Weakening of the disturbance severity effect
along the toxic load gradient may be related to several
reasons. Some investigations (Komulainen et al.
1994) have proved that, even if viable seeds are
present in the degraded communities, colonization
may be impeded by very high levels of soil toxicity. In
Forest Ecology
345
Table 3 List of species found on the disturbed plots in the three toxic load zones
Zone of pollution
Background
Buffer
Impact
Agrostis tenuis
Adenophora lilifolia, M
Adenophora lilifolia, M
Alchemilla sp., M
Adoxa moschatellina, M
Ajuga reptans, M
Angelica sylvestris, M
Agrostis tenuis
Athyrium filix-femina, M
Betonica officinalis, M
Alchemilla sp.
Carduus sp., M
Calamagrostis arundinaceae, M
Athyrium filix-femina
Carex sp.
Calamagrostis langsdorfii, M
Cacalia hastate, M
Cerastium pauciflorum
Carex montana
Cerastium sp., S
Deschampsia cespitosa
Carex sp.
Chrysosplenium alternifolium
Circaea alpina
Chamerion angustifolium, M
Coronaria flos-cuculi
Coronaria flos-cuculi
Chrysosplenium alternifolium
Dactylis glomerata, S
Dryopteris carthusiana
Cirsium sp., M
Dryopteris filix-max
Fragaria vesca, M
Coronaria flos-cuculi
Fragaria vesca
Galium sp., M
Crepis paludosa, M
Galeopsis bifida
Gymnocarpium dryopteris,
Dryopteris filix-max, M
Galium odoratum
M
Filipendula ulmaria, M
Galeopsis bifida
Galium uliginosum
Geranium sylvaticum, M
Impatiens noli-tangere, M
Lathyrus pratensis
Galium uliginosum
Gymnocarpium dryopteris
Lathyrus vernus, M
Geranium sylvaticum, M
Juncus sp., M
Luzula pilosa
Geum rivale
Lathyrus pratensis
Majanthemum bifolium
Glechoma hederacea, S
Phegopteris connectilis
Melica nutans, M
Goodyera repens, M
Poa sp., M
Phegopteris connectilis
Hieracium sp., M
Prunella vulgaris
Poa sp.
Lathyrus pratensis
Pulmonaria dacica
Ranunculus repens, M
Luzula palescens, M
Pyrola media, M
Rubus idaeus
Orthilia secunda, M
Silene sp.
Stellaria nemorum
Phegopteris connectilis
Solidago virgaurea
Taraxacum officinale
Plantago major, M
Stellaria sp., S
Thalictrum minus, M
Poa nemoralis
Stellaria nemorum
Tussilago farfara
Poa sp., M
Taraxacum officinale
Urtica dioica, M
Potentilla erecta, M
Prunella vulgaris
Thalictrum minus
Trientalis europaea
Veratrum lobelianum, M
Viola sp.
Ranunculus borealis, M
Tussilago farfara
Ranunculus cassubicus, M
Urtica dioica, M
Ranunculus repens, M
Valeriana wolgensis
Stachys sylvatica
Veratrum lobelianum, M
Taraxacum officinale
Veronica chamaedrys
Tussilago farfara
Vicia sylvatica
Urtica dioica
Viola canina, S
Veronica chamaedrys
Viola selkirkii
Vicia sylvatica, M
Viola canina, M
Species occurring at mildly disturbed plots are marked with M; S marks species of severely disturbed plots; the rest of the species
occurred at both types of disturbance
346
high-toxicity areas, colonization of vacant areas may
also be impeded by a very thick layer of forest litter
(more than 5 cm), which is typical for the degraded
forest ecosystems near this particular copper smelter
(Vorobeichik 1995). The negative influence of litter
on regeneration and survival of plants, composition,
and species richness of communities has been shown
in a range of studies (Peterson and Campbell 1993;
Sydes and Grime 1981a, b; Xiong and Nilsson 1999;
Weltzin et al. 2005; Sannikov 1992). Apparently,
disturbance of the thick and highly toxic forest litter
and exposure of the less contaminated soil lessens the
negative effect of the disturbance severity and led to
its weakening. The levelling of differences may also
result from a significant decrease of soil seed bank
diversity in the degraded communities, which often
takes place under long-term pollution (Ginocchio
2000; Meerts and Grommesch 2001; Salemaa and
Uotila 2001). The low number of post-disturbance
species in the heavily degraded communities
(Table 3) also confirms indirectly this supposition.
Another reason for the observed phenomenon may be
in the composition of these communities, e.g., the
high proportion of eurytopic species that are quite
indifferent to unfavorable conditions of the substrate,
but these questions would require a special study.
In the course of the recovery period, species
richness of the disturbed areas at certain stages was
significantly higher than in the undisturbed area. This
phenomenon is quite typical for post-disturbance
successions after disturbances in most forest communities (Barik et al. 1992; Cooper-Ellis et al. 1999;
Goldblum 1997; Mayer et al. 2004; Peterson and
Campbell 1993; Skvortsova et al. 1983) and implies
the presence of at least two time points during which
disturbed and undisturbed areas of communities will
not differ significantly in their species richness. The
first of these time points characterizes the duration of
the negative effects of the disturbances and is a very
important component of the resilience of communities, in addition to the period of positive effects and
the time of the final return (TR) to the initial state. The
negative-effect period, as a rule, seems to be shorter
after mild disturbance than after severe disturbance,
which is not unexpected, as severity of disturbance
largely defines colonization rate. Differences in the
duration of the negative effect after severe and mild
disturbance were greatest in the background zone and
were not observed in the impact zone. The possible
A.G. Van der Valk (ed.)
reasons for levelling of the differences between the
effects of severe and mild disturbances along the toxic
load gradient have been discussed above. However, the
following important points should be emphasized.
The duration of the negative effect of mild
disturbance was equal in the investigated communities. After severe disturbance, this period was
significantly longer in the background zone than in
the intermediately and heavily degraded communities,
despite the highest species richness and colonization
rates in the background zone. The results prove that it
is not only the species richness of communities that
determines the duration of the negative effects. It is
quite possible that the composition of communities or
functional diversity (Hooper et al. 2005) also plays a
decisive role in the rate of recovery. In the degraded
communities, the prevalence of pioneer species and
clonal plants (see above) with high colonization
abilities could have promoted faster overcoming of
the negative effects of disturbance.
Data from a few studies (e.g., Skvortsova et al.
1983) provide evidence that high species richness can
be preserved in disturbed areas for several decades.
Therefore, longer-term data is needed to determine
the time of the final return to the initial state.
Nevertheless, some suppositions about recovery of
species richness after disturbance can be made from
the given index rate of recovery changes and its
degree of deviation from the indices measured from
the undisturbed areas. After the 6 years of recovery
from mild disturbance, the greatest positive deviation
and the lowest rate of recovery were found in the
communities of the impact zone. One can suppose
that, after mild disturbance, communities of this zone
will return to the initial state more slowly than
communities of less contaminated habitats. However,
on the basis of the present data, hardly any suppositions can be made for the communities in the buffer
and background zones. However, if we consider the
complexes of the disturbed areas in general, and
suppose that the species richness of disturbed plots
will not increase significantly in the course of further
succession, we may conclude that the highest recovery rate of the initial species richness prevails in the
buffer-zone communities (TR = 1 year on observed
time interval), a lower rate prevails in the background-zone communities (TR = 3 years), and the
lowest rate prevails in the communities in the impact
zone (TR 6 years).
Forest Ecology
In general, the results suggest that the period of
final return to the initial state depends on the duration
of the negative effects of disturbance. It also depends
on the magnitude and duration of the positive effects.
The heavily degraded communities were characterized by the shortest period of negative effects of
disturbance, the greatest magnitude and duration of
positive effects, and as a result the greatest deviation
of species richness from the indices of the undisturbed areas after 6 years of recovery. The
intermediately degraded communities were characterized by an average duration of negative effect of
disturbances, the lowest magnitude and duration of
positive effects, and the highest recovery rate of
species richness on the whole. Recovery of the least
degraded communities occurred at an intermediate
rate owing to the longest period of negative effects
after severe disturbance. This fact once again confirms that resilience of communities is not determined
merely by their species richness.
Influence of short-term disturbances of various
severity levels on species richness of communities
under long-term pollution
In this study experimental disturbance promoted a
considerable increase in the total number of species in
the investigated communities. The observed results
coincide especially with the study results on the
influence of windfall disturbances to diversity of forest
communities (Barik et al. 1992; Cooper-Ellis et al.
1999; Goldblum 1997; Kuuluvainen 1994; Peterson
and Campbell 1993; Skvortsova et al. 1983; Ulanova
2000). At the same time, the degree of modifying
influence of disturbances depended on their severity, as
well as on the community degradation level.
