Development of microsatellite markers from the transcriptome of Erysiphe necator for analysing population structure in North America and Europe

Authors

  • O. Frenkel,

    1. Department of Plant Pathology & Plant-Microbe Biology, Cornell University, Ithaca, NY 14853, USA
    Search for more papers by this author
    • Present address: Department of Plant Pathology, Agricultural Research Organization, The Volcani Center, POB 6, Bet-Dagan 50250, Israel.

  • I. Portillo,

    1. Dipartimento di Protezione e Valorizzazione Agro-Alimentare, Università degli Studi di Bologna, Viale Fanin 46, 40100 Bologna, Italy
    Search for more papers by this author
  • M. T. Brewer,

    1. Department of Plant Pathology & Plant-Microbe Biology, Cornell University, Ithaca, NY 14853, USA
    Search for more papers by this author
    • Present address: Department of Plant Pathology, University of Georgia, Athens, GA 30602, USA.

  • J. P. Péros,

    1. Institut National de la Recherche Agronomique, UMR 1334, Equipe DAVEM, 2 place Viala, 34060 Montpellier, France
    Search for more papers by this author
  • L. Cadle-Davidson,

    1. United States Department of Agriculture–Agricultural Research Service, Grape Genetics Research Unit, Geneva, NY 14456, USA
    Search for more papers by this author
  • M. G. Milgroom

    Corresponding author
    1. Department of Plant Pathology & Plant-Microbe Biology, Cornell University, Ithaca, NY 14853, USA
    Search for more papers by this author

E-mail: mgm5@cornell.edu

Abstract

Transcriptome sequences of the grape powdery mildew fungus Erysiphe necator were used to develop microsatellite markers (EST-SSRs) to study its relatively unexplored population structure in its centre of diversity in eastern North America. Screening the transcriptome sequences revealed 116 contigs with candidate microsatellites, from which 11 polymorphic microsatellite markers were developed from 31 markers tested. Eight of these markers were used to genotype isolates from different regions and hosts in the eastern USA and compare them to samples from southern France and Italy. Genetic diversity in the eastern USA is much greater than in Europe. Bayesian cluster analyses showed that 10 isolates from North America have high affinities with, but differ from, European group A; these are referred to as A-like isolates. No isolates with close affinity to European group B were found in the eastern USA. Bayesian analyses also detected genetic differentiation between isolates from Vitis rotundifolia and isolates from other Vitis hosts. Genetic differentiation detected between the northeastern and southeastern USA was mostly attributable to the A-like isolates in the southeast, which are significantly more aggressive than the other populations. This research demonstrates that transcriptome sequencing of fungal pathogens is useful for developing genetic markers in protein-coding regions and highlights the role of these markers in population biology studies of E. necator.

Introduction

Studies of population genetics rely on genetic markers that reveal an appropriate amount of polymorphism for the questions being addressed. For example, when studying the species/population interface, sequencing multiple genes that are conserved between species has provided insights into the evolution of plant pathogens (Couch et al., 2005). To make inferences about population structure on a finer scale within a single species, it is often necessary to use markers that are more polymorphic, e.g. multilocus DNA fingerprinting or AFLPs (Collado-Romero et al., 2008).

A recent study of the grape powdery mildew fungus, Erysiphe necator (syn. Uncinula necator) showed, using multilocus sequencing, that eastern North America is more genetically diverse than populations in Europe and Australia, and is the likely source for introductions to these latter areas (Brewer & Milgroom, 2010). These results are consistent with historical records of grape powdery mildew and the fact that native grapevine species in North America are more resistant to it than the European grapevine, Vitis vinifera (Olmo, 1971; Cadle-Davidson et al., 2011), probably because they coevolved with E. necator. Moreover, along with variation in powdery mildew resistance among wild species, Frenkel et al. (2010) detected significant variation in pathogenicity and aggressiveness of E. necator isolates from wild species and different regions in North America. In particular, only isolates sampled from muscadine grapevines, Vitis rotundifolia (syn. Muscadinia rotundifolia), in the southeastern USA could establish colonies on an accession of V. rotundifolia.

