Microbial Ecology. Diversity of Peronosporomycete (Oomycete) Communities Associated with the Rhizosphere of Different Plant Species

Transcription

1 Microbial Ecology Diversity of Peronosporomycete (Oomycete) Communities Associated with the Rhizosphere of Different Plant Species Jessica M. Arcate, Mary Ann Karp and Eric B. Nelson Department of Plant Pathology, Cornell University, 334 Plant Science Building, Ithaca, NY 14853, USA Received: 15 September 2004 / Accepted: 12 January 2005 / Online Publication: 3 January 2006 Abstract Peronosporomycete (oomycete) communities inhabiting the rhizospheres of three plant species were characterized and compared to determine whether communities obtained by direct soil DNA extractions (soil communities) differ from those obtained using baiting techniques (bait communities). Using two sets of Peronosporomycete-specific primers, a portion of the 5 0 region of the large subunit (28S) rrna gene was amplified from DNA extracted either directly from rhizosphere soil or from hempseed baits floated for 48 h over rhizosphere soil. Amplicons were cloned, sequenced, and then subjected to phylogenetic and diversity analyses. Both soil and bait communities arising from DNA amplified with a Peronosporomycetidae-biased primer set (Oom1) were dominated by Pythium species. In contrast, communities arising from DNA amplified with a Saprolegniomycetidae-biased primer set (Sap2) were dominated by Aphanomyces species. Neighbor-joining analyses revealed the presence of additional taxa that could not be identified with known Peronosporomycete species represented in GenBank. Sequence diversity and mean sequence divergence ( ) within bait communities were lower than the diversity within soil communities. Furthermore, the composition of Peronosporomycete communities differed among the three fields sampled and between bait and soil communities based on F st and parsimony tests. The results of our study represent a significant advance in the study of Peronosporomycetes in terrestrial habitats. Our work has shown the utility of cultureindependent approaches using 28S rrna genes to assess the diversity of Peronosporomycete communities in association with plants. It also reveals the presence of potentially new species of Peronosporomycetes in soils and plant rhizospheres. Correspondence to: Eric B. Nelson; ebn1@cornell.edu Introduction The Peronosporomycetes are a large, ecologically, and phylogenetically distinct group of eukaryotes found most commonly in terrestrial and aquatic habitats. They include well-known genera of plant pathogens such as Aphanomyces, Peronospora, Phytophthora, and Pythium, most of which are soil-borne and infect subterranean plant parts such as seeds, roots, and hypocotyls. This group also includes other important genera such as Saprolegnia, Achlya, and Lagenidium, which are pathogenic to fish, insects, crustaceans, and mammals [17]. Although these organisms have received much attention in terms of the diseases they cause, few other details of their ecology are known. For many years, Peronosporomycetes were believed to be closely related to fungi. It was assumed, therefore, that they shared similar ecological traits. However, it is now quite clear that Peronosporomycetes share no close evolutionary relationships with the true fungi [4, 59, 62, 67]. Rather, they are closely related to the heterokont algae and hyphochytrids [5, 6, 68]. Peronosporomycetes are currently classified within the newly erected Kingdom Straminipila (previously known as Chromista) [8, 18], which is believed to represent one of the more diverse assemblages of organisms on earth [5]. Two distinct subclasses exist within the Peronosporomycetes: the Peronosporomycetidae and the Saprolegniomycetidae. This classification is well supported by numerous molecular phylogenetic studies based on 18S rdna [22], ITS sequences [12], cytochrome oxidase II [11, 31, 43], and, increasingly, 28S rdna [48 50]. Although estimates vary, there are now around 1200 known species in 88 genera [18]. The relationships among some genera and species are still uncertain [10, 27, 32, 37, 49, 73]. Despite the diversity and importance of the Peronosporomycetes, little is known of their distribution and 36 DOI: /s y & Volume 51, (2006) & * Springer Science+Business Media, Inc. 2006

2 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY 37 roles in various habitats, and few contemporary ecological studies of this group have been conducted. This is partly because of the fact that Peronosporomycetes seldom show up in standard culture-based methods commonly used for isolating true fungi from environmental samples. Instead, a variety of baiting techniques [26] have become the standard means of isolating Peronosporomycetes from environmental samples and determining their occurrence, distribution, and diversity. However, baiting has a number of shortcomings when used to assess species occurrence and distribution. First and most importantly are biases because of selective colonization and development on different types of baits [36, 55]. Even if a particular species can initially colonize baits, some species may competitively exclude others during the incubation process, leaving relatively few species to dominate baits. The number of species that have traditionally been described from baits is relatively limited; generally, fewer than 10 species have been described from a given sample in most studies. Second, the very nature of the baiting system selects species that produce zoospores under the conditions of the laboratory incubation [14], leaving nonzoospore-producing species or species not developmentally in a state to release zoospores to go undetected. Of particular importance in the latter case are oospore populations. Because oospores serve as survival structures and likely constitute a large proportion of Peronosporomycete biomass in soils [16], assessments of diversity based on baiting are likely to be underestimated. Many of the seminal studies of Peronosporomycete diversity were conducted between the 1920s and the 1970s [15]. These studies consisted largely of