Plant Pathology (2010) 59, 712–720 Doi: 10.1111/j.1365-3059.2010.02306.x Erysiphe trifolii – a newly recognized powdery mildew pathogen of pea R. N. Attanayakea, D. A. Glawea,b, K. E. McPheec, F. M. Dugand and W. Chend* a Department of Plant Pathology, Washington State University, Pullman WA 99164; bCollege of Forest Resources, University of Washington, Seattle WA 98195; cDepartment of Plant Sciences, North Dakota State University, Fargo ND 58108; and dUSDA-ARS, Washington State University, Pullman WA 99164, USA Diversity of powdery mildew pathogens infecting pea (Pisum sativum) in the US Pacific Northwest was investigated using both molecular and morphological techniques. Phylogenetic analyses based on rDNA ITS sequences, in combination with assessment of morphological characters, defined two groups of powdery mildews infecting pea. Group I (five field samples and three glasshouse samples) had ITS sequences 99% similar to those of Erysiphe pisi in GenBank and exhibited simple, mycelioid type of chasmothecial appendages typical of E. pisi. Erysiphe pisi is normally considered as the powdery mildew pathogen of pea. Group II (four glasshouse samples and two field samples) had ITS sequences 99% similar to those of E. trifolii and produced chasmothecia with dichotomously branched appendages similar to those of E. trifolii. There are fourteen nucleotide differences in the ITS region between the two groups. The correlation of rDNA ITS sequences with teleomorphic features for each of the two groups confirms their identity. Repeated samplings and artificial inoculations indicate that both E. pisi and E. trifolii infect pea in the US Pacific Northwest. Erysiphe trifolii is not previously known as a pathogen of pea. The existence of two distinct powdery mildew species infecting pea in both glasshouse and field environments may interfere with the powdery mildew-resistance breeding programmes, and possibly explains putative instances of breakdown of resistance in previously resistant pea breeding lines. Keywords: chasmothecial appendages, ITS sequences, Pisum sativum, powdery mildew, single nucleotide polymorphism, taxonomy Introduction Powdery mildew of pea (Pisum sativum), caused by Erysiphe pisi (in the past often reported as E. communis auct. p.p. or E. polygoni auct. p.p.), is a serious disease worldwide (Falloon & Viljanen-Rollinson, 2001). Two different Erysiphe pisi varieties have been described in Braun (1987), i.e. E. pisi var. pisi and E. pisi var. cruchetiana. In addition to species in Pisum, species in Medicago, Vicia, Lupinus, Lens and a small number of other genera are infected by E. pisi var. pisi, whereas Lathyrus and Ononis species are hosts for E. pisi var. cruchetiana (Braun, 1987). Powdery mildew adversely affects total biomass, number of pods per plant, number of seeds per pod, plant height and number of nodes (Gritton & Ebert, 1975). Yield loss of 10–65% due to the disease has been documented (Tiwari et al., 1997). Planting resistant cultivars is an economic and environmentally friendly means to manage the disease. Resistance to *E-mail: w-chen@wsu.edu Published online 10 May 2010 712 powdery mildew in pea is controlled by two recessive, independently inherited genes er-1 and er-2 (Heringa et al., 1969; Tiwari et al., 1997). Gene er-1 confers full resistance, whereas gene er-2 provides only leaf resistance (Heringa et al., 1969). A third gene, Er3, has been recently found in Pisum fulvum and also contributes to genetic resistance to E. pisi (Fondevilla et al., 2007). Although pea is always a field crop (glasshouse production is not commercially cost-effective), early generations of breeding materials are often produced in the glasshouse, particularly in temperate regions during winter months, in order to obtain two generations per year. Even though powdery mildew symptoms are easily recognized, it can be challenging to determine species assignment for a given powdery mildew (Glawe, 2008). Determining the pathogen to species level is very important for pea breeding programmes because different resistance genes may confer resistance to different species (Epinat et al., 1993). Taxonomic status of legume powdery mildews is incomplete in scope and not well understood. For example, E. trifolii has been referred to as a complex consisting of E. trifolii, E. baeumleri and E. asteragali (Braun, 1987). The nature of this No claim to original US government works Journal compilation ª 2010 BSPP Erysiphe trifolii on Pisum sativum species complex remains to be determined (U. Braun, Martin-Luther-Universität, Institut für Biologie, Germany, personal communication). Powdery mildew occurs on pea in both glasshouse and field conditions, and inconsistencies in resistance have been noted between the two. For example, some pea breeding lines were resistant to powdery mildew in the glasshouse, but were susceptible in the field (K.E. McPhee, unpublished data). Such inconsistent results could be due to different disease pressures in different environments, or glasshouse environments may alter the susceptibility of host plants to powdery mildew strains (Braun, 1987; Cunnington et al., 2005). Another possible explanation is the presence of powdery mildew