The increase in species richness in the backgroundzone communities derived from the emergence of
open and wet habitat species, ruderals, pioneer species
of initial stages of secondary successions or species of
other forest communities. The composition of species
growing in the disturbed sites of the background zone
is in agreement with data cited for the initial stages of
succession after treefalls in unpolluted areas (Skvortsova et al. 1983; Ulanova 2000). The positive effects
of disturbance increased considerably along the
gradient of toxic load. In the impact zone, the total
number of species in the communities after the
disturbances became seven times higher than in
347
undisturbed areas. Moreover, species richness
increases at the expense of both the characteristic
species of disturbed patches in the unpolluted forest
and of the most common unpolluted spruce forest
species that disappear from the communities as a
result of long-term pollution. Interestingly, a similar
phenomenon was observed in the studied communities after treefalls (unpublished data).
Such a large increase in the diversity of degraded
communities may be related to the preservation of
sufficiently diverse and viable diaspore bank in the
degraded communities, as well as to dispersal of
diaspores from outside. However, the role of the latter
factor may be small, as diversity and total cover of
vegetation were extremely low in this load zone.
It has been proved earlier that diaspore bank can
remain viable under long-term pollution by coppersmelter emissions (Ginocchio 2000; Huopalainen
et al. 2001; Komulainen et al. 1994; Meerts and
Grommesch 2001; Salemaa and Uotila 2001).
Increased species richness after mild disturbances
proves indirectly that the diaspore bank can preserve
certain diversity in heavily polluted zone. The results
also suggest that, after 60 years of copper smelter
functioning, only partial recovery of species richness
from the diaspore bank is possible in highly degraded
communities.
Mild disturbances decreased diversity differences
between the studied communities on both micro- and
mesoscale, and caused a certain reversion of the
degraded communities to the earlier state of anthropogenic succession. Particularly, after the second
post-disturbance year, species richness of mildly
disturbed plots in the impact zone was close to the
level observed in undisturbed plots of the buffer zone
on both scales. In the buffer zone, reversion to the
indices of undisturbed plots of the background zone
was observed only on the mesoscale. This is most
likely due to the different species elimination rates
from the communities on the micro- and mesoscale.
Particularly, the number of species in undisturbed
1 m2 plots (microscale) in the buffer zone was 2 times
lower compared with the background zone, whereas
in the area 20 m2 (mesoscale) was lower only 1.3
times. It was remarkable that severe disturbance only
slightly modified species richness of the communities
on the micro- and mesoscale and increased only the
total number of species. Moreover, the positive effect
of severe disturbance on the total number of species
348
was less pronounced than that of mild disturbance,
which again emphasizes the importance of diaspore
bank in the maintenance of high community
diversity.
Thus, our results suggest that short-term smallscale disturbance promotes species richness in plant
communities and that, under strong long-term pollution, it slows down the decline in species richness and
may also lead to certain reversion of degraded
communities to the previous state of succession.
However, the following important points should be
emphasized. Short-term disturbances, e.g., treefalls,
at the initial pollution stages can promote elimination
of typical forest species, because during the first years
of succession disturbed plots are actively occupied by
untypical species (Kuuluvainen 1994; Nakashizuka
1989; Rydgren et al. 2004; Skvortsova et al. 1983;
Ulanova 2000). The duration of negative effects after
severe disturbance in the background zone proves
that severe disturbance of vegetation and soil, such as
tree uprooting, could have led to the significant and
long-term decrease in species richness of communities on both micro- and mesoscales. In other words,
severe disturbance can accelerate the processes of
diversity decline in communities during the initial
stages of environmental contamination. The positive
effect of disturbance that is currently observed in the
degraded communities may subsequently decrease or
disappear due to the continuing input of toxicants. In
particular, germination of spores and seeds of species
intolerant to contamination and their subsequent
elimination because of persisting toxic influence will
lead to the exhaustion of the diaspore bank and to the
further decrease in diversity and recovery abilities of
the degraded communities as a whole.
Acknowledgements I am grateful to Sergey Kartavov for his
help in the creation of the experimental plots, and Irina
Mikhailova and Harri Hautala for revising the English of the
manuscript. This study was completed under the financial
support of the Russian Foundation for Basic Research (project
no. 08-04-91766-AF) and the program of the Russian Academy
of Sciences ‘‘Biodiversity and dynamics of genofunds’’
(project ‘‘Biodiversity changes and mechanisms of the
terrestrial ecosystems’ resistance along toxic load gradients’’).
References
Alexeyev VA (ed) (1990) Lesnye ekosystemy i atmosfernoe
zagryaznenie. Nauka, Leningrad
A.G. Van der Valk (ed.)
Barik SK, Pandey HN, Tripathi RS, Rao P (1992) Microenvironmental variability and species-diversity in treefall
gaps in a subtropical broadleaved forest. Vegetatio
103(1):31–40
Bobbink R, Hornung M, Roelofs JGM (1998) The effect of airborne nitrogen pollutants on species diversity in natural
and semi-natural European vegetation. J Ecol 86:717–738.
doi:10.1046/j.1365-2745.1998.8650717.x
Chernenkova TV (2002) Reaktsiya lesnoy rastitel’nosti na
promyshlennoe zagryaznenie. Nauka, Moskow
Cooper-Ellis S, Foster DR, Carlton G, Lezberg A (1999) Forest
response to catastrophic wind: results from an experimental hurricane. Ecology 80(8):2683–2696
Falinski JB (1978) Uprooted trees, their distribution and
influence in primeval forest biotope. Vegetatio 38(3):
175–183. doi:10.1007/BF00123268
Forest DR, Knight DH, Franklin JF (1998) Landscape patterns
and legacies resulting from large, infrequent forest disturbances. Ecosystems (NY, Print) 1:497–510. doi:10.1007/
s100219900046
Freedman B, Hutchinson TC (1980) Long-term effects of
smelter pollution on forest community composition. Can J
Bot 55(19):2123–2140
Frelich LE, Reich PB (1999) Neighborhood effects, disturbance severity and community stability in forests.
Ecosystems (NY, Print) 2:151–166. doi:10.1007/s100219
900066
Georgievsky AB (1992) Faza okon v korennykh elovykh
lesakh yuzhnoy taygi. Bot J 77(6):52–62
Ginocchio R (2000) Effects of a copper smelter on a grassland
community in the Puchuncavi Valley, Chile. Chemosphere 41:15–23. doi:10.1016/S0045-6535(99)00385-9
Gol’dberg IL (1997) Changes in the moss layer of Southern
taiga dark coniferous forests under conditions of technogenic pollution. Russ J Ecol 28(6):417–419
Goldblum D (1997) The effects of treefall gaps on understory
vegetation in New York state. J Veg Sci 8(1):125–132.
doi:10.2307/3237250
Hautala H, Tolvanen A, Nuortila C (2001) Regeneration
strategies of dominant boreal forest dwarf shrubs in
response to selective removal of understorey layers. J Veg
Sci 12:503–510. doi:10.2307/3237002
Hautala H, Tolvanen A, Nuortila C (2008) Recovery of pristine
boreal forest floor community after selective removal of
understorey, ground and humus layers. Plant Ecol
194:273–282. doi:10.1007/s11258-007-9290-0
Hooper DU, Chapin FSIII, Ewel JJ, Hector A, Inchausti P,
Lavorel S, Lawton JH, Lodge DM, Loreau M, Naeem S,
Schmid B, Setälä H, Symstad AJ, Vandermeer J, Wardle
DF (2005) Effects of biodiversity on ecosystems functioning: consensus of current knowledge. Ecol Monogr
75(1):3–35. doi:10.1890/04-0922
Huopalainen M, Tuittila E-S, Vanha-Majamaa I, Nousiainen H,
Laine J, Vasander H (2001) Effects of long-term aerial
pollution on soil seed banks in drained pine mires in
Southern Finland. Water Air Soil Pollut 125:69–79. doi:
10.1023/A:1005276201740
Jonsson BG, Essen P-A (1998) Plant colonization in small
forest-floor patches: importance of plant group and disturbance traits. Ecography 21:518–526. doi:10.1111/j.
1600-0587.1998.tb00443.x
Forest Ecology
Kaigorodova SY, Vorobeichik EL (1996) Changes in certain
properties of grey forest soil polluted with emissions from
a copper-smelting plant. Russ J Ecol 27(3):177–183
Kolesnikov BP, Zubareva RS, Smolonogov EP (1973) Lesorastitel’nye usloviya i lesnaya tipologiya Sverdlovskoy.
UNC AN SSSR, Sverdlovsk
Komulainen M, Vieno M, Yarmishko VT, Daletskaja TD,
Maznaja EA (1994) Seedling establishment from seeds
and seed banks in forests under long-term pollution stress:
a potential for vegetation recovery. Can J Bot 72:143–
149. doi:10.1139/b94-019
Kuuluvainen T (1994) Gap disturbance, ground microtopography and the regeneration dynamics of boreal coniferous
forests in Finland: a review. Ann Zool Fenn 31(1):35–51
Lavorel S (1999) Ecological diversity and resilience of Mediterranean vegetation to disturbance. Divers Distrib 5:3–
13. doi:10.1046/j.1472-4642.1999.00033.x
Lee JA (1998) Unintentional experiments with terrestrial ecosystems: ecological effects of sulfur and nitrogen
pollutants. J Ecol 86:1–12. doi:10.1046/j.1365-2745.1998.