surveys conducted in various terrestrial habitats, ranging from natural forested and grassland sites to swamps, ditches, littoral mud, and agricultural soils (e.g., [1, 19, 25, 34, 35]). What emerges from these and other studies is that Peronosporomycetes are worldwide in their distribution and are found in nearly all soil types and soil habitats. However, the factors that regulate their occurrence and distribution in terrestrial habitats are unknown. It has been suggested that some terrestrial Peronosporomycete species might be restricted to particular habitats because of the type of vegetation cover [3]. A growing body of evidence from other microbes indicates that plants themselves are a major selective force in determining the nature of the microbial communities with which they associate (e.g., [41, 56, 57]). Among Peronosporomycetes, few data are available, but there is good evidence that Pythium species have a profound impact on the spatial distribution of cherry trees [46, 47]. Recent molecular ecology studies of bacterial communities in terrestrial habitats have revealed an incredible diversity of previously nondescribed and potentially nonculturable organisms [61]. Similar discoveries are being made from analyses of fungal [53, 71] and other stramenopile communities [44, 45]. However, primer sequences used for analysis of these communities have not been effective in detecting Peronosporomycetes [58, 70], and no molecular-based approach has yet been used to study a broad range of Peronosporomycetes in terrestrial samples. The purpose of this study was to utilize a molecular ecological approach to test the hypothesis that direct DNA extractions and amplifications from rhizosphere soil give rise to Peronosporomycete communities that differ from those determined by traditional baiting. We tested this hypothesis in soils with different cropping histories and planted with different plant species. Materials and Methods Sampling Site. Samples were collected on 10 September 2003 from a Howard gravelly loam soil (ph 5.5) at the Cornell Vegetable Research Facility, Freeville, NY. Soils were planted to tomato (Lycopersicon esculentum), butternut squash (Cucurbita moschata), or sorghum (Sorghum spp.). Sampled soils were collected from adjacent fields, all of which were in nearly identical microclimates and with identical soil types. However, each field had a different cropping history. The tomato field had been in a tomato and winter rye (Secale cereale) rotation for several consecutive years. The winter rye had been plowed under before the tomato crop was planted. The sorghum field was in an alternating year rotation with potatoes (Solanum tuberosum). However, 2003 was the first year that sorghum was grown, and for all previous odd years, the crop was rye. The butternut squash field cropping history was inconsistent. Butternut squash and winter rye were planted in both 2002 and 2003, but the preceding years had been planted with a range of crops including melon (Cucumis melo), peppers (Capsicum spp.), sweet corn (Zea mays), tomato, sorghum, and rye. Rhizosphere soil samples from tomato and squash were collected from five randomly selected plants established in rows. Soil adjacent to the roots of individual plants in each of four replicate rows was removed to a depth of 15 cm. Sorghum rhizosphere soils were sampled on a diagonal transect, and four replicate samples were taken from randomly selected plants along the transect. Replicate samples from each plant species were combined for a total soil volume of approximately 0.5 L per plant species. These combined soils were each thoroughly mixed by repeated turning and shaking in polyethylene bags. Samples were transported to the laboratory and frozen until use. Baiting. Peronosporomycetes were obtained from each of the three rhizosphere soil samples using a mod-

3 38 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY ification of traditional baiting techniques [26]. Two grams of rhizosphere soil were placed into a sterile mm petri dish and flooded with 20 ml sterile water. Three scored and autoclaved hempseeds were added to each dish and were then incubated in darkness at 18-C for 48 h. Hempseeds were then transferred aseptically to an antibiotic solution [25] and were then incubated at 18-C in darkness for 14 days. The antibiotic solution contained per 1 L sterile distilled deionized H 2 O (ddw) 0.2 ml pimaricin to suppress fungi and 100 mg ampicillin to suppress bacteria. DNA was then extracted directly from baits. In some preliminary experiments, baits were plated on water agar containing 50 mg/ml rifampicin and penicillin G (to suppress bacteria and fungi, respectively) and were then incubated for 4 days to recover viable Peronosporomycete cultures. Peronosporomycete Primer Design. A5 0 region of the large subunit rdna that spans the variable D1 and D2 regions of the 28S rrna gene [7] was chosen for amplification using the primer sets Oom1 and Sap2. This region was chosen because there are well-developed, taxonomically defined datasets available for it and because it allows for phylogenetic examination at both broad and narrow taxonomic scales. The Oom1 primer set was designed from a consensus sequence derived from a broad range of Peronosporomycete species spanning the Peronosporomycetidae and the Saprolegniomycetidae. The Sap2 primer set, on the other hand, was designed from a consensus sequence derived solely from members of the Saprolegniomycetidae. Primers were designed using the program Primer Select 5.07 (DNASTAR, Madison, WI). The Oom1 primer set (Oom1F 5 0 -GTGCGAGACCGATAGCGA ACA-3 0 and Oom1R 5 0 -TCAAAGTCCCGAACAGCAAC AA-3 0 ) is located between positions homologous to 348 and 816 of the Phytophthora megasperma (GenBank X75631) 28S rrna gene [67] and amplifies a 468-bp product. The Sap2 primer set (Sap2F 5 0 -AGCATAGCGA TTTGGGATAAGTC-3 0 and Sap2R 5 0 -GTAGGCACCTC AGTCTCAACCA-3 0 ) is located between positions 123 and 567 of the Saprolegnia ferax (GenBank AF235953) 28S rrna gene [48] and amplifies a 444-bp product. Initial trials of the Oom1 primer set showed a high degree of