strains with different host ranges as reported in Cook & Fox (1992) or even different pathogen species in different environments. The objective of this study was to determine if more than one species of powdery mildew infects pea in the US 713 Pacific Northwest, and if so, whether infections from a given species are present in both glasshouse and the field. Materials and methods Powdery mildew samples A total of 18 powdery mildew samples (eight from glasshouses and 10 from fields or natural areas) were collected from pea and other legume plants (Table 1). The eight glasshouse samples (seven from pea and one from lentil) were from four disconnected glasshouses located at least 500 m apart from each other. Three of the four glasshouses (all with entrances facing north) are located on the Pullman campus of Washington State University and form a triangle. The fourth glasshouse is located 16 km away in Moscow, Idaho. There was minimum chance of transferring inoculum from one glasshouse to another because the glasshouses were usually not planted with Table 1 Sample designation, host plant, location and sampling date, appendage morphology, ITS sequence group (number of nucleotides considered), GenBank accession and species determination of powdery mildew (Erysiphe spp.) samples used in this study Sample designation Host plant Location Samples from glasshouses GH 04 Pisum sativum Collection date Greenhouse 112, 2004 Pullman, WA GH 05 P. sativum Greenhouse 112, 2005 Pullman, WA GH 06 P. sativum Greenhouse 112, 2006 Pullman, WA GH 07-119 P. sativum Greenhouse 119, 2007 Pullman, WA GH M 07 P. sativum Greenhouse, 2007 Moscow, ID GH N 07 P. sativum Plant Growth 2007 Facility Rm 134, Pullman, WA Lif 07 P. sativum Plant Growth 2007 Facility Rm 134, Pullman, WA LGHN 07 Lens culinaris Greenhouse 119, 2007 Pullman, WA Samples from agricultural fields or uncultivated areas EI 08-1 P. sativum Pullman, WA 2008 EI 08-2 P. sativum Pullman, WA 2008 FF 06 P. sativum Fairfield, WA 2006 GE 07 P. sativum Genesee, ID 2007 LI 08-1 P. sativum Pullman, WA 2008 LI 08-2 P. sativum Pullman, WA 2008 SP 07 P. sativum Pullman, WA 2007 Lathyrus Lathyrus sp. Pullman, WA 2007 Medicago Medicago lupulina Pullman, WA 2007 Melilotus Melilotus albus Colfax, WA 2006 a Not available. Not determined. b Plant Pathology (2010) 59, 712–720 Ratio of appendage length to ITS sequence chasmothecial Appendage group (number GenBank Species diameter type of nucleotides) accession determination December 1Æ5–2Æ5 Mycelioid I (646) FJ378870 E. pisi November N ⁄ Aa N⁄A II (646) FJ378874 E. trifolii November N ⁄ A N⁄A II (644) FJ378873 E. trifolii N⁄A N⁄A I (646) FJ378872 E. pisi December N ⁄ A N⁄A I (646) FJ378871 E. pisi December 4–6Æ5 Branched II (646) FJ378875 E. trifolii December 4Æ5–7 Branched II (280) N⁄A E. trifolii August 4–7 Branched II (568) FJ378882 E. trifolii July July August July July July July September September October N⁄A N⁄A N ⁄ Db N⁄D N⁄A N⁄A N⁄D 2–3 N⁄A N⁄A N⁄A N⁄A Mycelioid Mycelioid N⁄A N⁄A Mycelioid Mycelioid N⁄A N⁄A II (549) II (533) I (646) I (645) I (505) I (549) I (602) I (646) II (580) II (646) GU361633 GU361634 FJ378867 FJ378869 GU361635 GU361636 FJ378868 FJ378879 FJ378877 FJ378878 E. E. E. E. E. E. E. E. E. E. April trifolii trifolii pisi pisi pisi pisi pisi pisi trifolii trifolii 714 R. N. Attanayake et al. peas at the same season except one case in 2007 (see location and collection data in Table 1). At each sampling, specimens were taken from four to five well-separated plants in a given glasshouse and processed separately. If the specimens were identical in ITS sequences (usually the case), they were considered as one biological sample because they were likely asexually propagated from one source. Seven of the 18 samples were from pea plants from either commercial fields or experimental field plots. The remaining three samples were from other legume species found in uncultivated areas (roadsides and parks): black medick (Medicago lupulina), sweet pea (Lathyrus sp.) and sweet clover (Melilotus albus). Some of the samples were taken from the same glasshouse over the years (samples GH 04, GH 05 and GH 06, Table 1). However, there was little chance of carry-over of powdery mildew inoculum from the previous year’s glasshouse crop. Each year in August (the hottest month of the year), all glasshouses in the cool season grain legume breeding programme are routinely completely vacated, cleaned, and disinfected with GreenShield (Whitmire Micro-Gen Research Laboratories). Following disinfestations, they are allowed to naturally heat by solar radiation (with the cooling systems shut off) for 2 weeks to sanitize the glasshouse for insects and pathogens. It is assumed that each year the powdery mildew inoculum was from unidentified sources outside of the glasshouses. sequences of Erysiphe species found on fabaceous hosts from the NCBI GenBank database were included in the analysis. Sequences were aligned using BioEdit (Hall, 1999) and ambiguously aligned sites were removed. Phylogenetic analyses were conducted using PAUP* 4.0b8 (Swofford, 2002). Trees were obtained from maximum likelihood (ML) and parsimony (MP) methods. MP analysis was performed using the heuristic search option with 