Despite the high diversity and putative origin of E. necator in North America, until recently almost all population genetic studies of this species were conducted on populations in Europe and Australia, where two distinct genetic groups, A and B, have been found. Groups A and B are strongly differentiated genetically, with relatively little variability within each group (Délye et al., 1997; Miazzi et al., 2003; Péros et al., 2005). The occurrence of groups A and B in North America is an open question. Some isolates from the southeastern USA have the same multilocus sequence haplotype as group A isolates from Europe and Australia; however, isolates with haplotypes like group B were only found on the west coast of the USA, where they were the only haplotypes present (Brewer & Milgroom, 2010). Comparisons of genotypes based on multilocus sequencing may result in limited resolution because of the small number of polymorphic sites detected in the three genes sequenced. More variable markers are needed to address this and other questions about the population structure in North America with finer resolution.

Although some studies of population genetics of E. necator have used a small number of well-defined, species-specific markers (SCARs, SSRs and SNPs) (Amrani & Corio-Costet, 2006; Péros et al., 2006; Miazzi et al., 2008; Montarry et al., 2008; Brewer & Milgroom, 2010), most studies have been done with dominant and undefined markers (e.g. RAPDs, ISSRs and AFLPs), such that PCR products of the same size are assumed to be alleles that are identical by descent (Brown, 1996; Schlötterer, 2004). Undefined markers have the additional disadvantage that DNA from contaminants in powdery mildew colonies, e.g. mycoparasites, may go undetected and spuriously contribute to the genotype (unpublished data). During the last decade, microsatellites, or simple-sequence repeats (SSRs), and single-nucleotide polymorphisms (SNPs) have become the markers of choice for population genetic studies (Dutech et al., 2007). Microsatellites have the advantage over DNA fingerprinting with RFLPs, RAPDs or AFLPs because they identify unique loci and are codominant, highly reproducible and have high levels of polymorphism (Santana et al., 2009). However, efforts to develop microsatellites in E. necator so far have produced few useful markers (Péros et al., 2006; Dutech et al., 2007).

Until recently, the development of microsatellite markers involved the construction of plasmid libraries enriched for specific repeat motifs (Dutech et al., 2007; Santana et al., 2009). This method is difficult, time-consuming, expensive and has proven to be especially challenging for finding polymorphic markers in fungi relative to other eukaryotes (Dutech et al., 2007). An alternative method, which has proven successful for several fungi (Breuillin et al., 2006; Zhong et al., 2009), involves searching for repeat motifs in nucleotide sequences from EST libraries, transcriptomes or whole genomes. Recently, the availability and decreased costs of massively parallel sequencing, such as pyrosequencing, have exponentially increased in silico databases of nucleotide sequences (Santana et al., 2009) from which microsatellites can be mined. Although microsatellite mining is not usually the first priority of such sequencing projects, whole genome or transcriptome sequences can be used for this purpose (Cheung et al., 2008; Parchman et al., 2010).

Sequencing the transcriptome provides a valuable starting point for characterizing functional genetic variation in non-model organisms, especially where whole-genome sequencing efforts are too costly or time-consuming (Parchman et al., 2010). In the Erysiphales, large numbers of repetitive elements throughout the genome present a serious challenge for whole-genome sequencing and assembly (Spanu et al., 2010). Therefore, transcriptome sequencing is a reasonable compromise for initiating studies of the genomes of powdery mildew fungi. The transcriptome of one North American isolate of E. necator was recently sequenced (L. Cadle-Davidson & M.G. Milgroom, unpublished data). The objectives of this study were to: (i) develop polymorphic microsatellite markers from transcriptome sequences; (ii) investigate the genetic structure of E. necator from different regions and species of Vitis in North America; (iii) test for genetic differentiation between populations sampled from the northeastern and southeastern USA, between cultivated and wild hosts, and between isolates from V. rotundifolia and other host species; (iv) revisit the question of whether groups A and B are present in the eastern USA; and (v) evaluate microsatellite markers for distinguishing genetic groups A and B in Europe.