specificity for Peronosporomycetes over other eukaryotes and prokaryotes at annealing temperatures above 56-C. Although this primer pair amplifies 28S rdna sequences from a broad range of Peronosporomycete genera from both the Peronosporomycetidae and Saprolegniomycetidae subclasses when screened in the laboratory, amplifications from soils and baits were somewhat biased toward members of the subclass Peronosporomycetidae. Similarly, the Sap2 primer pair was biased toward members of the Saprolegniomycetidae but also amplified 28S rrna genes from members of the Peronosporomycetidae at annealing temperatures below 56-C. DNA Extraction and PCR Conditions. Peronosporomycete DNA was isolated both from rhizosphere soil samples and baits using Ultracleani Soil DNA Isolation Kits (MoBio Laboratories, Inc.) according to the manufacturer s instructions. For DNA extractions from rhizosphere soil, 0.5 g of soil was used. DNA was also extracted from three hempseed baits for each rhizosphere soil sample. DNA was further purified with the Ultracleani PCR Clean-Up Kit (MoBio Laboratories, Inc.). Two replicate extractions were conducted for subsequent amplification, cloning, and sequencing. Polymerase chain reactions (PCRs) for the Oom1 primer set contained 10 mm Trizma HCl, ph 8.3, 50 mm KCl, 2.5 mm MgCl 2, 200 mm of each deoxyribonucleotide triphosphate (dntp), 0.2 mm each of Oom1f and Oom1r, 2 U of Taq DNA polymerase, and 1.0 ml template DNA per 25-mL reaction. DNA was amplified with a Bio-Rad MyCycleri thermal cycler using the following program: initial denaturation at 94-C for 5 min, followed by 35 cycles of denaturation at 94-C for30s, annealing at 58-C for 30 s, extension at 72-C for 30 s, and a final extension at 72-C for 5 min. PCR products were either used immediately or stored at 4-C prior to subsequent analyses. Two replicate reactions were run for each sample. PCRs for the Sap2 primer set contained 10 mm Trizma HCl, ph 8.3, 50 mm KCl, 2.5 mm MgCl 2, 0.2 mm of each of Sap2f and Sap2r, 200 mm of each dntp, 4 U of Taq DNA polymerase, and 1.0 ml template DNA per 50-mL reaction. DNA was amplified using the following PCR program: initial denaturation at 94-C for 5 min, followed by 35 cycles of denaturation at 94-C for 30 s, annealing at 56-C for 30 s, extension at 72-C for 30 s, and a final extension at 72-C for 10 min. Two replicate extractions and amplifications with the Oom1 primer set were conducted to determine whether the sampled communities were the same. Both F st statistics and parsimony tests (see below) were not significant (data not shown), indicating that the communities arising from the separate extractions were not significantly different. Sequences from each of these experiments were therefore combined for all subsequent analysis. Amplicons from the Sap2 primer set were only obtained from one DNA extraction. Cloning and Sequencing of 28S rrna Genes. Polymerase chain reaction products from rhizosphere soils and baits were cloned using INVaF 0 competent cells with the pcr \ 2.1 vector from the TA Cloning \ Kit (Invitrogen, Carlsbad, CA). To concentrate the DNA, two 50-mL PCRs for each sample were pooled and were then concentrated during purification using the MoBio UltraCleani PCR Clean-up Purification Kit (MoBio

4 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY 39 Laboratories, Inc.). This step insured that the PCR product was free of excess salts, dntps, and primers, and that we could obtain at least 6 ng DNA in 1 ml ofproduct.a ligation protocol of 1:1 vector-to-insert ratio yielded the best cloning efficiency results. Forty-eight clones of each sample were picked into 96-well deep-well plates containing Luria Bertani and kanamycin and were incubated for 16 hat37-c shaking at 225 rpm. Plasmids were then purified using the Wizard \ SV 96 Plasmid DNA Purification System (Promega, Madison, WI). To determine if the insert was present, plasmid DNA was amplified using the M13 primer set [M13f (5 0 - GTAAAACGACGGCCAG-3 0 ) and M13r (5 0 -CAGGAAA CAGCTATGAC-3 0 )]. PCRs for the M13 primer set consisted of 10 mm Trizma HCl, ph 8.3, 50 mm KCl, 2.5 mm MgCl 2, 200 mm of each dntp, 0.2 mm of each of M13f and M13r primers, 1 U of Taq DNA polymerase, and 0.5 ml template DNA per 25-mL reaction. DNA was amplified using the following program: initial denaturation at 94-C for 5 min, followed by 35 cycles of denaturation at 94-C for 30 s, annealing at 50-C for 30 s, extension at 72-C for 30 s, and a final extension at 72-C for 5 min. PCR products were electrophoresed to screen for the presence of the insert. Purified plasmid DNA that contained the insert was mixed with the M13f primer (forward direction) and was then submitted to the Cornell Bioresource Center Sequencing Facility. Sequencing was performed on an Applied Biosystems Automated 3730 DNA Analyzer using Big Dye Terminator chemistry and AmpliTaq- FS DNA Polymerase. Sequences were compiled in Sequencher 4.1 (Gene Codes Corp, Ann Arbor, MI, USA) or EditSeq 5.07 (DNASTAR), and selected representatives were submitted to GenBank with assigned accession numbers of AY AY Sequence Alignments and Phylogenetic Analyses. Raw sequences were edited using EditSeq to manually remove vector sequences and eliminate poorly resolved regions. These edited sequences were aligned using either MegAlign 5.07 (DNASTAR) or Clustal X version 1.83 [9], both using the Clustal W algorithms [60]. Separate alignments were generated for sequences obtained from DNA extracted from baits, sequences obtained from DNA extracted directly from soil, and combined bait and soil sequences. A number of tomato rhizosphere sequences obtained from DNA extracted directly from soil could not be used in these analyses because they were too divergent to be aligned to other soil and bait sequences. Although BLAST analysis suggested that they were Peronosporomycete sequences (best BLAST hits were AY Pythium sp. AR235 and AF Pythium sp. AR100), they were eliminated from subsequent analyses. Ribosomal DNA sequences from reference taxa (Table 1) were included in alignments used for subsequent phylogenetic analysis. They were selected to broadly