1000 random addition sequences to increase the likelihood of finding the most parsimonious tree. The branch-swapping algorithm used was tree-bisectionreconnection (TBR) with ‘MulTrees’ option in effect. Branches collapsed (creating polytomies) if branch length was zero. Gaps were treated as missing data. Strength of internal branches of the resulting trees was tested with Bootstrap analysis using 1000 replications (Felsenstein, 1985). In the ML method, the most appropriate evolution model was determined for a given data set using PAUP* 4.0b8 (Swofford, 2002) and DT-ModSel (Minin et al., 2003). A starting tree was obtained with the neighbourjoining (NJ) (Saitou & Nei, 1987) method with the JC69 model of evolution. With this tree, likelihood scores were calculated for 56 alternative models of evolution by PAUP. Once the model of evolution was chosen, it was used to construct a phylogenetic tree with the Minimum Evolution (ME) method from the heuristic search option in PAUP* 4.0b8. Starting branch lengths were obtained using the Rogers-Swofford approximation method. ITS sequencing Morphological characterization Total genomic DNA was isolated from about 100 mg of powdery mildew (conidia and mycelia) using the FastDNA Kit (BIO 101 Inc.), and ITS sequences were obtained from all samples as described in Attanayake et al. (2009). Polymerase Chain Reaction (PCR) was performed with total genomic DNA using the ITS1 and ITS4 primer pair (White et al., 1990) or Erysiphe-specific ITS primer pair, EryF (5¢-TACAGAGTGCGAGGCTCA GTCG-3¢) and EryR (5¢-GGTCAACCTGTGATCCA TGTGACTGG-3¢) (Attanayake et al., 2009). Amplified DNA fragments were first cloned and transformed into competent Escherichia coli cells (Invitrogen Crop). Following blue-white colony selection, plasmids were isolated from positive colonies and at least five clones from each sample were sequenced using one of the six primers: EryF, EryR, ITS1, ITS4, M13F and M13R (Attanayake et al., 2009). Nucleotide sequences were determined from both strands using an ABI PRISM 377 automatic sequencer (Applied Biosystems) at the Sequencing Core Facility of Washington State University. Sequences were used as queries in BLAST (http://www.ncbi.nlm.nih.gov/BLAST) searches to identify most similar sequences available in the GenBank databases. Chasmothecia (when available) and conidia were removed from leaves with an insect needle, mounted in water and examined at 100–1000· using bright field microscopy (Carl Zeiss Model Axioskop 40). Taxonomic characters such as chasmothecial appendages and diameters, number of asci per chasmothecium, number of ascospores per ascus, lengths and widths of asci, ascospores, conidia and conidiophore foot cells were examined and recorded. Five plants from each glasshouse (one from each of the four corners and one from the middle) were observed. At least 50 measurements were made for each character from each sample and results were compared with the species descriptions in Braun (1987). Phylogenetic analyses Thirteen of the 18 ITS sequences determined in this study that were at least 568 bp long (Table 1) along with 15 ITS Pathogenicity assays A detached leaf assay was carried out to confirm pathogenicity of E. trifolii on pea. Fresh inoculum (conidia) was obtained from Lens culinaris, M. albus and P. sativum, and used in cross inoculation of the same three host species as described below. The ITS sequences of the three inoculum sources were determined before the inoculation experiment. Powdery mildew-infected M. albus plants were obtained from the Boyer Park, Washington, and subsequently transplanted to a separate glasshouse to maintain fresh powdery mildew inoculum. Powdery mildew conidia from P. sativum and Lens culinaris were Plant Pathology (2010) 59, 712–720 Erysiphe trifolii on Pisum sativum obtained from naturally infected glasshouse-grown plants. Fresh leaves of Lens culinaris cv. Crimson, M. albus (PI 90186) and P. sativum cv. Dark Skin Perfection, grown in a separate glasshouse where no powdery mildew was observed, were surface disinfected with 70% ethanol for 30 s followed by three serial washings with sterilized distilled water (Spurr, 1979). Leaves of a given plant were then placed in three replicate Petri dishes (moist chambers) as described in Attanayake et al. (2009). Moist chambers were made with 9 cm diameter Petri dishes and sterilized wet filter papers. Sterilized 200 lm metal mesh (4 · 4 cm) was kept between filter paper and the leaf to prevent the leaf from directly contacting water. Each leaf petiole was inserted into a 200 lL pipette tip (narrow end flame sealed) filled with 1% sucrose solution, thereby prolonging greenness of the leaves. Conidia from P. sativum were used to inoculate detached leaves of M. albus and Lens culinaris, and conidia from M. albus and Lens culinaris were used separately to inoculate detached leaves of P. sativum. Conidia were applied on the abaxial leaf surfaces using a fine paint brush (Lim, 1973) until a white powdery appearance was visible. The paint brush was disinfested by rinsing in 95% ethanol, followed by air-drying, between each treatment. Aseptic techniques were used during the inoculation procedure to minimize cross contamination, and