Materials and methods

Normalized cDNA library and transcriptome sequencing

Erysiphe necator isolate G14 was collected in Geneva, NY, USA, from the commercial grape hybrid cv. Rosette in September 2007. Conidia and mycelia of isolate G14 were harvested from colonies on grape leaves 17 days after inoculation as described previously (Wang et al., 1995; Cadle-Davidson et al., 2010). RNA was extracted with the RNeasy Plant Mini kit (QIAGEN) and provided to Bio S&T for preparation of an uncloned normalized cDNA library. Normalization was done to equalize sequencing of abundant and less abundant transcripts. One half-run of 454-FLX sequencing (454 Life Sciences) was done at the Cornell University Life Sciences Core Laboratory Center. mira software (Chevreux et al., 2004) was used to assemble the reads into contigs de novo.

Data mining for microsatellites

Approximately 32 000 contigs from the E. necator transcriptome (data not shown) were searched for microsatellite motifs using the software tandem repeat finder (Benson, 1999). Mono- and dinucleotide repeats were eliminated because of the difficulty of scoring alleles based on differences of one or two nucleotides during fragment analysis. Trinucleotide motifs 18 bp or longer and tetra- and pentanucleotide motifs 20 bp or longer were screened for. To ensure independence of markers, sequences containing microsatellites were compared using multiple alignments of the nucleotide sequences with the software ClustalX (Larkin et al., 2007) to identify and exclude sequences with high similarity. Primers flanking repeats were designed using the web-based software Primer3 (Rozen & Skaletsky, 2000). The general primer-picking conditions were set for an optimal primer of approximately 20 bases. Primer pairs were developed to produce amplicons with expected lengths of 130–350 bp.

PCR and fragment analysis of microsatellite markers

To facilitate fluorescent detection of fragments and to increase post-PCR multiplexing flexibility, an M13-specific sequence (5′-CACGACGTTGTAAAACGAC-3′) was added to the 5′-end of each forward primer as described by Schuelke (2000). Along with locus-specific primers (Table 1), PCR was performed with complementary M13-specific primers, 5′-labelled with a fluorescent dye (FAM, VIC, or NED; Applied Biosystems) for detection of fragments (see below). PCR for all primer pairs was carried out in a total volume of 12·5 μL. Reactions included 1·25 μL of 10× PCR buffer (Takara Bio, Inc.), 1·25 μL of 2·5 mm dNTPs, 0·2 μL of 10 μm forward primers, 0·4 μL of 10 μm reverse primer, 0·5 μL of 10 μm 5′-dye-labelled M13 primer (Applied Biosystems), 0·375 U ExTaq (Takara Bio, Inc.) and 1 μL DNA template (approximately 5 ng μL−1). Reactions were conducted on a PTC-100 thermocycler (Bio-Rad). Cycling conditions included an initial denaturation at 95°C for 2 min, followed by 35 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 30 s and extension at 72°C for 30 s, followed by a final extension at 72°C for 5 min.

Table 1.   Diversity of Erysiphe necator assessed with eight microsatellite markers (EnMS2–EnMS9)
PopulationNNo. polymorphic lociMean no. alleles per locusGene diversityaNo. private allelesGenotypic richnessGenotypic diversitybIA (P-value)b
  1. aUnbiased gene diversity was estimated as [N/(N−1)](1−Σpi2), where pi is the frequency of the ith allele and N is the sample size; estimated using GenAlEx v6·4 (Peakall & Smouse, 2006).

  2. bGenotypic diversity and index of association (IA) were estimated using Multilocus 1·2 (Agapow & Burt, 2001). Genotypic diversity was estimated as [N/(N−1)](1−Σpi2), where pi is the frequency of the ith haplotype. IA was estimated in the southeastern USA population excluding the A-like isolates (N = 24).

  3. cSampled from wild and cultivated Vitis species, except for V. rotundifolia (Table S1).

  4. dIA not estimated in populations with low genotypic diversity and/or small sample size.

  5. eGroup A and group B were sampled mostly from southern France (= 29 and 24, respectively) by Péros et al. (2005). Three additional isolates in each group were sampled from south and central Italy (Table S2).