represent the diversity of known Peronosporomycetes and to complement (based on BLAST searches) specific groups of our observed sequences. The sequences were derived from a variety of Peronosporomycetes, Table 1. Reference taxa used in phylogenetic analyses Name a GenBank accession no. Achlya treleaseana AF Aphanomyces cochlioides AF Aphanomyces euteiches AF Aphanomyces laevis AF Aphanomyces stellatus AF Aplanes androgynus AF Aplanopsis spinosa AF Apodachlya brachynema AF Apodachlya minima AF Brevilegnia bispora AF Brevilegnia megasperma AF Calyptralegnia achlyoides AF Dictyuchus monosporus AF Dictyuchus sterilis AF Hyphochytrium catenoides X80345 Isoachlya toruloides AF Leptolegnia caudata AF Leptomitus lacteus AF Pachymetra chaunorhiza AF Peronophythora litchii CBS a Peronospora parasitica AY Phytophthora erythroseptica b CBS a Phytophthora fragariae AF Phytophthora infestans AF Phytophthora megasperma X75631 Plectospira myriandra AF Pythiopsis cymosa AF Pythium aphanidermatum AY Pythium arrhenomanes b AY Pythium capillosum AY Pythium conidiophorum AY Pythium cylindrosporum b AY Pythium graminicola b AY Pythium insidiosum b AY Pythium intermedium b AY Pythium macrosporum b AY Pythium monospermum AY Pythium multisporum b AY Pythium oligandrum b AY Pythium paroecandrum AY Pythium salpingophorum AY Pythium sylvaticum b AY Pythium torulosum b AY Pythium ultimum var. ultimum AY Pythium vanterpoolii AY Saprolegnia anisopora AF Saprolegnia litoralis AF Sclerospora graminicola AY Scoliolegnia asterophora AF Thraustotheca clavata AF a Sequences courtesy of Dr. Andre Levesque, Agriculture and Agri-food Canada, Ottawa. b Type strain.

5 40 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY especially Pythium species for which no GenBank sequences were available. These sequences were kindly provided by Dr. André Levesque (Agriculture and Agri- Food Canada, Ottawa) and have recently been described [38]. Additionally, we included selected GenBank sequences, which were obtained from our BLAST search results. The 28S rdna sequence from Hyphochytrium catenoides (GenBank X80345) was used as an outgroup in this study [68]. After initial alignments in MegAlign, sequence alignments were manually edited using BioEdit [29] to correct misaligned sequences and ambiguous base designations. During this final editing, all sequences were trimmed to a fixed length of 492 bp (gaps included). In initial alignments, all known 28S rdna sequences from Pythium species were included [38]. For subsequent phylogenetic analyses, only those standards with apparent associations with our unknown sequences were included. Phylogenetic analyses were conducted using the neighbor-joining (NJ) method [51] as implemented in TreeCon 1.3b [63, 66]. Branch support was based on 1000 bootstrap replications. Gaps and missing or ambiguous data were ignored. Nucleotide substitution rates for each alignment were calibrated as described by Van de Peer and De Wachter [64, 65] using TreeCon. Comparative Analysis of Peronosporomycete Communities. We used phylogenetic methods described by Martin [42] for comparing directly extracted and baited communities in the three different plant rhizospheres. Various measures of diversity were calculated for each Peronosporomycete community. Sequence diversity, nucleotide diversity, and were calculated using the program Arlequin [54]. For the purpose of these analyses, operational taxonomic units (OTUs) were defined as any set of sequences that differed by 1% or less. The degree of differentiation among communities was estimated by calculating the F st statistic as follows: F st =( T W )/ T, where T is the genetic diversity for all of the communities combined (e.g., all combined plant species communities or all combined soil and bait communities) and W is the genetic diversity within each of the individual communities averaged over all of the communities [42]. To calculate this statistic, aligned sequences without reference taxa were directly imported into Arlequin The statistical significance of F st was estimated from 1000 permutations at a significance level of The parsimony test, for which the theory has been described by Maddison and Slatkin [40], was carried out by generating sets of 100 phylogenetic trees on which plant species or extraction source was optimized as a discrete character using parsimony with the aid of Mesquite 1.02 [39]. The covariation between phylogeny and plant species or extraction source was then determined as the number of character changes between the two extraction sources or between two different plant species that could explain the observed distribution of Peronosporomycetes. The significance of the observed covariation was established by determining the expected number of changes under the null hypothesis that the communities from which sequences were sampled do not covary with phylogeny. The null expectation can be estimated by assuming that the community identity of individual sequences remains fixed, and that the relationships among sequences are randomized [40]. If the observed number of transitions from one community to another is less than 95% of the values generated from randomized data, then microbial composition differs significantly between the two communities [42]. The following direct DNA extraction comparisons were analyzed: tomato vs squash communities, squash vs sorghum communities, and sorghum vs tomato communities. Additionally, comparisons were made between baited communities and direct DNA extracted communities from each of the plant species comparisons and from datasets where sequences from all plant species were combined. In all cases, the null expectation was based on 100 random trees. Results Our analyses using the Oom1 primer set were based on two replicate DNA extractions, whereas analyses from the Sap2 primer set were based on only one extraction. We analyzed the two replicate sets of Oom1 sequences to determine whether they were sampled different portions of the rhizosphere communities. F st analysis of the combined soil and bait datasets indicated that no differences in sequence composition existed between the two replicate extractions and amplifications (F st = 0.007, P = 0.13). Therefore sequences from both extractions were combined for all subsequent analyses. Phylogenetic Analyses of Oom1-derived Soil Communities. Our phylogenetic analyses of Oom1-amplified soil communities revealed the presence of a number of well-supported clusters, many of which were closely related to 28S rdna sequences of known Peronosporomycete species. Sequences grouped largely into two main clusters of Pythium species (Pythium cluster I and Pythium cluster II; Fig. 1). However, some sequences, exclusively from tomato rhizospheres, grouped with a Phytophthora/Peronospora cluster, whereas only one sequence from squash grouped with the Saprolegniomycetidae (see arrow). Pythium Cluster I. A number of taxa largely from tomato rhizospheres clustered with Pythium torulosum