mock-inoculated leaves (a paint brush without conidia was applied to the leaf surface) served as controls to monitor potential contamination. All inoculation work was conducted in a biological safety cabinet. Inoculated leaves were incubated at room temperature under white fluorescence light with a 12 h photoperiod (Warkentin et al., 1995). Symptom development was monitored using a dissecting microscope (Carl Zeiss Model Stemi 2000 C) and recorded as presence or absence of powdery mildew colonies at 2-day intervals until leaves become senescent (usually 20 days). DNA from freshly developed powdery mildew colonies was isolated using the microwave method as described in Attanayake et al. (2009), and used in PCR and DNA sequencing as described above. The experiment was performed twice. An observational study was carried out in a glasshouse to see if the pea powdery mildew pathogen can infect soybean. Two soybean genotypes L84-2237 (PI 547870) and Harosoy (PI548573) were planted along with four pea cultivars: Dark Skin Perfection, Lifter, Medora and Radly. The two soybean genotypes are known to be susceptible to E. diffusa (Dunleavy, 1978; Lohnes & Nickell, 1994). Twelve seeds of each genotype (four seeds per 15-cm pot and three pots per genotype) were arranged randomly, and maintained at 16 h photoperiod, 19–24C day temperature and 15C night temperature. Upon first appearance of symptoms, development of powdery mildew, revealed by the presence of white powdery colonies, was observed and recorded at 2-day intervals for 7 weeks, then at weekly intervals until plants matured (in about 10 weeks). The experiment was repeated once. Plant Pathology (2010) 59, 712–720 715 Results ITS sequences PCR using primers ITS1 and ITS4 generated products of about 650 bp from most samples. Full length (646 bp) or partial sequence of the ITS region was obtained from all the samples and representative sequences were used in BLAST searches (Table 1). All ITS sequences obtained from a single glasshouse at a given time were virtually identical (similarity more than 99Æ4%) and therefore considered as one biological sample. Sequences of samples obtained from pea fell into two discrete groups (referred to as Group I and Group II in Table 1 and hereafter). Fourteen single nucleotide polymorphisms (SNPs) were found between these two groups when 602 nucleotides of the ITS region were compared. Group I included five field samples (FF 06, SP 07, GE 07, LI 08-1 and LI 08-2) and three glasshouse samples (GH 04, GH 07-119 and GH M 07). Group II included four glasshouse samples (GH 05, GH 06, GH N 07 and Lif 07) and two field samples (EI 08-1 and EI 08-2). Although sequences for a few samples (e.g. Lif 07, Table 1) were incomplete, assignment to one of the two groups was easily accomplished based on some of the 14 unique SNPs. BLAST search using the FF 06 sample (Group I) as query had 99% similarity (one nucleotide difference) to E. pisi (Accession AF011306 from Lathyrus latifolius deposited by Saenz & Taylor (1999)). The sequences in GenBank showing the next highest similarities (fourteen base pair differences) were eleven identical sequences (EF196666 to EF196675 and AY739112) for E. diffusa deposited by Almeida et al. (2008), and several ‘Oidium sp.’ (AB078800 etc.) deposited by Takamatsu et al. (2002). In the BLAST search using the GH N 07 sequence (Group II) as a query, the sequences in GenBank that showed the highest similarity (one base-pair difference) were three identical sequences (AB079853 to AB079855) of Oidium sp. from three different hosts from Japan which were in an E. bauemleri ⁄ E. trifolii clade (Fig. 1 in Okamoto et al., 2002). The sequences in GenBank that showed the next highest similarity (three base pair differences) were five identical sequences (AB015913, AB163926, AB167523, AB167524 and AF298542) of E. trifolii (Takamatsu et al., 1999; Cunnington et al., 2003; Matsuda et al., 2005) and another sequence (AB015933) of E. baeumleri (Takamatsu et al., 1999). Phylogenetic analyses Of 570 total characters (nucleotides), 93 characters were phylogenetically informative for parsimony analysis. Parsimony analysis using PAUP* 4.0b10 generated 24 equally most parsimonious trees. Tree topologies were almost consistent among all the trees except for branch lengths and branching orders of the terminal branches. The majority rule consensus tree is shown in Fig. 1. For ML analysis, DT-ModSel selected shape parameter alpha, 0Æ295144, HKY85 + G model, transition ⁄ transversion 716 R. N. Attanayake et al. Figure 1 Majority rule consensus tree (unrooted) based on the internal transcribed spacer (ITS) sequences from 28 taxa of Erysiphe spp. showing the relationship among strains of several powdery mildew species found on legumes. Bootstrap values based on 1000 replications are shown above the branches. Roman numerals at the right of taxa indicate the ITS sequence grouping in Table 1. Taxa in bold are the sequences determined in this study. ratio 1Æ30921 (kappa = 2Æ6403874) and nucleotide frequencies as A = 0Æ19757, C = 0Æ28096, G = 0Æ27493, T = 0Æ24653. Tree topology of the ML tree was similar to that of the MP tree (not shown). Thirteen sequences obtained in this study