00257.x
Liechty HO, Jurgensen MF, Mroz GD, Gale MR (1997) Pit and
mound topography and its influence on storage of carbon,
nitrogen and organic matter within an old-growth forest. Can
J Res 27(12):1992–1997. doi:10.1139/cjfr-27-12-1992
Lukina NV, Nikonov VV (1993) Sostoyanie elovykh biogeotzenozov Severa v uscloviyakh tekhnogennogo
zagryazneniya. Kola Scientific Centre, Apatity
Makhnev AK, Trubina MR, Pryamonosova SA (1990) Lesnaya
rastitel’nost’ v okrestnostyakh predpriyatiy tsvetnoy metallurgii. In: Shiyatov SG, Shilova II (eds) Estestvennaya
rastitel’nost’ promyshlennykh i urbanizirovannykh territoriy Urala. UD AS USSR, Sverdlovsk
Mayer P, Abs C, Fischer A (2004) Colonization by vascular
plants after soil disturbance in the Bavarian Forest—key
factors and relevance for forest dynamics. For Ecol
Manage 188:279–289
McNaughton SJ (1977) Diversity and stability of ecological
communities: a comment on the role of empiricism
in ecology. Am Nat 111(979):515–525. doi:10.1086/
283181
Meerts P, Grommesch C (2001) Soil seed banks in heavy-metal
polluted grassland at Prayon (Belgium). Plant Ecol
155:35–45. doi:10.1023/A:1013234418314
Nakashizuka T (1989) Role of uprooting in composition and
dynamics of an old-growth forest in Japan. Ecology
70(5):1273–1278. doi:10.2307/1938186
Peterson CJ, Campbell JE (1993) Microsite differences and
temporal change in plant-communities of treefall pit and
mounds in an old-growth forest. Bull Torrey Bot Club
120(4):451–460. doi:10.2307/2996750
Peterson CJ, Pickett STA (1995) Forest reorganization—a
case-study in old-growth forest catastrophic blowdown.
Ecology 76(3):763–774. doi:10.2307/1939342
Peterson CJ, Carson WC, McCarthy BC, Pickett STA (1990)
Microsite variation and soil dynamics within newly created treefall pits and mounds. Oikos 58(1):39–46. doi:
10.2307/3565358
Peterson G, Allen CR, Holling CS (1998) Ecological resilience, biodiversity, and scale. Ecosystems (NY, Print)
1:6–18. doi:10.1007/s100219900002
349
Prokaev VI (1976) Fiziko-geograficheskoe rayonirovanie
Sverdlovskoy oblasti. Part 1. Sverdlovsk State Pedagogical Institute, Sverdlovsk
Putz FE (1983) Treefall pits and mounds, buried seeds, and the
importance of soil disturbance to pioneer trees on Barro
Colorado Island, Panama. Ecology 64(5):1069–1074. doi:
10.2307/1937815
Rydgren K, Økland RH, Hestmark G (2004) Disturbance
severity and community resilience in a boreal forest.
Ecology 85(7):1906–1915. doi:10.1890/03-0276
Salemaa M, Uotila T (2001) Seed bank composition and
seedling survival in forest soil polluted with heavy metals.
Basic Appl Ecol 2:251–263. doi:10.1078/1439-179100055
Salemaa M, Vanha-Majamaa I, Derome J (2001) Understory
vegetation along a heavy-metal pollution gradient in SW
Finland. Environ Pollut 112:339–350. doi:10.1016/S02697491(00)00150-0
Sannikov SN (1992) Ekologiya i geografiya estestvennogo
vozobnovleniya sosny obyknovennoy. Nauka, Moscow
Skvortsova EB, Ulanova NG, Basevich BF (1983) Ekologichskaya rol’ vetrovalov. Forestry Industry, Moscow
Smith YK (1985) Les i atmosfera. Progress, Moscow
Stepanov AM (ed) (1992) Kompleksnaya ekologicheskaya otsenka
tekhnogennogo vozdeystviya na ekosistemy yuzhnoy taygi.
TsEPL, Moskow
Sydes C, Grime TP (1981a) Effects of tree leaf litter on herbaceous
vegetation in deciduous woodland. 1. Field investigations.
J Ecol 69(1):237–248. doi:10.2307/2259828
Sydes C, Grime TP (1981b) Effects of tree leaf litter on herbaceous vegetation in deciduous woodland. 2. An
experimental investigation. J Ecol 69(1):249–262. doi:
10.2307/2259829
Tilman D, Downing JA (1994) Biodiversity and stability in
grasslands. Nature 367:363–365. doi:10.1038/367363a0
Trubina MR (2003) Kolichestvennye pokasateli iskusstvenno
narushennykh uchastkov temnokhvoynykh lesov na
nachal’nykh stadiyakh vosstanovitel’nykh mikrosuktzessiy
v usloviyakh zagryazneniya. In: Mamaev SA, Makhnev
AK, Chibrik TS (eds) Biologicheskaya rekul’tivatsiya
narushennykh zemel’. Ural Division of Russian Academy
of Science, Ekaterinburg
Trubina MR, Makhnev AK (1997) Dynamics of ground vegetation in forest phytocenoses under conditions of chronic
pollution by fluorine. Russ J Ecol 28(2):73–77
Turner MG, Baker WL, Peterson CJ, Peet RK (1998) Factors
influencing succession: lessons from large, infrequent
natural disturbances. Ecosystems (NY, Print) 1:511–523.
doi:10.1007/s100219900047
Ulanova NG (2000) The effects of windthrow on forests at
different spatial scales: a review. For Ecol Manage
135:155–167
Vorobeichik EL (1995) Change of thickness of forest litter
under chemical pollution. Russ J Ecol 26(4):252–258
Vorobeichik EL, Hantemirova EV (1994) Reaction of forest
phytocenoses to technogenic pollution: dose-effect
dependencies. Russ J Ecol 25:171–180
Vorobeichik EL, Sadykov OF, Farafontov MG (1994) Ecologicheskoe normirovanie tekhnogennykh zagryaznenii
nazemnykh ekosistem (lokal’nyi uroven’). Nauka,
Ekaterinburg
350
Weltzin JF, Keller JK, Bridgham SD, Pastor J, Allen PB, Chen
J (2005) Litter controls plant community composition in a
northern fen. Oikos 110:537–546. doi:10.1111/j.00301299.2005.13718.x
Whitford WG, Rapport DJ, Soyza AG (1999) Using resistance
and resilience measurements for ‘fitness’ tests in
A.G. Van der Valk (ed.)
ecosystem health. J Environ Manage 57:21–29. doi:
10.1006/jema.1999.0287
Xiong S, Nilsson C (1999) The effects of plant litter on vegetation: a meta-analysis. J Ecol 87:984–994. doi:10.1046/
j.1365-2745.1999.00414.x
Hurricane disturbance in a temperate deciduous forest:
patch dynamics, tree mortality, and coarse woody detritus
R. T. Busing Æ R. D. White Æ M. E. Harmon Æ
P. S. White
Originally published in the journal Plant Ecology, Volume 201, No. 1, 351–363.
DOI: 10.1007/s11258-008-9520-0 Springer Science+Business Media B.V. 2008
Abstract Patch dynamics, tree injury and mortality,
and coarse woody detritus were quantified to examine
the ecological impacts of Hurricane Fran on an oakhickory-pine forest near Chapel Hill, NC. Data from
long-term vegetation plots (1990–1997) and aerial
photographs (1998) indicated that this 1996 storm
caused patchy disturbance of intermediate severity
(10–50% tree mortality; Woods, J Ecol 92:464–476,
2004). The area in large disturbance patches
([0.1 ha) increased from \1% to approximately 4%
of the forested landscape. Of the forty-two 0.1-ha
plots that were studied, 23 were damaged by the
storm and lost 1–66% of their original live basal area.