6 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY 41

7 42 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY (Fig. 1; Pythium cluster IA). Another major group of sequences exclusively from sorghum rhizospheres represented closely related strains of P. arrhenomanes or P. aristosporum (only P. arrhenomanes shownintree;fig.1; Pythium cluster IC). Other groups of sequences in Pythium cluster I could not be grouped with any 28S rdna sequence from known Pythium species. Sequences in Pythium cluster IB were exclusively from squash rhizospheres, whereas those from Pythium clusters ID and IE were from mixed plant rhizospheres and sorghum rhizospheres, respectively. Despite the close affinities of taxa in Pythium cluster ID to P. aphanidermatum and P. monospermum and taxa in Pythium cluster IE to P. insidiosum, they all likely represent distinct species of Pythium that are either novel or are known species for which no 28S rdna sequences are available. Phytophthora/Peronospora Cluster. Some sequences from tomato rhizospheres grouped with species of Phytophthora and Peronospora (Fig. 1). One sequence was essentially identical to Phytophthora infestans, whereas four other tomato sequences were similar to that of Peronospora parasitica and likely represent other Peronospora species. Pythium Cluster II. All sequences in this cluster represented unknown Pythium species (Fig. 1; Pythium cluster IF). Although they grouped loosely with P. intermedium, P. macrosporum, P. sylvaticum, P. paroecandrum, and P. cylindrosporum, they likely represent distinct species of Pythium that are either novel or are known species for which no 28S rdna sequences are available. Saprolegniomycetidae. Only one sequence from squash rhizospheres clustered with taxa in the Saprolegniomycetidae (Fig. 1; see arrow). This sequence was most closely related to Apodachlya brachynema but likely represents a distinct genus within this group. Phylogenetic Analyses of Oom1-Amplified Bait Communities. Similar to soil communities, neighborjoining trees of Oom1-amplified bait communities also revealed a number of unique clusters, many of which were closely related to 28S rdna sequences of known Peronosporomycete species (Fig. 2). As with soil communities, sequences from baits fell into the same two major Pythium clusters (clusters I and II). Pythium Cluster I. This cluster corresponds to the same Pythium cluster I found among the soil communities. As with soil communities, bait-derived sequences in this cluster were dominated by a large group of nearly identical sequences that came from the rhizospheres of all three plant species and were very closely related to P. torulosum (Fig. 2; Pythium cluster IA). Another group of sequences exclusively from tomato rhizospheres (Pythium cluster IC) was nearly identical to P. monospermum, whereas most other sequences within this cluster (e.g., Pythium cluster IB and other individual sequences) could not be assigned to any known taxa. Phytophthora/Peronospora and Saprolegniomycetidae Clusters. One tomato sequence grouped in the Phytophthora/Peronospora cluster and nearly identical to Phytophthora infestans (Fig. 2). No Peronospora or Saprolegniomycetidae sequences were found on baits from rhizospheres. Pythium Cluster II. Sequences from this cluster fell largely into two groups, neither of which was closely affiliated with any known Pythium species. Pythium cluster IID represented a group of taxa from tomato rhizospheres (Fig. 2), whereas those in Pythium cluster E consisted exclusively of sequences from sorghum rhizospheres with close relationships to P. intermedium, P. paroecandrum, P. cylindrosporum, and P. sylvaticum. Phylogenetic Analyses of Sap2-Amplified Communities. Analysis of Sap2-derived communities revealed a general lack of sequence diversity among all rhizosphere samples. The majority of sequences fell into two major clusters, regardless of whether they originated from baits or directly from soils (Fig. 3). Many of the squash and sorghum sequences amplified from baits or directly from soils clustered with species of Aphanomyces, Plectospira, and Pachymetra (Aphanomyces cluster I), with the greatest similarity (and BLAST hits) to Aphanomyces cochlioides. However, some squash and all tomato bait sequences clustered with members of the Peronosporomycetidae most closely aligned with Pythium ultimum var. ultimum (Fig. 3; Pythium cluster III). Comparison of Bait and Soil Peronosporomycete Community Diversity. In all community comparisons, the number of OTUs shared between the two com- Figure 1. Neighbor-joining analysis of 86 partial 28S rdna sequences obtained from Oom1 amplifications of DNA extracted directly from rhizosphere soils (soil communities). Ribosomal DNA sequences from reference taxa were included to anchor sequences of unknown affiliation, where possible, to known Peronosporomycete species. Genetic distances were calculated using calibrated substitution rates (0.34) as described previously [64, 65]. Trees were rooted with the hyphochytrid Hyphochytrium catenoides (X80345). Sequence designations beginning with 1 are from the first extraction; those beginning with 2 are from the second extraction. Numbers at each node represent bootstrap values based on 1000 replications.