fell into two major clades (Fig. 1). Clade I (97% bootstrap support) consisted of E. pisi GenBank sequences obtained from Lathyrus latifolius (AF011306) and P. sativum (AF073348) hosts from USA and Australia, respectively, and most of the pea field samples, some glasshouse samples and a sample from Lathyrus sp. obtained in this study. Clade II had 93% bootstrap support and comprised the rest of the glasshouse and field samples obtained in this study and GenBank sequences E. trifolii and E. baeumleri obtained from Vicia (AB015919 and AB015933) and E. trifolii (AB015913 and AF298542) from Japan and Switzerland, respectively. Erysiphe diffusa formed a separate distinct clade with 100% bootstrap support. Surprisingly, the ITS sequence (AB104519) of a fungus deposited as E. pisi from Melilotus sativa from Iran is distinct from all the other E. pisi sequences. Morphological observations All samples used in this study displayed typical powdery mildew symptoms. Mycelia of all samples were mainly epiphyllous, in white, effuse patches often covering the entire adaxial and abaxial surfaces of leaves, stems and sometimes pods. Hyphae were branched, septate, hyaline, thin-walled; lobed appressoria were solitary or in opposite pairs. Single conidia formed terminally on conidiophores. In all samples conidiophore foot cells were erect, straight to sometimes flexuous, and cylindrical. After detecting ITS sequence identity among the several samples obtained from a given glasshouse, one set of morphological measurements was taken into account. Conidial dimensions were graphically depicted using box plots in Fig. 2. Conidial lengths and widths for E. pisi ranged from 23Æ5–60 · 7Æ5–20Æ5 lm, whereas those for E. trifolii were 23–60Æ5 · 9Æ5–19 lm (Fig. 2). Similarly, the conidial dimensions for the two species in Braun (1987) also overlapped (24–55 · 13Æ5–22 lm for E. pisi and 30– 45 · 16–21 lm for E. trifolii). Conidiophore foot cell measurements also varied considerably amongst the samples (data not shown). Chasmothecia were observed in six P. sativum samples (FF 06, GH 04, GE 07, GH N 07, Lif 07 and SP 07). Chasmothecia were scattered or gregarious, with irregularly polygonal peridial cells. Mature dark brown chasmothecia enclosed several sessile or short-stalked asci. Ascospores were ellipsoid to ovoid. Two kinds of chasmothecial appendages were observed (Table 1). Chasmothecial appendages of the samples FF 06, GH 04, GE 07 and SP 07 (from Group I) were short, mycelioid, simple, septate, brown coloured at the base, becoming pale towards the tip and hyaline at the upper half and often interwoven with each other and with other mycelia (Fig. 3a). Chasmothecia of samples GH N 07 and Lif 07 (from Group II) displayed long, flexuous, dichotomously branched appendages (Fig. 3b). Appendage apices were straight, 3–5 times loosely branched, diffuse, and often deeply cleft (Fig. 3c). Removal of Group II chasmothecia from the leaf surface was accomplished more easily than was the case for Group I samples because the latter had appendages interwoven with the surrounding Plant Pathology (2010) 59, 712–720 Erysiphe trifolii on Pisum sativum 717 and for Group II was 4–7, whereas in Braun (1987) it is 0Æ5–3Æ5 and 2–6 for E. pisi and E. trifolii, respectively. The highly branched chasmothecial appendage apices observed in Group II samples differed from descriptions of E. trifolii in Braun (1987, 1995) but agreed with the descriptions of E. trifolii on lentils in Attanayake et al. (2009) and with apices on an authentic specimen of E. trifolii as discussed below. Pathogenicity assays Figure 2 Box plots showing variations of conidial lengths and widths among 14 samples of Erysiphe pisi (hatched boxes) and E. trifolii (open boxes) used in this study. The boxes and middle lines represent the middle 50 percentiles and medians, respectively. The whiskers represent upper and lower limits and asterisks represent outliers. hyphae. Most of the other characters measured (such as length and width of conidiophore foot cells) displayed overlapping values between the two groups as well as amongst the individual samples (data not shown). However, the ratio of chasmothecial appendage length to chasmothecial diameter is clearly different between the two groups. Ratios for Group I and Group II agreed with the descriptions for E. pisi and E. trifolii, respectively, in Braun (1987). The ratio of chasmothecial appendage length to chasmothecial diameter for Group I was 1–3 (a) (b) The ITS sequences of the three samples that were used for inoculation belonged to Group II (E. trifolii). In the detached leaf assay, signs of powdery mildew were visible on leaves of M. albus and Lens culinaris about 10 days after inoculation with conidia obtained from P. sativum. Similarly, signs of powdery mildew were visible after the same time interval on P. sativum leaves when conidia from M. albus or Lens culinaris were used. Mock-inoculated controls remained free of symptoms during the entire period of the experiment. Repeating the experiment gave the same results. ITS sequences of conidia from the infected leaves were identical to those of the inoculum used. In the observational study in the glasshouse all four pea