Although the remaining 19 plots gained basal area
(1–15% increase), across all 42 stands basal area
decreased by 17% because of storm impacts. Overall
mortality of trees [10 cm dbh was 18%. The basal
area of standing dead trees after the storm was
0.9 m2/ha, which was not substantially different from
the original value of 0.7 m2/ha. In contrast, the
volume and mass of fallen dead trees after the storm
R. T. Busing (&) R. D. White P. S. White
Department of Biology, University of North Carolina,
Chapel Hill, NC 27599-3280, USA
e-mail: rtbusing@aol.com
M. E. Harmon
Department of Forest Science, Oregon State University,
Corvallis, OR 97331-5752, USA
(129 m3/ha; 55 Mg/ha) were 6.1 and 7.9 times
greater than the original levels (21 m3/ha; 7 Mg/ha),
respectively. Uprooting was the most frequent type of
damage, and it increased with tree size. However, two
other forms of injury, severe canopy breakage and
toppling by other trees, decreased with increasing tree
size. Two dominant oak species of intermediate
shade-tolerance suffered the largest losses in basal
area (30–41% lost). Before the storm they comprised
almost half of the total basal area in a forest of 13%
shade-tolerant, 69% intermediate, and 18% shadeintolerant trees. Recovery is expected to differ with
respect to vegetation (e.g., species composition and
diversity) and ecosystem properties (e.g., biomass,
detritus mass, and carbon balance). Vegetation may
not revert to its former composition; however,
reversion of biomass, detritus mass, and carbon
balance to pre-storm conditions is projected to occur
within a few decades. For example, the net change in
ecosystem carbon balance may initially be negative
from losses to decomposition, but it is expected to be
positive within a decade after the storm. Repeated
intermediate-disturbance events of this nature would
likely have cumulative effects, particularly on vegetation properties.
Keywords Canopy gap dynamics Coarse
woody debris Forest ecosystem Intermediate
disturbance Net ecosystem carbon balance
North Carolina Piedmont Snag dynamics
Wind disturbance
A.G. Van der Valk (ed.), Forest Ecology. DOI: 10.1007/978-90-481-2795-5_26
351
352
Introduction
Wind is a leading agent of disturbance in the
temperate deciduous forests of eastern North America
(White 1979; Lorimer 1980; Runkle 1985). Within the
eastern deciduous forest (sensu Barbour et al. 1980)
small patches (or canopy gaps) created by the death of
one or a few trees are the most frequent form of
disturbance (Runkle 1982, 1985; Lorimer 1989). Yet,
hurricanes and other violent windstorms can create
larger patch disturbances having strong ecological
impacts that differ quantitatively and qualitatively
from those of small-patch disturbances (Dunn et al.
1983; Canham and Loucks 1984; Foster 1988; Peart
et al. 1992; Boose et al. 1994; Peterson and Pickett
1995; Greenberg and McNab 1997; Frelich 2002;
Woods 2004). Hurricane disturbance in the eastern
deciduous forest is not fully characterized with respect
to immediate impacts and long-term impacts on
vegetation and ecosystems. What types of forest
damage occur over the landscape and within stands,
and the implications for dynamics of vegetation and
ecosystems require attention. Consideration of vegetation and ecosystem recovery processes and rates
following hurricane disturbance contributes toward
understanding of long-term dynamics of these forests.
In this article, we examine the impacts of Hurricane
Fran, which passed through the North Carolina Piedmont in September 1996, on community and ecosystem
attributes of an eastern deciduous forest. Windstorm
impact is usually assessed through basal area loss, but
one can include other measures such as estimates of
the size, abundance and dynamics of disturbance
patches (e.g., Platt et al. 2000) as well as the changes
in canopy cover (e.g., Peart et al. 1992), coarse woody
detritus (CWD) (e.g., Whigham et al. 1991), and
ecosystem carbon dynamics to better quantify disturbance severity and impacts on the forest community
and ecosystem (Everham and Brokaw 1996).
Relying primarily on a set of permanent vegetation
plots established prior to Hurricane Fran in oakhickory-pine forest, we consider changes in forest
structure, composition, and ecosystem properties following the storm. Our general hypothesis is that the
hurricane disturbance differs from small-gap disturbances in its impacts on vegetation and ecosystems.
We ask the following questions: (1) How different is
the size of disturbance patches created by the storm
compared to other disturbances in this system? (2) Do
A.G. Van der Valk (ed.)
the type and degree of damage differ by tree size, by
tree species, and by ecological guild (e.g., shadetolerance class)? (3) How are the forest composition
and succession affected? (4) To what degree are
detritus levels and forest ecosystem processes related
to carbon dynamics altered? In addressing these
objectives and questions, we assess both live and dead
trees, allowing evaluation of vegetation and ecosystem
impacts. Long-term impacts and the role of cumulative
disturbance effects due to successive hurricanes are
projected and discussed.
Study area
The North Carolina Botanical Garden is a 242-ha
tract of oak-hickory-pine forest (Braun 1950; Greller
1988) in Chapel Hill, North Carolina (35530 N,
7920 W). The development, dynamics, and environment of the Piedmont hardwood forest of North
Carolina are well studied (Oosting 1942; Peet and
Christensen 1980). It is an area defined by undulating
topography, soils of poor to good quality, and a
temperate climate. Soils of the study area include
Wedowee sandy loam and Goldston slaty silt loam
(Dunn 1977). The climatic regime of the area fits
Thornthwaite’s (1948) humid mesothermal class,
with a mean annual temperature of 16C and a mean
annual precipitation of 116 cm (NOAA 1974).
Human history of the study area is incompletely
known. Some productive sites (e.g., floodplains and
lower slopes) of the study area were farmed starting
from the mid- to late 1700s and ending between the
late 1800s and 1920. It is likely that some of the
upland forests studied here survived as woodlots in
the farm landscape, with occasional cutting of trees
for firewood or lumber and with understory grazing;
oaks were considered a valuable source of forage for
livestock. Fire was used by Native Americans prior to
1700 and by farmers thereafter. Some lower slope
forest patches are about 80 years old, whereas most
forests are probably 120 years old, and the older
woodlots have been continuously in the forest for
hundreds of years and support occasional trees that
are 200–250 years old.
Hurricane Fran, a category three storm, passed
through the region on the morning of 6 September
1996. The eye passed about 7 km east of the study
area. Wind data from the closest meteorological
station at Raleigh-Durham International Airport
Forest Ecology
40 km east of Chapel Hill indicated sustained winds
of 72 km/h and gusts of up to 128 km/h during the
storm (NOAA, unpublished data).
Methods
Sampling prior to the hurricane
Forty two 0.1-ha (20 9 50 m) plots were established
in upland forests of the North Carolina Botanical
Garden from March 1990 to May 1991. A 100 9
100 m grid was surveyed across garden lands prior to
the establishment of the plots. The grid of 1-ha cells
covered a landscape ca. 50 ha in area. Plots were
dispersed so that no more than one plot occurred in
each 1-ha grid cell. Design of the permanently
marked plots followed that of the North Carolina
Vegetation Survey (Peet et al. 1998), featuring nested
subplots for multiscale sampling of composition,
structure, and diversity.
During the initial sampling, we laid out each 0.1ha plot with ten contiguous10 9 10 m subplots.
Within each subplot, all trees larger than 1 cm dbh
(diameter at breast height) were identified according
to species and measured for diameter. In addition, all
live and dead trees over 10 cm dbh were mapped to
allow subsequent data collectors to track individual
stems. Each fallen dead stem of this size was mapped
and assigned to a 10-cm diameter class. Vegetation
type (pine, mixed, or hardwood), canopy height,
elevation, aspect, slope, and soil characteristics such
as nutrient content and density of soil were measured
and documented for each plot during pre-hurricane
sampling (White et al. 1991, 1992).
Sampling after the hurricane
In the summer of 1997, we re-sampled all of the 42
original upland plots. We re-measured every tree
greater than 1 cm dbh and assigned one of the four
damage-type codes to hurricane-damaged individuals: uproot (H1, if uprooted by wind), breakage (H2,
if canopy was damaged by the wind), leaner angle
(H3, if the tree was leaning), and leaner support (H4,
if the tree was supporting another tree). A damage
severity level (1–3 or 1–4) was assigned as well,
depending on the damage-type code (e.g., H1 = 3 for
353
a tree completely uprooted by wind) yielding a total
of 15 classes of damage type and severity (H1 = 1–3,
H2 = 1–4, H3 = 1–4, and H4 = 1–4). All trees
greater than 10 cm dbh in the previous survey were
incorporated into a dataset summarizing the fates of
individual large stems (White 1999).
To quantify differences in damage among plots,
the amount of basal area severely damaged by the
hurricane was determined within each plot (Everham
and Brokaw 1996). We considered severely damaged
stems to be those that had been completely tipped up
(H1 = 3), had lost [35% of canopy from breakage
(H2 = 3 or 4), or had fallen with their bole lying on
the ground (trees fallen, H3 = 4; or toppled and
pinned by other trees, H4 = 4). If any of these
categories applied to the individual tree in question, it
was effectively eliminated from the canopy of the
forest because it was no longer fully present as a live
tree in the canopy. We excluded any tree considered
severely damaged from our canopy live basal area
estimates after the hurricane.
Fallen CWD (diameter [10 cm) and canopy
disturbance were measured after the hurricane. In
1997, all fallen boles and branches were sampled
using the planar transect method (Brown 1974;
Harmon and Sexton 1996). Pieces intercepted by
the 50-m centerline of each plot were measured for
diameter at point of intercept. In addition, each piece
was classified as input before or after the hurricane.