8 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY 43

9 44 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY munities was low, generally less than 5 10%. Sequence diversity was high and significantly greater in soil communities than in bait communities. For example, significantly greater sequence diversity was observed in soil communities of all three combined plant species and from sorghum and squash soil communities than from the respective bait communities (Table 2). This was true for both primer sets. Nucleotide diversity was low among all Oom1- derived communities but much higher among Sap2- derived communities. Significantly greater nucleotide diversity was observed among Oom1 sequences from squash soil communities than from squash bait communities. However, these differences were not observed using the Sap2 primer set. With the Sap2 primer set, significantly greater nucleotide diversity was found in the sorghum and tomato soil communities than in the respective bait communities. The mean sequence divergence, (a measure of community diversity), was generally lower among Oom1- derived communities than with Sap2-derived communities. Diversity of Oom1-derived bait and soil communities from sorghum and tomato rhizospheres did not differ. However, significantly greater diversity was found in squash soil communities than in squash bait communities. Similarly, significantly greater diversity was found in Sap2-derived tomato and sorghum soil communities than in the respective bait communities. F st statistics for pairwise comparisons of bait and soil communities in all plant rhizospheres were highly significant (P G 0.001), regardless of the primer set used. Furthermore, the F st statistics of combined sequences from rhizospheres of all plant species were also significant (P G 0.001), indicating that bait communities were significantly different from soil communities. Parsimony tests of differentiation between soil and bait communities were all highly significant (P G 0.001), regardless of the plant species or primer set tested. Each of the pairwise comparisons among plant species was also highly significant (P G 0.03; data not shown). Comparison of Peronosporomycete Community Diversity in the Rhizosphere of Different Plant Species Analyses of all pairwise comparisons among the three plant species are shown in Table 3. As with the soil vs bait comparisons, the number of OTUs shared between the rhizosphere communities of each of the paired plant species was low and was in the range of less than 5 10%. Sequence diversity was similar among all pairwise comparisons using the Oom1 primer set. However, the Sap2 primer set yielded lower levels of sequence diversity. No differences in nucleotide diversity were observed among soil communities using the Oom1 primers, regardless of the plant species. However, among bait communities, nucleotide diversity was greatest in tomato rhizospheres, followed by sorghum then squash rhizospheres. With Sap2 primers, greater nucleotide diversity was observed in bait communities from tomato rhizospheres than in those from sorghum or squash rhizospheres. Much higher levels of nucleotide diversity were found among soil communities than among bait communities using the Sap2 primers. However, no differences were observed among the three plant species. Overall sequence divergence was generally greater from soil communities than from bait communities (Table 3). Differential influences of the plant/cropping system on Peronosporomycete communities were observed for both bait and soil communities, regardless of the primer set used. For example, Oom1-derived bait communities from tomato and sorghum rhizospheres were significantly more diverse than those from squash rhizospheres. However, with the Sap2 primer set, bait communities from tomato rhizospheres were more diverse than those from either sorghum or squash. Among soil communities, no differences in overall sequence divergence were apparent among the different plants/cropping systems, regardless of the primer set used. However, Sap2-derived soil communities from tomato rhizospheres were significantly more diverse than those from squash but not sorghum rhizospheres. F st statistics varied considerably depending on the specific plant comparison and on the bait vs soil community comparison. F st values calculated from comparisons of tomato and sorghum communities showed significant community divergence when determined with Oom1- derived soil communities and Sap2-derived bait and soil communities but not with Oom1-derived bait communities. F st values calculated from comparisons of tomato and squash communities showed significant community divergence when determined with Oom1-derived soil and bait communities and Sap2-derived soil communities but not with Sap2-derived bait communities. F st values calculated from comparisons of sorghum and squash rhizosphere communities indicated significant community divergence only with Oom1- and Sap2-derived bait communities and not with soil communities. Figure 2. Neighbor-joining analysis of 89 partial 28S rdna sequences obtained from Oom1 amplifications of DNA extracted from hempseed baits incubated in rhizosphere soils (bait communities). Ribosomal DNA sequences from reference taxa were included to anchor sequences of unknown affiliation, where possible, to known Peronosporomycete species. Genetic distances were calculated using calibrated substitution rates (0.36) as described previously [64, 65]. Trees were rooted with the hyphochytrid H. catenoides (X80345). Sequence designations beginning with 1 are from the first extraction; those beginning with 2 are from the second extraction.

10 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY 45

11 46 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY Table 2. Comparative diversity estimates of directly-extracted and baited Peronosporomycete communities from the rhizospheres of different plant species Primer set Community pairwise Usable comparison a OTUs b loci c Shared OTUs d Sequence diversity (TSD) e Nucleotide diversity (TSD) f Mean sequence divergence g (TSD) Oom1 Combined baits (91) (0.013) (0.061) (29.54) (P G 0.001) Combined soils (86) (0.006) (0.061) (29.18) Oom1 Sorghum baits (33) (0.043) (0.060) (28.65) (P G 0.001) Sorghum soils (28) (0.015) (0.057) (26.42) Oom1 Tomato baits (32) (0.015) (0.054) (25.29) (P G 0.001) Tomato soils (27) (0.033) (0.065) (30.83) Oom1 Squash baits (31) (0.051) (0.007) 5.92 (3.23) (P G 0.001) Squash soils (26) (0.027) (0.061) (28.43) Sap2 Combined baits (55) (0.019) (0.167) (77.45) (P G 0.001) Combined soils (55) (0.009) (0.236) (120.61) Sap2 Sorghum baits (17) (0.072) (0.046) (23.08) (P G 0.001) Sorghum soils (19) (0.022) (0.215) (109.85) Sap2 Tomato baits (20) (0.035) (0.144) (74.10) (P G 0.001) Tomato soils (14) (0.031) (0.315) (160.45) Sap2 Squash baits (19) (0.071) (0.175) (90.30) (P G 0.001) Squash soils (20) (0.033) (0.166) (83.08) a Numbers in parentheses represent the total number of sequences analyzed. b OTU = operational taxonomic unit defined as sequences with Q99% nucleotide similarity. c All Oom1 sequences were truncated to 492 nucleotides (gaps included); Sap2 sequences were trimmed to 517 nucleotides (gaps included); number of nucleotides used in the analyses to calculate the various statistics. d OTUs shared between the two samples. e The probability that two randomly chosen sequences are different. f The probability that two randomly chosen homologous nucleotides are different; a measure of diversity within the usable loci. g Total