  6. fDiversity estimates in parentheses are based on data from 11 loci (EnMS1–EnMS11). Data for EnMS10 and EnMS11 were only obtained for European isolates.

Northeastern USAc2785·60·737271·00−0·085 (0·797)
Southeastern USAc3386·00·728300·990·109 (0·209)
V. rotundifolia (southeastern US)1021·50·160 50·84d
European group Ae320 (0)f1·0 (1·0)0·00 (0·00)0 (0)1 (1)0 (0)
European group Be271 (3)f1·1 (1·4)0·01 (0·08)2 (4)2 (5)0·07 (0·62)

For fragment analysis, PCR products from three reactions, each with a different fluorescent dye (FAM, VIC or NED), were pooled and mixed gently. One microlitre of the pooled reactions was added to 8·5 μL HIDI formamide (Applied Biosystems), 0·3 μL ddH2O and 0·2 μL GeneScan 500 LIZ size standard (Applied Biosystems) and heated for 5 min at 95°C. Fragment analyses were conducted at the Cornell University Life Sciences Core Laboratories Center using an Applied Biosystems 3730xl DNA Analyzer. Allele sizes were analysed using the GeneMapper software v3·0 (Applied Biosystems). Fragment analyses for all European isolates that showed polymorphism and a subset of the common North American haplotypes were repeated at least once.

Isolate collection and DNA isolation

A panel of 15 isolates was used to test PCR primers and assess polymorphism of candidate microsatellite markers. This panel comprised three isolates from the northeastern USA (Ith5, NY91, NY47); four from the southeastern USA (NCaes1, NCaes6, NClab4, vir1); two from V. rotundifolia (PCF1, PCF2); three from group A in France (AD2, AD18, FA01); and three from group B in France (JMD9, JMD23, PA06) (Tables S1 and S2). After identifying repeatable, polymorphic markers, 129 isolates were genotyped for nine markers; isolates from Europe were also genotyped for two additional markers, as explained below. The 129 isolates included 70 from North America and 59 from Europe (Table 1). North American isolates were collected and described previously in the eastern USA by Frenkel et al. (2010) or Brewer & Milgroom (2010) and chosen for this study to maximize the potential diversity of E. necator with respect to location and host species. Isolates were sampled from cultivated V. vinifera (= 22) and from four other Vitis species: V. rotundifolia (= 10), V. labrusca (= 6), V. riparia (= 15) and V. aestivalis (= 17). Forty-five of these same isolates were previously genotyped by multilocus sequencing by Brewer & Milgroom (2010), including 19 isolates from V. vinifera, five from V. rotundifolia, five from V. labrusca, seven from V. riparia and nine from V. aestivalis (Table S1). The eastern USA samples were divided into two subpopulations on the basis of significant genetic differentiation (Brewer & Milgroom, 2010) and differences in pathogenicity or aggressiveness (Frenkel et al., 2010): northeastern USA (= 27), including isolates from Massachusetts, New Hampshire, New York, New Jersey, Ohio, Pennsylvania and Michigan; and southeastern USA (= 33), including isolates from Virginia, North Carolina, Georgia and Texas. Isolates sampled from V. rotundifolia growing in the southeastern USA, both wild and cultivated, were considered a third population (= 10) because they were genetically differentiated from other North American isolates (Brewer & Milgroom, 2010) and they were the only isolates sampled from diverse Vitis hosts that were pathogenic on V. rotundifolia var. munsoniana (USDA-ARS accession DVIT 2735) (Frenkel et al., 2010), which is in a different subgenus from the other Vitis species sampled (Olmo, 1971). All North American isolates were cultured on leaves of V. vinifera cv. Cabernet Sauvignon as described previously (Brewer & Milgroom, 2010; Frenkel et al., 2010). Mycelium and conidia were collected from powdery mildew colonies for DNA extraction using sticky tape and extracted in 5% Chelex as described by Brewer & Milgroom (2010).