The amount of decay was noted for each piece using
the two-stage classification of Brown (1974). Total
volume was calculated following VanWagner (1968):
V ¼ p2 R D2 8L
where V is the volume in m3/m2, L is the transect
length (50 m), and D is the diameter of individual
pieces (m). Mass was calculated using the approximate density of fresh wood (0.46 Mg/m3—averaged
across all species) and decomposed wood (0.29
Mg/m3—averaged across all species and decay
classes). Density was determined by sawing out
sections from recently fallen trees as well as those in
various states of decomposition. Volume of the
sections was calculated from surface measurements
and mass was determined by weighing the entire
section and then subsampling it to determine moisture content. Density was calculated as the dry mass
(oven dried at 55C) divided by the undried volume.
354
A.G. Van der Valk (ed.)
Changes in net ecosystem carbon balance (NECB,
Chapin et al. 2006) after the 1996 hurricane were
projected using net primary production (NPP) and
mass decay of CWD:
NECB = NPPb
DCWD ;
where NPPb is bole NPP of the relatively undisturbed
forest plots and DCWD are the losses due to the decay
of CWD. NPPb was estimated using measurements of
stem diameter growth and allometric equations for
stem biomass of trees grouped by genus or species
(Ter-Mikaelian and Korzukhin 1997). The losses
from decomposition were calculated as:
DCWD ¼ MCWDt
MCWDt 1 ;
where MCWDt is the mass of CWD at time t calculated
with a negative exponential model (Olson 1963):
MCWDt ¼ MCWD0 e
kt
;
where k is the decomposition rate constant, assumed
to range between 0.1 and 0.2/year (Onega and
Eickmeier 1991).
Canopy disturbance after the hurricane was
assessed at the plot level and at the landscape level.
The amount of canopy loss in each plot was
quantified with a canopy densitometer, taking measurements at 10-m intervals along the plot centerline.
For landscape-level estimates, stereoscopic aerial
photographs taken in April 1998 were examined for
large canopy gaps (C0.1 ha) across a 45-ha area
covering undeveloped garden lands. We only
included openings with an unobstructed view of the
ground surface. The length and width of all canopy
gaps approximately 0.1 ha or larger were measured.
Area of individual gaps was estimated using the
formula for the area of an ellipse (Runkle 1982):
A ¼ p L W=4;
where A is gap area (m2), L is gap length (m), and W
is gap width (m). Upper and lower estimates of the
size of each gap were obtained using a tolerance of
6 m for gap length and width. The tolerance level
represented the precision of gap length and width
measurements from the photographs. Using these
methods, upper, intermediate, and lower estimates of
total land area in gaps C0.1 ha were generated.
Several plot-level variables such as basal area,
dead tree density, CWD volume, and CWD mass,
were compared before and after the hurricane.
Statistical differences before and after were assessed
with paired t-tests (SAS Institute Inc 1985) using
plot-level values from before and after. Two-tailed
probabilities were used to assess the significance of
changes.
Results
Landscape disturbance
The initial, pre-hurricane survey of forested lands in
the study area indicated that large, naturally created
gaps (C0.1 ha) were either rare or absent. Circa 1990,
prior to the hurricane, large gaps occupied less than
1% of the undeveloped land area. Two years after the
hurricane, the estimated total land area in large gaps
(C0.1 ha) was 4%. The lower and upper bounds for
this estimate based on the measurement tolerances
were 1 and 7%, respectively (see Methods).
Physical structure of stands
Basal area
Over the entire study area, live basal area declined
significantly by 17% (Table 1). The coefficient of
variation, indicating the variability among plots,
increased from 19 to 33%. Of the 42 plots visited
after the hurricane, 23 lost 1–66% of their original
live basal area (Fig. 1). The mean amount of basal
area lost in these damaged stands was 25%. In the 19
plots that were relatively undamaged by the hurricane, basal area increased on average by 8% over the
sampling interval (ca. 1990–1997). Basal area gains
ranged from 1 to 15% in these plots over this interval.
Canopy cover
Although canopy cover was not measured prior to the
hurricane, comparison of cover between damaged
and undamaged stands provided an indication of the
amount of cover lost in the storm. Mean canopy cover
was 92% in the 19 undamaged plots and 81% in the
23 damaged plots (see Appendix Table A1), respectively. Canopy cover ranged from 89 to 95% in the
undamaged plots, and from 60 to 93% in the damaged
plots.
Forest Ecology
355
Table 1 Live trees, standing dead trees, and fallen trees
before and after the 1996 hurricane (ca. 1990 vs. 1997)
Mean Std.
Dev.
Min. Max. Coeff.
Var.
N
Basal area of live trees (m2/ha)
Before storm 27
5
19
43
19
42
After storm
7
6
43
33
42
22**
Basal area of standing dead trees (m2/ha)
Before storm 0.7
1
0
3.1 108
42
After storm
0.9
1
0
4.9 112
42
0.9
1
0
4.6 110
42
Added by
storm
Density of standing dead trees [10 cm DBH (stems/ha)
Before storm 22
18
0
70
83
42
After storm
30**
20
0
80
66
42
29
38
0
80
67
42
Added by
storm
Density of standing dead trees [30 cm DBH (stems/ha)
Before storm 2
4
0
10
226
42
After storm
4
9
0
40
225
42
4
8
0
30
213
42
Added by
storm
Density of standing dead trees [50 cm DBH (stems/ha)
Before storm 0.5
2
0
10
453
42
After storm
0.5
2
0
10
453
42
0.2
2
0
10
648
42
Added by
storm
Volume of fallen trees [10 cm diameter (m3/ha)
Before storm 24
After storm
Added by
storm
26
3
110
107
40
129** 126
6
532
97
40
105
0
522
120
40
126
exceeded 30 cm dbh (2 stems/ha). As a result of
hurricane damage, across the study area an average of
0.9 m2/ha of new dead basal area was added. An
average of 29 stems/ha of new standing dead trees
[10 cm dbh was added, and a significantly higher
mean value of 30 m2/ha was attained. The associated
coefficient of variation declined from 80 to 66%
indicating decreased variability among plots.
Fallen trees
Prior to the hurricane the volume and mass of
downed CWD in the study forest averaged 24 m3/ha
and 7 Mg/ha, respectively (Table 1). Although the
range of values was quite large, for example, volume
ranged from 3 to 110 m3/ha, the average value is
quite typical for a warm temperate deciduous forest
(Muller and Liu 1991). The hurricane increased
CWD volume approximately sixfold to an average of
129 m3/ha (Table 1). Mass increased to a greater
degree, approximately eightfold to 55 Mg/ha,
because of the higher density of fresh wood added
by the disturbance.
The volume of CWD after the storm was much
greater in the stands that lost live basal area (see
Appendix Table A1). Nonetheless, the hurricane did
not substantially alter the relative variability of
downed CWD among plots as the coefficients of
variation for pre- versus post-hurricane volume were
107 and 97, respectively (Table 1).
Tree damage
Mass of fallen trees [10 cm diameter (Mg/ha)
Before storm 7
After storm
Added by
storm
7
1
32
106
40
55**
58
2
244
105
40
48
58
0
240
120
40
For several variables the amount added by the storm was
measured directly; this amount did not necessarily equal the
difference between 1990 and 1997 values. Significant
differences between before and after values are noted
(** significance at the p \ 0.01 level)
Standing dead trees
Pre-hurricane basal area of standing dead trees (or
snags) across the study area was low (0.7 m2/ha)
(Table 1). Density of standing dead trees ([10 cm
dbh) averaged 24 stems/ha. Few of the dead trees
Of all stems greater than 10 cm dbh, 18% were
severely damaged by the hurricane event (Table 2).
Certain types of injury were dependent on tree size.
For example, the occurrence of full uprooting
increased with tree size (Fig. 2a). By contrast, the
occurrence of severe canopy breakage (Fig. 2b) and
of toppling by other trees (Fig. 2c) decreased with
tree size. When all of these forms of damage were
considered, the tendency was for small-sized trees to
suffer the least damage (Fig. 2d).
Two moderately shade-tolerant species, red oak
(Quercus rubra) and black oak (Quercus velutina),
suffered the largest average basal area losses per plot
(41% and 30%, respectively), as shown in Table 3.
Conversely, shade-intolerant pine species lost only
7% of their basal area on average over all plots
356
A.G. Van der Valk (ed.)