cultivars were heavily infected by powdery mildew, exhibited chasmothecial production, and the ITS sequence of the samples from the pea plants were of Group II (E. trifolii). However, the two soybean genotypes (L84-2237 and Harosoy, susceptible to E. diffusa) remained symptomless during the entire period of the experiment. The same results were obtained when the experiment was repeated. Discussion Powdery mildew is a recurring problem in glasshousegrown pea in the US Pacific Northwest. Such frequent occurrence of powdery mildew in glasshouses has facilitated selection of resistance materials in breeding programmes. However, some of the resistant materials (c) 0.1 mm 50 µm 50 µm Figure 3 Comparison of chasmothecial appendages of Erysiphe pisi and E. trifolii. (a) Short myceliod chasmothecial appendages of E. pisi. (b) Long flexuous chasmothecial appendages (arrow) of E. trifolii. (c) Highly branched appendage apex of E. trifolii (arrow). Plant Pathology (2010) 59, 712–720 718 R. N. Attanayake et al. selected in the glasshouse are susceptible to powdery mildew in the fields (K.E. McPhee, unpublished data). The inoculum sources in the glasshouse are often unknown, but because of disinfestation procedures they must originate from outside the glasshouse. Both teleomorphic characters and ITS sequences were used to distinguish pathogen species found on pea in both glasshouse and field in the US Pacific Northwest. Based on ITS sequences, two distinct groups of powdery mildew pathogens were found. The two groups differed in 14 nucleotide positions in the ITS region and also exhibited readily distinguishable chasmothecial appendage morphology. The correlation of morphological differences with ITS sequence differences confirms that the two groups belong to different species of the genus Erysiphe. Group I had ITS sequences most similar to those of previously deposited sequences of E. pisi (Saenz & Taylor, 1999; Cunnington et al., 2003), and produced chasmothecia with short mycelioid, simple appendages conforming to the descriptions of E. pisi, a well-documented pathogen of pea. However, anamorphic characters could not be used to differentiate the two species due to intraspecific variations (Fig. 2). The variations could be due to genetic differences and environmental conditions such as relative humidity. For example, relatively larger conidia were consistently observed in detached leaf assays conducted in moist chambers (high relative humidity) than those on the field samples (R.N.Attanayake, unpublished data). Determination of the species identity of the Group II samples was more complicated. ITS sequences of Group II were most similar to those of the E. trifolii complex (Takamatsu et al., 1999; Okamoto et al., 2002; Cunnington et al., 2003; Khodaparast et al., 2003). However, members of this group produced branched chasmothecial appendages, resembling those described for E. diffusa or E. pisi var. cruchetiana in Braun (1987). Neither E. diffusa, E. pisi var. cruchetiana nor E. trifolii has previously been reported to infect pea. Erysiphe diffusa produces dichotomously branched, rigid chasmothecial appendages (Braun, 1987), whereas E. trifolii was not reported to produce dichotomously branched chasmothecial appendages in Braun (1987). A recent study (Attanayake et al., 2009) employing ITS sequences, plus morphological observation of an authentic specimen of E. trifolii (WSP 70928) originating from GZU Dupla Fungorum that was determined by U. Braun (Scheuer, 2003), demonstrated that E. trifolii can produce long, flexuous and dichotomously branched chasmothecial appendages. Erysiphe trifolii differs from E. pisi var. cruchetiana by producing frequently dichotomously branched long, flexuous appendages, whereas the latter has irregularly branched, short appendages. These findings help differentiate E. trifolii from E. pisi on pea. Erysiphe trifolii can also be differentiated from E. diffusa, in spite of the fact that both species may produce highly dichotomously branched appendage apices, on the basis of the flexuous nature of appendages in E. trifolii (Fig. 3b). Erysiphe diffusa is a well docu- mented pathogen of soybean (Dunleavy, 1978; Lohnes & Nickell, 1994) and was also tentatively identified (on the basis of appendage morphology) as a causal agent of powdery mildew on lentil in Canada (Banniza et al., 2004). The powdery mildew pathogen from pea did not cause any visible disease symptoms on soybean genotypes known to be susceptible to E. diffusa, and ITS sequence of E. diffusa from wild soybean was distinct from those of the pea powdery mildew pathogens (Attanayake et al., 2009). These findings indicate that the pea powdery mildew samples constituting Group II are not referable to E. pisi or E. diffusa, but are in fact E. trifolii. Phylogenetic analysis further supported this conclusion. The phylogenetic analysis in this study revealed two groups of powdery mildews: Group I was in congruence with the morphological characters of E. pisi, and formed a clade with all previously identified E. pisi sequences except the Iranian sample (AB104519) from Melilotus sativa, whereas Group II was in congruence with E. trifolii. In addition, E. baeumleri was also