genetic variation in the sample. h F st =( T W )/ T, where T is the genetic diversity of the combined community samples and W is the genetic diversity within each community averaged over the two communities being compared. F st is a measure of the level of differentiation between each community and the total combined community. F st h Discussion The major hypothesis underlying our work was that terrestrial Peronosporomycete communities described using a molecular phylogenetic approach based on direct DNA extraction and amplification of 28S rrna genes would be different and more diverse than those using conventional baiting strategies. We reasoned that, in addition to the vegetative life stages [zoospores, sporangia, hyphal swellings (or gemmae)] that are typically present and able to colonize baits, we should also detect populations of Peronosporomycetes that may exist only as oospores or other populations that might not be isolated using baiting procedures. In our work, 28S rrna genes were amplified directly from baits as well as from soil to avoid problems associated with the morphological identification of isolates from baits and also to facilitate comparative analyses. Although our sample sizes were small (about sequences per sample), our data clearly show that communities obtained from hempseed baits exhibit a lower level of diversity than those obtained from DNA extracted directly from rhizosphere soil. This conclusion is based on several lines of evidence. First, our phylogenetic analysis of Oom1-derived communities revealed the presence of several clusters found among soil communities that were absent among bait communities. For example, a cluster of sorghum rhizosphere sequences nearly identical to Pythium arrhenomanes was found among soil communities (Fig. 2, cluster IC) but not in bait communities from the same rhizosphere soil (see also Fig. 2, clusters IB and II). Second, sequence diversity was much greater in soil communities than in bait communities in all rhizospheres except tomato, regardless of the primer set used. This was also reflected in the F st statistics, which indicated that soil communities, regardless of the plant species, were significantly different from the respective bait communities. Furthermore, relatively few OTUs were present in both bait and soil communities, indicating that each method samples the underlying rhizosphere community differently. Although the source of the increased diversity in soil communities is unknown, we hypothesize that it might be because of oospore populations of Peronosporomy- Figure 3. Neighbor-joining analysis of 65 partial 28S rdna sequences obtained from Sap2 amplifications of DNA extracted directly from rhizosphere soils and from hempseed baits incubated in rhizosphere soils. Ribosomal DNA sequences from reference taxa were included to anchor sequences of unknown affiliation, where possible, to known Peronosporomycete species. Genetic distances were calculated using calibrated substitution rates (0.28) as described previously [64, 65]. Trees were rooted with the hyphochytrid H. catenoides (X80345).

12 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY 47 Table 3. Diversity estimates of Peronosporomycete communities from the rhizospheres of different plant species based on all pairwise comparisons of each plant species Primer set Community pairwise Usable comparison a OTUs b loci c Shared OTUs d Sequence diversity (TSD) e Nucleotide diversity (TSD) f Mean sequence divergence g (TSD) Bait communities Oom1 Tomato (33) (0.008) (0.055) (25.97) (P = 0.068) Sorghum (25) (0.011) (0.027) (12.73) Oom1 Squash (31) (0.008) (0.007) 5.60 (3.08) (P G 0.001) Tomato (33) (0.008) (0.057) (26.42) Oom1 Sorghum (25) (0.011) (0.022) (12.12) (P = 0.058) Squash (31) (0.008) (0.007) 5.60 (3.08) Sap2 Tomato (14) (0.034) (0.011) (5.79) (P G 0.001) Sorghum (19) (0.022) (0.002) 1.48 (1.06) Sap2 Squash (9) (0.052) (0.006) 4.61 (2.83) (P = 0.911) Tomato (12) (0.034) (0.011) (5.79) Sap2 Squash (9) (0.052) (0.006) 4.61 (2.83) (P G 0.001) Sorghum (16) (0.022) (0.002) 1.48 (1.06) Soil communities Oom1 Tomato (27) (0.008) (0.064) (30.41) (P = 0.014) Sorghum (33) (0.008) (0.055) (25.91) Oom1 Squash (26) (0.008) (0.063) (29.78) (P = 0.052) Tomato (27) (0.008) (0.065) (30.91) Oom1 Sorghum (33) (0.008) (0.055) (25.95) (P = 0.358) Squash (26) (0.008) (0.060) (28.43) Sap2 Tomato (14) (0.031) (0.315) (160.45) (P G 0.002) Sorghum (19) (0.022) (0.215) (109.85) Sap2 Squash (22) (0.028) (0.182) (91.16) (P G 0.006) Tomato (14) (0.031) (0.315) (160.45) Sap2 Sorghum (19) (0.022) (0.215) ( (P = 0.634) Squash (22) (0.028) (0.182) (91.16) a Numbers in parentheses represent the total number of sequences analyzed. b OTU = operational taxonomic unit defined as sequences with Q99% nucleotide similarity. c All sequences were truncated to 492 nucleotides (gaps included); number of nucleotides used in the analyses to calculate the various statistics. d OTUs shared between the two samples. e The probability that two randomly chosen sequences are different. f The probability that two randomly chosen homologous nucleotides are different; a measure of diversity within the usable loci. g Total genetic variation in the sample. h F st =( T W )/ T, where T is the genetic diversity of the combined community samples and W is the genetic diversity within each community averaged over the two communities being compared. F st is a measure of the level of differentiation between each community and the total combined community. F st h cetes that have gone undetected using baiting procedures. Based on our preliminary laboratory studies (unpublished), we have shown that our methods can effectively lyse oospores of a variety of Peronosporomycete species, extract DNA, and amplify our target sequences. We therefore expected that we were effective in extracting oospores present in our test soils. However, it is possible that other Peronosporomycete populations may be less responsive to baits or may be less competitive under the bait incubation conditions. Clearly, the choice of baiting material can influence the types of Peronosporomycete species that can be detected [14, 52]. Therefore, it is possible that the choice of different baits would yield different results of Peronosporomycete diversity. Results of this study demonstrate the feasibility of using molecular approaches to study Peronosporomycete communities and to uncover hidden levels of biodiversity. Our data suggest that molecular characterization is detecting a greater level of diversity than previously known for this group of organisms. For example, we observed several unique clusters of Pythium-like sequences that could not be associated with any known Pythium species for which 28S rdna sequences are available. These unique groups were found not only from DNA extracted directly from soil but also from DNA extracted from baits. Traditionally, species have been identified from baits based on the reproductive structures formed under the conditions of the incubation or after isolation in culture [26]. It is likely, therefore, that other species are present on colonized baits, which are not detected because they lack diagnostic structures. Furthermore, given that some clusters were found among bait communities that were not detected in soil communities (e.g., Fig. 3, cluster II sequences nearly identical to Pythium ultimum var. ultimum), the use of DNA extractions both from baits and from soils will likely provide a more complete picture of the diversity of Peronosporomycete communities in soil and aquatic habitats.