Fig. 1 Basal area change in
the set of study plots before
and after the 1996 hurricane
(ca. 1990 to 1997)
14
12
FREQUENCY
10
8
6
4
2
0
15%
5%
-5%
-15%
-25%
-35%
-45%
-55%
-65%
BASAL AREA CHANGE
Table 2 Comparison of all stems and larger-sized stems with
respect to frequency snapped (H2 = 3 or 4), frequency
uprooted (H1 = 3), and frequency severely damaged (H1 = 3,
H2 = 3 or 4, H3 = 4, or H4 = 4)
Damage parameter
All stems
[1 cm dbh
(n = 10,547)
All stems
[10 cm dbh
(n = 1,899)
Frequency snapped
297 stems
117 stems
Percent snapped
2.8%
6.2%
Frequency uprooted
325 stems
210 stems
Percent uprooted
3.1%
11.2%
Frequency severely damaged
1014 stems
332 stems
Percent severely damaged
9.6%
17.7%
Values are for the entire 4.2-ha area sampled in summer 1997
after the September 1996 hurricane
relative basal area. For shade-tolerant species the loss
of basal area in damaged plots was partially offset by
gains in undamaged plots. Overall, damage caused
little change in the relative basal area of various
shade-tolerance classes.
Projected changes in NECB
The NPP of boles in undamaged stands was 2.5 Mg/
ha/year. Assuming that level of NPP is maintained
despite hurricane damage, the losses of mass due to
decomposition were projected to exceed forest inputs
during 5–10 years (Fig. 3). After that point the forest
should have a positive NECB.
Discussion
(Table 3). Deciduous canopy species such as tulip
poplar (Liriodendron tulipifera), red maple (Acer
rubrum), beech (Fagus grandifolia), white oak
(Quercus alba), and ash (Fraxinus sp.) suffered light
to moderate damage (7–19%). Consequently, they
showed large gains in importance (relative basal area)
after the hurricane. Their shade-tolerance classifications range from intolerant to tolerant.
Despite the differences in damage among species,
damage was not restricted to a particular shadetolerance class (Table 4). However, in undamaged
stands all but the shade-intolerant class increased in
Disturbance patterns, patch dynamics,
and succession
Hurricane disturbance on the upland landscape was
patchy. About half of the study stands suffered basal
area loss. Even in those stands, canopy disturbance
was incomplete at the scale of 0.1 ha. Stands that
were damaged lost, on average, one-quarter of their
original live basal area and about one-tenth of their
canopy cover. The lowest canopy cover estimate after
the disturbance was 60% across a 0.1-ha area.
Nonetheless, inputs of CWD in damaged stands were
Forest Ecology
Full
357
Partial
(a) Uprooted
None
Toppled
90%
FREQUENCY IN CATEGORY
FREQUENCY IN CATEGORY
100%
80%
60%
40%
20%
80%
70%
60%
50%
40%
30%
20%
10%
0%
<10
10 to 20
20 to 30
30 to 40
40 to 50
50 to 60 60 to 70
0%
> 70
<10
10 to 20
DIAMETER CLASS (cm DBH)
35-90% canopy loss
10-35% canopy loss
20 to 30
30 to 40
40 to 50
50 to 60 60 to 70
> 70
DIAMETER CLASS (cm DBH)
(b) Broken
>90 canopy loss
(d) Either uprooted, broken or toppled
None
100%
Damage noted
None
100%
90%
90%
80%
FREQUENCY IN CATEGORY
FREQUENCY IN CATEGORY
(c) Toppled
None
100%
70%
60%
50%
40%
30%
20%
10%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0%
<10
10 to 20
20 to 30
30 to 40
40 to 50
50 to 60 60 to 70
> 70
DIAMETER CLASS (cm DBH)
<10
10 to 20
20 to 30
30 to 40
40 to 50
50 to 60 60 to 70
> 70
DIAMETER CLASS (cm DBH)
Fig. 2 Hurricane damage by type and tree size for a uprooted trees, b broken trees, c toppled trees, and d all damaged trees. Sample
sizes vary by class (left to right, n = 8648, 938, 371, 250, 203, 89, 37 & 11)
substantial. Forest-wide, including both damaged and
undamaged stands, live basal area and canopy cover
declined only moderately, but necromass levels
increased markedly.
Whereas long-term studies of tree mortality in the
eastern deciduous forest give a mean rate of nearly
1% of the population dying per year (Parker et al.
1985; Runkle 2000; Busing 2005), the storm produced much higher levels of mortality. For stems
[10 cm dbh, 18% were classified as severely damaged; most were uprooted (11%) or had broken boles
(6%) (Table 2). Given that the mean return interval of
hurricanes of category three or higher is at least
40 year in the Chapel Hill area (NOAA, unpublished
data) and assuming a mean annual mortality rate of
nearly 1%, hurricanes probably account for less than
half of the total long-term mortality of forest trees.
Overall, larger trees suffered the greatest damage
(cf. DeCoster 1996). The pattern of increasing injury
and mortality with tree size did not fully conform to
Everham and Brokaw’s (1996) two generalized
conceptual models of hurricane disturbance effects
on forests. Neither the unimodal response model,
wherein intermediate-sized trees suffer the most
damage, nor the bimodal response model, wherein
intermediate-sized trees suffer the least amount of
damage, was followed. However, the commonly
observed tendency of minimal damage to small stems
(Everham and Brokaw 1996) was exhibited in this
case. The observed pattern contrasted sharply with
that of tree mortality between storm events, where
mortality of small trees is relatively high (Peet and
Christensen 1987).
Damage also varied by species and by patch
composition prior to the storm. The broad-leaved
deciduous species tended to suffer higher losses of
basal area than the needle-leaved coniferous species.
The fact that these deciduous species were in leaf at
the time of the storm was important. Their relatively
broad leaves and crowns made them susceptible to
wind damage. By contrast, early successional patches
of needle-leaved coniferous species (e.g., Pinus) were
358
A.G. Van der Valk (ed.)
Table 3 Frequency of trees in plots, basal area before and after the 1996 hurricane, basal area change, and percent of total basal area
lost or gained between the two sampling periods (ca. 1990 vs. 1997)
Frequency
Basal area before
storm (m2)
Basal area after
storm (m2)
Change in basal
area (%)
Shade
tolerance
Canopy species
Acer barbatum
18
0.38
0.29
-24
Tolerant
Acer rubrum
42
8.50
7.39**
-13
Tolerant
Carya species
39
16.42
13.06**
-21
Intermediate
Fagus grandifolia
29
4.53
3.83
-16
Fraxinus species
30
1.29
1.20
-7
Liriodendron
34
8.79
7.12*
-19
Intolerant
Pinus species
23
18.43
17.11**
-7
Intolerant
Quercus alba
42
39.87
36.37*
-9
Intermediate
Quercus rubra
36
9.19
5.39**
-41
Intermediate
Quercus velutina
Sub-canopy species
22
3.46
2.43
-30
Intermediate
Carpinus caroliniana
19
0.09
0.09
0
Tolerant
Cercis canadensis
16
0.06
0.06
0
Tolerant
Cornus florida
42
2.16
1.63**
-25
Tolerant
Crataegus species
11
0.01
0.02
50
Unknown
Ilex decidua
13
0.04
0.05*
25
Tolerant
Juniperus
36
0.72
0.82
14
Intolerant
Liquidambar
23
1.31
1.19
-9
Intolerant
Morus rubra
17
0.05
0.06
20
Tolerant
Nyssa sylvatica
40
1.69
1.41*
-17
Tolerant
Ostrya virginiana
30
0.68
0.62
-9
Tolerant
Oxydendron
39
4.98
4.86
-2
Tolerant
Ulmus species
17
0.11
0.13
18
Intermediate
Tolerant
Intermediate
Basal area is the total over the 4.2 ha area sampled. Significant differences are noted (** significance at the p \ 0.05 level,
* significance at the p \ 0.10 level). Shade tolerance classifications (1 is the highest tolerance class) are according to Baker (1949) or
Burns and Honkala (1990) (Liriodendron = Liriodendron tulipifera, Liquidambar = Liquidambar styraciflua, Juniperus = Juniperus
virginiana, Oxydendron = Oxydendron arboreum)
less affected. The variation in susceptibility among
species has implications for community dynamics.
First, the initial impacts of the storm altered forest
composition directly by reducing the abundance of
certain dominant deciduous species. Second,
resources (e.g., light and nutrients) made available
by disturbance appear to have enhanced the growth of
some species.
Overall, the forest continued to be dominated by
intermediate and shade-tolerant species after the
storm despite the newly created disturbance patches.
Taken as a group, shade-tolerant species have
increased in basal area in undamaged stands, whereas
shade-intolerant species have not increased in these
same stands. Thus, the trends in undamaged stands
are consistent with patterns in mid-successional
forests, as shade-intolerant species are giving way
to shade-tolerant species. In damaged stands, the loss
in basal area included species from all shade-tolerance classes. Yet, some intolerant species were
unaffected by the storm, potentially stalling or setting
succession back to an earlier stage, at least in the
disturbance patches.
Ecosystem dynamics
The loss of tree basal area (and biomass) resulting
from the storm is a disruption to ecosystem development in this otherwise aggrading forest. Large
amounts of organic debris were transferred to the
Forest Ecology
359
Table 4 Basal area of
major species before and
after the 1996 hurricane by
tolerance grouping and
stand damage (ca. 1990 vs.