grouped with E. trifolii. There are taxonomic ambiguities in the placement of E. baeumleri. It is not clear whether E. trifolii and E. baeumleri are conspecific or two distinct species (U. Braun, Martin-Luther-Universität, Institut für Biologie, Germany, personal communication). Erysiphe trifolii has been regarded as a complex of similar species consisting of E. trifolii, E. baeumleri and E. asteragali (Braun, 1987). However, in this analysis the bootstrap support to separate the E. baeumleri clade was only 71%. The species identity of the Iranian sample (AB104519) from M. sativa needs reassessment because the ITS sequence was drastically different from those of the other E. pisi sequences. Results of this study demonstrated that both E. pisi and E. trifolii are present and cause disease on pea in both field and glasshouse conditions. The sample Lif 07 was from pea cv. Lifter which is resistant to E. pisi (McPhee & Muehlbauer, 2002), but severely infected by E. trifolii in the glasshouse. Ondřej et al. (2005) reported a similar situation: Pisum sativum germplasm lines resistant to E. pisi were susceptible to E. baeumleri in fields in the Czech Republic. Therefore, powdery mildew fungi infecting pea are more diverse than previously assumed (Braun, 1987; Falloon & Viljanen-Rollinson, 2001). These findings may explain some inconsistent responses of individual pea breeding lines to powdery mildew infection. Glasshouse infections may not adequately reflect host ranges in natural or field situations. Cook & Fox (1992) observed that Vicia faba (faba bean) grown in glasshouses were infected with E. pisi var. pisi while such infection is not reported in fields in Britain. Okamoto et al. (2002) also found Oidium subgenus Pseudoidium causing powdery mildew on its non-natural host, Eustoma grandiflorum (prairie gentian), under glasshouse conditions. Taylor (2008) suggested that climatic change may have favoured increased aggressiveness of existing pathogen races, causing the breakdown of red clover resistance to powdery mildew. Nevertheless, in the present work E. trifolii was observed on pea in both fields and glasshouses and Plant Pathology (2010) 59, 712–720 Erysiphe trifolii on Pisum sativum formed the teleomorph on pea (Table 1), indicating that infection of pea by E. trifolii is not a glasshouse artifact. During winter months only a small acreage of winter pea is grown in the field in the US Pacific Northwest and is at a significant distance from the glasshouse facilities, suggesting that powdery mildew inoculum for the glasshouse plants likely comes from volunteer pea plants or alternative hosts, or from resting states (chasmothecia) on plant debris. Both E. pisi and E. trifolii were found on other legumes (Lathyrus sp., Medicago lupulina, Melilotus albus and Lens culinaris) commonly found in the US Pacific Northwest. These legume species could be inoculum sources for glasshouse grown peas during winter months. Although both E. pisi and E. trifolii were detected in glasshouses, at a given time only one of the species (based on ITS sequences) was detected in a given glasshouse. This suggests that the winter inoculum, originating external to glasshouses, was very limited during the colder months but propagated rapidly upon gaining entry to a suitable internal environment. This report is the first to document E. trifolii causing powdery mildew on pea, and documents the disease in both field and glasshouse conditions. It is likely that E. trifolii has been a pathogen of pea for a long time, but has not been recognized until now. Recognition of E. trifolii as a pea pathogen is significant for pea breeding programmes. Pathogen species identity is important because different species may interact with pea genotypes differently. Although it may be difficult to completely exclude a given powdery mildew species from a glasshouse, attempts can be made to introduce the desired powdery mildew species into glasshouse screenings or at least determine which species is being screened at a given time. Acknowledgements The authors want to thank Professor Uwe Braun, MartinLuther-Universität, Institut für Biologie, Bereich Geobotanik, Herbarium, Neuwerk 21, D-06099 Halle ⁄ S. Germany for advice concerning the taxonomic status of Erysiphe trifolii, Dr Eric H. Roalson, Department of Biology, Washington State University, Pullman, WA for helping in phylogenetic analysis, and Shari Lupien, Plant Introduction, USDA ARS, Pullman for maintaining Melilotus albus plants infected with powdery mildew. The research was funded in part by the USDA-CSREES Cool Season Food Legume Research Program. References Almeida AMR, Binneck E, Piuga FF, Marin SRR, Ribeiro do Valle PRZ, Silveira CA, 2008. Characterization of powdery mildew strains from soybean, bean, sunflower and weeds in Brazil using rDNA-ITS sequences. Tropical Plant Pathology 33, 20–6. Attanayake RN, Glawe DA, Dugan FM, Chen W, 2009. Erysiphe trifolii causing powdery mildew of lentil (Lens culinaris). Plant Disease 93, 797–803. Plant Pathology (2010) 59, 712–720 719 Banniza S, Parmelee JA, Morrall RAA, Tullu A, Beauchamp CJ, 2004. First record of powdery mildew on lentil in Canada. Canadian Plant Disease Survey 84, 102–3. Braun