13 48 J.M. ARCATE ET AL.: PERONOSPOROMYCETE COMMUNITY DIVERSITY A prominent observation from our data was the significant effect of the plant/cropping system on rhizosphere Peronosporomycete communities. For example, a number of individual clusters within the NJ trees of either bait or soil communities were composed of sequences from only one plant/cropping system. This is particularly true for groups of sequences from tomato and sorghum rhizospheres. The significant impact of plant/cropping system in influencing the Peronosporomycete community is further indicated by the parsimony test, which points to a significant (P G 0.001) covariation of Peronosporomycete lineage with plant species. Second, significant F st statistics for comparisons between squash or sorghum and tomato indicate that these communities include significantly different sets of taxa. Although there are subtle differences in nucleotide diversity and overall sequence divergence between squash and sorghum communities, we are not confident in concluding that these communities differ. There are at least two explanations for these observations: either plants actively select for specific Peronosporomycete communities in their rhizospheres, or, because of the different cropping histories, each of the three fields sampled contained significantly different Peronosporomycete communities. Although each of the three plant species sampled was grown adjacent to each other in soils with identical soil properties and presumably homogeneous Peronosporomycete communities, the different cropping histories unique to each field prevent us from being able to reject the possibility that Peronosporomycete communities differed in these soils, making it unclear whether plants per se select for specific Peronosporomycete communities. Experiments are underway to test this hypothesis more directly. Despite our abilities to clearly discern differences between bait and soil communities and between rhizosphere communities associated with different plant/ cropping systems, our results are still likely to underestimate the diversity of Peronosporomycete communities in the rhizosphere. Our sampling intensity was considerably limited, with relatively few clones sequenced per sample. A greater sampling of our clone libraries will undoubtedly reveal additional community diversity beyond what we detected in our current study. Peronosporomycetes show distinct seasonal fluctuations in species abundance [1, 19, 21, 34], and some species may be highly aggregated [13, 14]. These observations indicate the need for more intensive sampling in time and space to capture the full range of diversity in a particular habitat. One of the striking characteristics of rhizosphere Peronosporomycete communities revealed in this study was the dominance by Pythium species. This is not surprising because nearly all surveys of Peronosporomycetes have revealed an abundance of Pythium species over a wide range of habitats [2, 24, 25, 30, 34, 35]. Sequences obtained from rhizospheres of the three plant species were closely related to P. ultimum, P. torulosum, P. arrhenomanes/p. aristosporum, P. monospermum, P. intermedium, P. sylvaticum, P. paroecandrum, and P. cylindrosporum. Those sequences related to P. arrhenomanes/p. aristosporum all came from sorghum rhizospheres. This would be predicted because these species are common pathogens of grasses and cereals [69]. Similarly, sequences closely related to P. torulosum came largely from the rhizosphere of tomato, which is a common host for P. torulosum [69]. Strikingly absent from our rhizosphere Peronosporomycete communities was a greater diversity of sequences from the Saprolegniomycetidae. Although sequences related to Aphanomyces species were detected largely from sorghum and squash rhizospheres, and one sequence related to Apodachlya was detected, we anticipated a higher frequency of members of this group because they have been described in association with seeds [74, 75], and species of related genera are known plant pathogens [20, 23, 28, 33, 72, 76]. The absence of these species in our current study may indicate either that (1) they indeed are not present in the rhizospheres of these plant species, (2) their absence is related to either low seasonal abundance discussed above or defined microhabitats that were not sampled in the course of this study, or (3) our primers were not effective in detecting these species, despite the ability of these primer sets to detect these species in laboratory controls. However, current studies with Agrostis stolonifera Peronosporomycete communities using these same two primer sets indicate an abundance of taxa from the Saprolegniomycetidae (Nelson, unpublished data). This suggests that our primer sets adequately detect these species if they are present. The results of our study have shown the utility of culture-independent approaches using 28S rrna genes to assess diversity of Peronosporomycete communities in association with plants. This approach has revealed the presence of as yet unexplored diversity of Peronosporomycetes in plant rhizospheres. Further work in this area will provide a better understanding of the distribution and functional roles of Peronosporomycetes in the environment as well as their pathogenic interactions in soils and water bodies. Acknowledgments Special thanks are due to Dr. André Levesque for providing 28S rdna sequences of a wide range of Peronosporomycete species, especially those of Pythium spp., many of which represented the type or neotype strain. We also thank Kathie Hodge for her helpful discussions and review of the manuscript and Michael Milgroom, Fernando Ponce, Angelika Rumberger, and Sofia Wind-