Fifty-nine isolates from Europe (Table S2) were genotyped to test whether the SSR markers were polymorphic within and between genetic groups A and B. Fifty-three of these were previously collected in the Héraut and Ardèche districts of southern France, genotyped for RAPDs and SNPs in two genes (ITS and CYP51) and assigned to groups A and B (Péros et al., 2005); they comprised 29 isolates from group A and 24 from group B chosen to maximize the diversity of RAPD genotypes, grapevine cultivar, sampling date and geographic origin as reported by Péros et al. (2005). In addition, four isolates from southern Italy assigned to group A (isolates Am288 and Am221) or group B (isolates Bm316 and Bm329) using SCAR markers (Miazzi et al., 2008) and two isolates (4909-4 and 4909-51) from central Italy (Reggio Emilia) assigned to groups A and B, respectively, on the basis of SNPs in beta-tubulin and IGS (I. Portillo & M. G. Milgroom, unpublished data) were genotyped.

Statistical analysis

GenAlEx v6·4 (Peakall & Smouse, 2006) was used to estimate allele frequencies, unbiased gene diversities, observed number of alleles and number of private alleles within each of the five populations. Multilocus 1·2 (Agapow & Burt, 2001) was used to calculate the observed number of multilocus haplotypes and genotypic diversity, defined as = [N/(N−1)](1−Σpi2), where pi is the frequency of the ith haplotype and N is the sample size. GenAlEx v6·4 was also used to estimate ΦST to test for population structure in the eastern USA. Differentiation between isolates from the northeastern and southeastern USA, between isolates on cultivated V. vinifera and wild Vitis hosts, and between isolates from V. rotundifolia and other Vitis hosts was tested for.

Jost (2008) raised concerns regarding ΦST for estimating differentiation between populations when markers with high mutation rates such as SSRs are being used. Therefore, a complementary Bayesian model-based clustering method on multilocus genotype data implemented in the program structure v2·2·3 (Pritchard et al., 2000) was also used to assign individuals to populations. An admixture model using independent allele frequencies was adopted, and the analysis was performed using a burn-in period of 5 × 104 and a run length of 5 × 105 Markov chain Monte Carlo (MCMC) repetitions. Log-likelihood values and posterior probabilities were estimated assuming one to seven clusters (= 1, 2 . . . 7). Five independent runs were performed for each analysis to verify the convergence of parameter estimates. The most probable number of clusters was estimated using the method described by Evanno et al. (2005), based on the rate of change in the log likelihood between successive values of k. This analysis assumes that populations are randomly mating and that markers are unlinked, i.e. there is no linkage (or gametic) disequilibrium (Jombart et al., 2009). Because progenies from crosses are not available to test for linkage, gametic disequilibrium was estimated for all loci together using the index of association (IA) and for all pairs of loci for samples from the northeastern and southeastern USA with the program Multilocus 1·2 (Agapow & Burt, 2001). A principal coordinates analysis (PCoA) was used on individual pairwise genetic distances between haplotypes implemented in GenAlEx v6·4. Analysis with PCoA has the advantage that it does not require strong assumptions about the underlying genetic model (Jombart et al., 2009).

Results

Microsatellite marker development

A total of 82 Mbp of transcriptome sequence of E. necator was obtained in approximately 32 000 contigs and used for mining for microsatellites. There were 116 unique contigs containing microsatellites with six or more trinucleotide repeats, five or more tetranucleotide repeats and four or more pentanucleotide repeats. Trinucleotide repeats were the most common (47·4%), followed by tetranucleotide (27·6%) and pentanucleotide repeats (25%). Many of these contigs were not suitable for marker development either because the regions flanking the repeat motifs were too short for designing PCR primers or they contained other repetitive sequences unsuitable for primer design. Therefore, PCR primers were designed for 31 contigs. Nineteen of the 31 primer pairs produced a single PCR product when screened against 15 isolates in the test panel. No PCR products were obtained from the primer pairs designed to the 12 remaining sequences. Among the 19 primer pairs that yielded PCR products, eight were monomorphic among the test panel isolates. Nine of the remaining markers (EnMS1EnMS9) showed reproducible amplification and clear polymorphisms. Two markers (EnMS10 and EnMS11) produced consistent results only for isolates in the test panel from Europe but not from the eastern USA. With USA isolates, single PCR products were obtained, but sometimes not in sufficient quantity to score reliably. Repeat motifs, PCR primers and allele sizes for these 11 markers are shown in Table 2. Sequences of the 31 contigs are available upon request from the authors.