1997)
Shade tolerance class
Basal area before
storm (m2 ha-1)
Basal after
storm (m2 ha-1)
Change in basal
area (%)
Damaged stands (n = 23)
Tolerant
Intermediate
Intolerant
4.0
2.9**
-27
20.2
3.9
13.5**
2.6**
-33
-34
8
Undamaged stands (n = 19)
Damaged stands are defined
as those with lower live
basal area after the storm.
Mean basal area values and
percent change are
provided. Significant
differences are noted
(** significance at the
p \ .01 level)
(a)
Tolerant
Intermediate
Intolerant
Tolerant
FLUX (Mg/ha/yr)
5
Intolerant
10
15
20
-2
-4
-6
CWD decay, k=0.1/yr
NPP
NECB, k=0.1/yr
CWD decay, k=0.2/yr
NECB, k=0.2/yr
-8
-10
-12
TIME SINCE DISTURBANCE (yr)
(b) 60
CWD MASS (Mg/ha)
3
6.4
6.3
-0.1
3.5
Intermediate
2
0
3.2**
17.3**
All stands (n = 42)
4
0
3.0
16.8
50
k=0.1/year
k=0.2/year
40
30
20
10
3
-14
18.6
15.1**
-19
5.0
4.3**
-15
(Harmon et al. 1986). Much of the detritus is wood,
which decomposes slowly in temperate forests
(\15% mass lost per year).
NECB, the overall change in organic matter, was
likely negative immediately following the hurricane.
Negative NECB would have been caused by the large
input of newly decomposing wood with losses
exceeding forest gains by net primary production
(NPP) (Fig. 3). The duration of the period of negative
NECB through losses to the atmosphere depends on
the decomposition rate and the time required for NPP
to recover to pre-hurricane levels. It is possible that
NPP of boles was temporarily reduced by the
hurricane and this may delay the switch from
negative to positive NECB. However, alternative
calculations with delays in NPP recovery did not alter
our conclusions regarding the time required to go
from negative to positive NECB as long as NPP
reached pre-hurricane levels within a decade. In
contrast, delaying the recovery of NPP promoted a
negative NECB because the lower NPP failed to
offset decomposition losses.
0
0
5
10
15
20
TIME SINCE DISTURBANCE (yr)
Fig. 3 Projected net ecosystem carbon balance (NECB) and
major components after the 1996 hurricane showing a fluxes
and b detritus mass decay
forest floor during the storm, particularly in heavily
damaged stands with new CWD. The rate of decomposition of detritus can have important consequences
for ecosystem energetics and nutrient dynamics
Long-term consequences
The changes in forest patch structure, composition,
and coarse detritus brought about by this disturbance
event are expected to last for decades. Reversion
toward the pre-hurricane state is expected for at least
some parameters, however. For ecosystem parameters such as live biomass and necromass, a direct but
potentially slow, recovery toward pre-hurricane
360
A.G. Van der Valk (ed.)
levels is anticipated. With the recovery of leaf area
and the additional resources made available by
disturbance, NPP will be maintained or increased
during the recovery period (Beard et al. 2005). Based
on published rates of CWD decomposition in similar
ecosystems (Onega and Eickmeier 1991; Busing
2005), CWD added by the hurricane should be
largely gone within 20–30 years (Fig. 3).
In contrast to biomass, forest composition and
diversity may initially diverge further from predisturbance levels as a result of new colonization and
recruitment. The direction and duration of community-level dynamics are potentially complex given
that much of the pre-hurricane forest was in midsuccession. A simple projection, based on community
resilience through positive feedback mechanisms, is
that after initial divergence, composition and diversity will revert to their pre-hurricane states. For
example, seeds, seedling banks, and sapling banks
generated by existing adult trees would be expected
to maintain recruitment of existing canopy species. If
post-hurricane recruitment of shade-tolerant seedlings, presumably established before the storm, is
relatively successful then succession may be accelerated (Abrams and Scott 1989); however, elevated
recruitment of shade-tolerant tree seedlings was not
detected shortly after the storm (White 1999). An
increase in exotic plants was evident within the first
2 years after the storm (White 1999). If the hurricane
disturbance facilitates invasion (Crawley 1987),
novel composition and dynamics may result. For
these reasons, full recovery of pre-hurricane composition and dynamics is unlikely.
Although hurricane disturbance is rarely catastrophic in Piedmont forests, episodic events of this
nature may have important, lasting impacts on forests
(Foster et al. 1998). Yet, the long-term effects of
hurricane disturbances in the Piedmont are not well
studied. It is increasingly clear that partial damage to
stands, as observed in this study, is typical of the
regional disturbance regime. The long-term response
of ecological parameters to intermediate-disturbance
events similar to the one described here is less clear.
Responses are likely to vary among community and
ecosystem parameters. It would be particularly useful
to know which parameters exhibit delayed recovery
or no recovery at all. If recovery times approach or
exceed the return interval for disturbances of this
severity, then the possibility of cumulative effects of
multiple intermediate-disturbance events on forest
dynamics must be considered.
Acknowledgements We are grateful to Julia Larke, Jon
Harrod, Jay Sexton, and Becky Fasth for assistance with
collection and processing of the data. This research was funded
by the Institute for Museum Studies, the University of North
Carolina, the Ward and Kaye Richardson Endowment, NSF
support to the Andrews LTER (DEB-9632921 and DEB0218088), and a Bullard Fellowship from Harvard University.
Appendix
Table A1 Forest stand parameters by 0.1 ha study plot in the North Carolina Botanical Garden before and after the 1996 hurricane
Plot
number
Basal area before
storm (m2/ha)
Basal area after
storm (m2/ha)
Basal area
change (%)
Densitometer reading
after storm (Cover, %)
CWD before
storm (m3/ha)
CWD after
storm (m3/ha)
50
22.1
7.6
-65.6
ND
74.6
170.7
25
28.0
10.9
-61.1
60.3
20.5
363.6
72
28.2
12.9
-54.2
63.8
8.49
412.2
8
22.6
13.2
-41.5
82.3
40.8
105.3
51
33.2
20.2
-39.1
86.0
38.9
413.6
24
35.5
21.9
-38.2
87.0
75.4
207.6
36
32.5
22.2
-31.8
63.8
20.9
287.9
33
25.7
18.0
-30.1
81.0
6.3
101.6
47
26.7
20.1
-24.8
83.2
17.8
183.1
28
33.3
25.1
-24.5
80.5
15.3
209.5
65
30.0
23.1
-22.9
89.0
ND
ND
42
25.2
20.3
-19.6
71.0
11.9
211.2
67
34.9
28.3
-19.0
90.3
36.5
36.9
Forest Ecology
361
Table A1 continued
Plot
number
Basal area before
storm (m2/ha)
Basal area after
storm (m2/ha)
Basal area
change (%)
Densitometer reading
after storm (Cover, %)
CWD before
storm (m3/ha)
CWD after
storm (m3/ha)
54
35.3
28.7
-18.6
81.7
4.8
160.4
40
28.1
22.9
-18.4
70.0
25.1
190.8
71
38.2
31.4
-17.8
88.3
6.5
21.8
34
27.0
23.1
-14.3
90.2
ND
ND
41
28.9
24.9
-13.7
92.8
11.9
84.3
4
32.1
28.9
-9.9
ND
18.1
233.3
32
29.2
27.2
-6.9
83.7
8.9
90.5
44
33.1
31.1
-6.0
86.2
12.0
90.4
70
22.5
21.4
-5.0
80.8
10.3
532.0
35
30.2
29.9
-1.1
ND
108.6
132.1
37
27.2
27.5
1.0
89.8
4.2
9.4
39
25.7
26.1
1.6
90.0
10.5
47.7
55
44.6
46.2
3.6
91.7
7.2
18.3
14
48
28.8
30.6
29.9
31.8
3.7
4.0
90.2
90.5
16.7
11.3
63.2
100.5
16
26.8
28.0
4.6
89.8
17.5
61.2
3
28.4
29.9
5.4
93.0
5.9
20.8
22
30.3
32.1
5.8
93.2
2.9
59.4
43
20.8
22.0
6.0
89.2
20.6
75.4
31
20.4
21.8
7.1
92.3
39.0
39.3
30
26.1
28.0
7.2
ND
6.3
8.3
46
21.6
23.7
9.7
88.7
22.7
26.3
2
27.1
29.9
10.2
93.7
26.7
99.5
49
28.9
31.9
10.3
94.8
4.4
5.7
52
37.7
41.7
10.6
90.7
20.5
83.0
1
29.9
33.2
11.0
90.5
11.4
13.1
68
21.2
23.9
12.7
92.3
33.8
51.4
9
26.0
29.7
14.4
95.3
17.8
17.8
17
28.8
33.1
14.9
92.2
109.8
124.1
Plots are ranked by change in basal area. Basal area change represents basal area losses from severe damage and gains due to growth
between sampling periods (ca. 1990 and 1997). Densitometer readings consist of the average of six measurements along the
centerline to determine cover of the canopy in each plot (ND = no data). Coarse woody detritus (CWD) is the amount of fallen wood
bisecting the 50 m centerline plane (ND = no data)
References
Abrams MD, Scott ML (1989) Disturbance-mediated accelerated succession in two Michigan forest types. For Sci
35:42–49
Baker FS (1949) A revised tolerance table. J For 47:179–181
Barbour MG, Burk JH, Pitts WD (1980) Terrestrial plant
ecology. Benjamin Cummings, New York
Beard KH, Vogt KA, Vogt DJ, Scatena FN, Covich AP, Sigurdardottir R, Siccama TG, Crowl TA (2005) Structural
and functional responses of a subtropical forest to
10 years of hurricanes and droughts. Ecol Monogr
75:345–361. doi:10.1890/04-1114
Boose ER, Foster DR, Fluet M (1994) Hurricane impacts to
tropical and temperate forest landscapes. Ecol Monogr
64:369–400. doi:10.2307/2937142
Braun EL (1950) Deciduous forests of eastern North America.