U, 1987. A monograph of the Erysiphales (Powdery Mildews). Beiheft zur Nova Hedwigia 89, 1–700. Braun U, 1995. The Powdery Mildews (Erysiphales) of Europe. New York, USA: Gustav Fischer Verlag. Cook RTA, Fox RTV, 1992. Erysiphe pisi var. pisi on faba beans and other legumes in Britain. Plant Pathology 41, 506–12. Cunnington JH, Takamatsu S, Lawrie AC, Pascoe IG, 2003. Molecular identification of anamorphic powdery mildews (Erysiphales). Australasian Plant Pathology 32, 421–8. Cunnington JH, Lawrie AC, Pascoe IG, 2005. Molecular identification of Golovinomyces (Ascomycota, Erysiphales) anamorphs on the Solanaceae in Australia. Australasian Plant Pathology 34, 51–5. Dunleavy JM, 1978. Soybean seed yield losses caused by powdery mildew. Crop Science 18, 337–9. Epinat C, Pitrat M, Bertrand F, 1993. Genetic analysis of resistance of five melon lines to powdery mildews. Euphytica 65, 135–44. Falloon RE, Viljanen-Rollinson SLH, 2001. Powdery mildew. In: Kraft JM, Pfleger FL, eds. Compendium of Pea Disease and Pests. St Paul, MN, USA: APS Press, 28–9. Felsenstein J, 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39, 783–91. Fondevilla S, Torres AM, Moreno MT, Rubiales D, 2007. Identification of a new gene for resistance to powdery mildew in Pisum fulvum, a wild relative of pea. Breeding Science 57, 181–4. Glawe DA, 2008. The powdery mildews: a review of the world’s most familiar (yet poorly known) plant pathogens. Annual Review of Phytopathology 46, 27–51. Gritton ET, Ebert RD, 1975. Interaction of planting date and powdery mildew on pea plant performance. American Society of Horticultural Science 100, 137–42. Hall TA, 1999. BioEdit: a user friendly biological sequence alignment editor and analysis program for Windows 95 ⁄ 98 ⁄ NT. Nucleic Acids Symposium Series 41, 95–8. Heringa RJ, Van Norel A, Tazelaar MF, 1969. Resistance to powdery mildew (Erysiphe poligoni DC.) in pea (Pisum sativum L.). Euphytica 18, 163–9. Khodaparast SA, Hedjaroude GA, Takamatsu S, 2003. Phylogenetic relationships between Iranian isolates of Microsphaera and Erysiphe s. lat. based on rDNA internal transcribed spacers sequences. Rustaniha 4, 79–86. Lim TM, 1973. A rapid laboratory method of assessing susceptibility of Hevea clones to Oidium heveae. Experimental Agriculture 9, 275–9. Lohnes DG, Nickell CD, 1994. Effects of powdery mildew alleles Rmd-c, Rmd, and rmd on yield and other characteristics in soybean. Plant Disease 78, 299–301. Matsuda Y, Sameshima T, Moriura N et al., 2005. Identification of individual powdery mildew fungi infecting leaves and direct detection of gene expression by single conidium polymerase chain reaction. Phytopathology 95, 1137–43. McPhee KE, Muehlbauer FJ, 2002. Registration of Lifter green dry pea. Crop Science 42, 1377–8. Minin V, Abdo Z, Joyce P, Sullivan J, 2003. Performance-based selection of likelihood models for phylogeny estimation. Systematic Biology 52, 674–83. 720 R. N. Attanayake et al. Okamoto J, Limkaisang S, Nojima H, Takamatsu S, 2002. Powdery mildew of prairie gentian: characteristics, molecular phylogeny and pathogenicity. Journal of General Plant Pathology 68, 200–7. Ondřej M, Dostălovă R, Odstrčilovă L, 2005. Response of Pisum sativum germplasm resistant to Erysiphe pisi to inoculation with Erysiphe baeumleri, a new pathogen of peas. Plant Protection Science 41, 95–103. Saenz GS, Taylor JW, 1999. Phylogeny of the Erysiphales (powdery mildews) inferred from internal transcribed spacer ribosomal DNA sequences. Canadian Journal of Botany 77, 150–68. Saitou N, Nei M, 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Molecular Biology and Evolution 4, 406–25. Scheuer C, 2003. Dupla Fungorum, Supplementum (2003), verteilt vom Institut für Botanik der Universität Graz (GZU). Fritschiana (Graz) 40, 1–51. Spurr HW, 1979. Ethanol treatment: a valuable technique for foliar biocontrol studies of plant disease. Phytopathology 69, 773–6. Swofford DL, 2002. PAUP*: Phylogenetic Analysis Using Parsimony (and other methods) 4Æ0 Beta. Sunderland, MA, USA: Sinauer Associates. Takamatsu S, Hirata T, Sato Y, Nomura Y, 1999. Phylogenetic relationships of Microsphaera and Erysiphe section Erysiphe (powdery mildews) inferred from the rDNA ITS sequences. Mycoscience 40, 259–68. Takamatsu S, Shin HD, Paksiri U et al., 2002. Two Erysiphe species associated with recent outbreak of soybean powdery mildew: results of molecular phylogenetic analysis based on nuclear rDNA sequences. Mycoscience 43, 333–41. Taylor NL, 2008. A century of clover breeding developments in the United States. Crop Science 48, 1–13. Tiwari KR, Penner GA, Warkentin TD, Rashid KY, 1997. Pathogenic variation in Erysiphe pisi, the causal organism of powdery mildew of pea. Canadian Journal of Plant Pathology 19, 267–71. Warkentin TD, Rashid KY, Zimmer RC, 1995. Effectiveness of a detached leaf assay for determination of the reaction of pea plants to powdery mildew. Canadian Journal of Plant Pathology 17, 87–9. White TJ, Bruns T, Lee S, Taylor J, 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: Innis MA, Gelfand DH, Sninsky JJ, White TJ, eds. PCR Protocols: A Guide to Methods and Applications. San Diego, CA, USA: Academic Press Inc, 315–22. Plant Pathology (2010) 59, 712–720