Table 2.   Microsatellite primer sequences, repeat motifs, allele size ranges and numbers of alleles observed in populations of Erysiphe necator from the eastern USA, France and Italy
LocusRepeat motifPrimer sequence (5′→ 3′)Range of allele sizesaNumber of observed alleles
  1. aObserved range of allele sizes in samples of E. necator from eastern USA, France and Italy. Allele sizes include 19 bp of the M13-specific sequence in primers used for PCR and fragment analysis (see Materials and methods).

  2. bMicrosatellite markers EnMS10 and EnMS11 did not yield consistent or reliable results in the initial screening of isolates from the eastern USA, so alleles reported here are only from isolates from France and Italy. These two markers were later found to be reliable for USA isolates depending on the quality and quantity of template DNA (unpublished data).

EnMS1(TCA)10F: TCACGACCTTTCCAAAATCC236–30217
R: TGTCCGTTTTGAACTCCAGA
EnMS2(GTA)3(GTG)2(GAT)6F: TCTGAGTCGGAATTTGGAATG185–2006
R: TATAACAGCCAACCGCCATC
EnMS3(GCT)8F: TGTGTCGATGCCACGTTATT219–2518
R: AAATTGGATCCCCACCTCTC
EnMS4(TGT)9F: ATATCGAGTTGCCTGCATGA284–3209
R: CTCATGAGCTTCGGTTCAAAG
EnMS5(GAA)13F: TCAAGTCAAGGAGATCCTGTCATA184–2079
R: TCCATCTTCGTCTTCCGAGA
EnMS6(GAT)7F: CAAGAGGCGTTCCAGAAAAG257–2786
R: GAACTCATTGACCCACCAGTC
EnMS7(ATG)11F: AGGATGCCAACAAGAGCCTA186–2109
R: TTTGCCCCTCGATTATCAAC
EnMS8(TGA)8F: GCGCAGGAGGTGAAATAAAA189–2056
R: GAGGTGGCGGTTCTGAAGTA
EnMS9(TTAT)7F: TTTACATCCCCCAGCTTACG172–1844
R: CGTGGCTTCAATCATAGGAA
EnMS10b(TAC)21F: TATCTCGGGTTGGACGATGT269, 2722
R: CATCCCCAATCAGTCTACCAA
EnMS 11b(TTA)9F: GCAAATGCTGGGACATCTTC185, 1912
R: GAAATCTTTGCGGCTAATGC

Genetic diversity and population structure

There were four to 19 alleles per locus for the nine markers (EnMS1–EnMS9) that were polymorphic in the eastern USA (Table 2). Seven of these nine loci had six to nine alleles; one had four alleles. EnMS1 had 19 alleles in the USA (Table 3) and was excluded from analyses in the USA because of being hypervariable. EnMS1 was used for some analyses of European populations because only two alleles were found there. For the other eight loci (EnMS2–EnMS9), the mean numbers of alleles per locus in the northeastern and southeastern USA populations were 5·6 ± 0·73 and 6·0 ± 0·60, respectively (Table 1). Two of the eight loci were polymorphic in the small sample from V. rotundifolia (= 10), with three alleles at each locus. Mean unbiased gene diversities for northeastern and southeastern USA were 0·73 ± 0·04 and 0·72 ± 0·03, respectively, compared to 0·16 ± 0·11, 0 and 0·01 ± 0·009 for samples from V. rotundifolia, group A and group B, respectively (Table 1).

Table 3.   Microsatellite allele frequencies and unbiased gene diversity (h) in five populations of Erysiphe necator collected from the eastern USA (northeastern, southeastern USA and Vitis rotundifolia) and southern France and Italy (group A and group B)
LocusAlleles sizes (bp)aNortheastern USA (= 27)Southeastern USA (= 33)V. rotundifolia (southeastern US) (= 10)European group A (= 32)European group B (= 27)
  1. aMicrosatellite alleles are defined by fragment length determined using ABI 3730xl DNA Analyzer. Sizes include 19 bp in the M13 specific sequence in PCR primers (see Materials and methods). Alleles are sorted by frequency in the northeastern USA population.