Blakiston, Philadelphia
Brown JK (1974) Handbook for inventorying downed woody
material. USDA Forest Service GTR INT-16, Ogden,
Utah
Burns RM, Honkala BH (1990) Silvics of North America.
USDA Forest Service, Washington
Busing RT (2005) Tree mortality, canopy turnover and woody
detritus in old cove forests of the southern Appalachians.
Ecology 86:73–84. doi:10.1890/04-0410
362
Canham CD, Loucks OL (1984) Catastrophic windthrow in the
presettlement forests of Wisconsin. Ecology 65:803–809.
doi:10.2307/1938053
Chapin FS et al (2006) Reconciling carbon-cycle concepts,
terminology, and methods. Ecosystems 9:1041–1050. doi:
10.1007/s10021-005-0105-7
Crawley MJ (1987) What makes a community invasible? In:
Gray AJ, Crawley MJ, Edwards PJ (eds) Colonization,
succession and stability. Blackwell Scientific, Oxford, pp
429–453
DeCoster J (1996) Impacts of tornados and hurricanes on the
community structure and dynamics of North and South
Carolina forests. Ph.D. Dissertation, University of North
Carolina, Chapel Hill, North Carolina, USA
Dunn J (1977) Soil survey of Orange County, North Carolina.
Department of Agriculture, Soil Conservation Service, in
cooperation with the North Carolina Agricultural Experiment Station and the Orange County Board of
Commissioners
Dunn CP, Gutenspergen GR, Dorney JR (1983) Catastrophic
wind disturbance in an old-growth hemlock-hardwood
forest, Wisconsin. Can J Bot 61:211–217. doi:10.1139/
b83-023
Everham EM, Brokaw NVL (1996) Forest damage and
recovery from catastrophic wind. Bot Rev 62:113–185.
doi:10.1007/BF02857920
Foster DR (1988) Species and stand response to catastrophic
wind in central New England, U.S.A. J Ecol 76:135–151
Foster DR, Knight DH, Franklin JF (1998) Landscape patterns
and legacies resulting from large, infrequent forest disturbances. Ecosystems 1:497–510. doi:10.1007/s1002199
00046
Frelich LE (2002) Forest dynamics and disturbance regimes.
Cambridge University Press, Cambridge
Greenberg CH, McNab WH (1997) Forest disturbance in
hurricane-related downbursts in the Appalachian mountains of North Carolina. For Ecol Manage 104:179–191.
doi:10.1016/S0378-1127(97)00246-6
Greller AM (1988) Deciduous forest. In: Barbour MG, Billings
WD (eds) North American terrestrial vegetation. Cambridge University Press, New York, pp 287–316
Harmon ME, Sexton J (1996) Guidelines for measurements of
woody detritus in forest ecosystems. Publication no. 20.
U.S. LTER Network Office. University of Washington,
Seattle, Washington
Harmon ME, Franklin JF, Swanson FJ, Sollins P, Gregory SV,
Lattin JD, Anderson NH, Cline SP, Aumen NG, Sedell JR,
Lienkaemper GW, Cromack K, Cummins KW (1986)
Ecology of coarse woody debris in temperate ecosystems.
Adv Ecol Res 15:133–302. doi:10.1016/S0065-2504(08)
60121-X
Lorimer CG (1980) Age structure and disturbance history of a
southern Appalachian virgin forest. Ecology 61:1169–
1184. doi:10.2307/1936836
Lorimer CG (1989) Relative effects of small and large disturbances on temperate hardwood forest structure.
Ecology 70:565–567. doi:10.2307/1940207
Muller RN, Liu Y (1991) Coarse woody debris in an oldgrowth deciduous forest on the Cumberland Plateau,
southeastern Kentucky. Can J For Res 21:1567–1572. doi:
10.1139/x91-218
A.G. Van der Valk (ed.)
NOAA (1974) Climates of the states: eastern states. Water
Information Center, Port Washington
Olson JS (1963) Energy storage and the balance of producers
and decomposers in ecological systems. Ecology 44:322–
331. doi:10.2307/1932179
Onega TL, Eickmeier WG (1991) Woody detritus inputs and
decomposition kinetics in a southern temperate deciduous
forest. Bull Torrey Bot Club 118:52–57. doi:10.2307/
2996975
Oosting HJ (1942) An ecological analysis of the plant communities of the Piedmont, North Carolina. Am Midl Nat
28:1–126. doi:10.2307/2420696
Parker GR, Leopold DJ, Eichenberger JK (1985) Tree
dynamics in an old-growth deciduous forest. For Ecol
Manage 11:31–57. doi:10.1016/0378-1127(85)90057-X
Peart DR, Cogbill CV, Palmiotto PA (1992) Effects of logging
history and hurricane damage on canopy structure in a
northern hardwoods forest. Bull Torrey Bot Club 119:29–
38. doi:10.2307/2996917
Peet RK, Christensen NL (1980) Hardwood vegetation of the
North Carolina Piedmont. In: Leith H, Landolt E (eds)
Contributions to the knowledge of flora and vegetation in
the Carolinas. Veroff. Geobot. ETH, Zurich, pp 14–38
Peet RK, Christensen NL (1987) Competition and tree death.
Bioscience 37:586–595. doi:10.2307/1310669
Peet RK, Wentworth TR, White PS (1998) A flexible, multipurpose method for recording vegetation composition and
structure. Castanea 63:262–274
Peterson CJ, Pickett STA (1995) Forest reorganization: a case
study in an old-growth forest catastrophic blowdown.
Ecology 76:763–774. doi:10.2307/1939342
Platt WJ, Doren RF, Armentano TV (2000) Effects of Hurricane Andrew on stands of slash pine (Pinus elliottii var.
densa) in the everglades region of south Florida (USA).
Plant Ecol 146:43–60. doi:10.1023/A:1009829319862
Runkle JR (1982) Patterns of disturbance in some old-growth
mesic forests in eastern North America. Ecology
63:1533–1546. doi:10.2307/1938878
Runkle JR (1985) Disturbance regimes in temperate forests. In:
Pickett STA, White PS (eds) The ecology of natural disturbance and patch dynamics. Academic Press, New York,
pp 17–33
Runkle JR (2000) Canopy tree turnover in old-growth mesic
forests of eastern North America. Ecology 81:554–567
SAS Institute Inc. (1985) SAS/STAT guide for personal
computers. SAS Institute Inc., Cary
Ter-Mikaelian MT, Korzukhin MD (1997) Biomass equations
for sixty-five North American tree species. For Ecol
Manage 97:1–24. doi:10.1016/S0378-1127(97)00019-4
Thornthwaite CW (1948) An approach towards a rational classification of climate. Geogr Rev 38:55–94. doi:10.2307/
210739
VanWagner CE (1968) The line intersect method in forest fuel
sampling. For Sci 14:20–26
Whigham DF, Olmstead I, Harmon ME (1991) The impact of
hurricane Gilbert on trees, litterfall, and woody debris in a
dry tropical forest in the northeastern Yucatan Peninsula.
Biotropica 23:434–441. doi:10.2307/2388263
White PS (1979) Pattern, process, and natural disturbance in
vegetation. Bot Rev 45:229–299. doi:10.1007/BF0286
0857
Forest Ecology
White RD (1999) The impacts of Hurricane Fran on a North
Carolina Piedmont woodland. M.A. Thesis, University of
North Carolina, Chapel Hill
White PS, Busing RT, Larke JO (1991) Conservation project:
the William Lanier Hunt Arboretum and Gray Bluff
Garden. Technical Report, University of North Carolina,
Chapel Hill
363
White PS, Busing RT, Larke JO (1992) Conservation project:
Mason farm biological reserve. Technical Report, University of North Carolina, Chapel Hill
Woods KD (2004) Intermediate disturbance in a late-successional hemlock-northern hardwood forest. J Ecol 92:464–
476. doi:10.1111/j.0022-0477.2004.00881.x