  2. bUnbiased gene diversity, h, was estimated using GenAlEx v6·4 (Peakall & Smouse, 2006).

  3. cMicrosatellite markers EnMS10 and EnMS11 did not yield consistent or reliable results in the initial screening of isolates from the eastern USA, so were only used to genotype isolates from France and Italy. These two markers were later found to be reliable for USA isolates depending on the quality and quantity of template DNA (unpublished data).

EnMS12510·160·100·100·000·00
2450·120·000·000·000·00
2660·120·100·000·000·00
2390·080·500·001·000·74
2570·080·130·000·000·00
2630·080·030·000·000·00
2690·080·000·200·000·00
2720·080·000·100·000·00
2420·040·030·000·000·00
2540·040·000·000·000·00
2600·040·030·000·000·00
2800·040·000·000·000·00
2990·040·000·200·000·00
2360·000·070·000·000·26
2870·000·000·200·000·00
2900·000·000·100·000·00
3020·000·000·100·000·00
hb0·940·720·930·000·40
EnMS21910·440·421·000·000·00
1880·260·270·001·000·00
1940·260·210·000·000·00
2000·040·000·000·000·00
1850·000·030·000·001·00
1970·000·060·000·000·00
h0·690·72000
EnMS32360·460·110·000·000·00
2330·290·331·000·000·00
2390·130·040·000·001·00
2300·040·070·000·000·00
2420·040·040·000·000·00
2510·040·000·000·000·00
2190·000·040·000·000·00
2220·000·370·001·000·00
h0·7140·758000
EnMS42870·320·421·000·000·00
2930·180·120·000·000·00
2960·180·040·000·000·00
2840·140·120·001·000·04
2990·140·150·000·000·00
2900·050·080·000·000·00
3020·000·040·000·000·00
3050·000·000·000·000·96
3200·000·040·000·000·00
h0·830·79000·07
EnMS51920·240·19000
1860·160·471·001·001·00
1980·160·060·000·000·00
1840·120·000·000·000·00
1890·080·100·000·000·00
1950·080·030·000·000·00
2070·080·030·000·000·00
2010·040·030·000·000·00
2040·040·000·000·000·00
h0·890·730·000·000·00
EnMS62630·420·390·000·001·00
2600·390·551·001·000·00
2660·120·000·000·000·00
2570·040·030·000·000·00
2750·040·000·000·000·00
2780·000·030·000·000·00
h0·680·570·000·000·00
EnMS71890·220·070·000·000·00
2010·220·170·101·000·00
1860·190·000·000·000·00
1950·110·310·000·001·00
1980·110·070·000·000·00
1910·070·070·600·000·00
2040·040·100·300·000·00
2070·040·100·000·000·00
2100·000·100·000·000·00
h0·870·860·600·000·00
EnMS82010·480·400·501·000·00
1980·330·230·300·000·00
1920·190·230·200·000·00
1890·000·030·000·000·00
1950·000·100·000·000·00
2050·000·000·000·001·00
h0·650·750·690·000·00
EnMS91720·650·471·000·000·00
1760·270·400·001·001·00
1800·040·060·000·000·00
1840·040·060·000·000·00
h0·520·630·000·000·00
EnMS10268c0·000·93
2711·000·07
h   0·000·14
EnMS111851·000·11
1910·000·89
h   0·000·20

Microsatellite markers clearly distinguished groups A and B, which were genotyped for 11 markers (Table 3). Two loci (EnMS5 and EnMS9) were monomorphic for the same allele in both groups. Only two alleles were found at each of nine polymorphic loci. Five loci were fixed for different alleles in group A and group B (EnMS2, EnMS3, EnMS6, EnMS7 and EnMS8). Group A was monomorphic at all loci. Group B was polymorphic at four loci (EnMS1, EnMS4, EnMS10 and EnMS11), but shared one allele at each of these loci with group A. Only two private alleles (at loci EnMS4 and EnMS8) were found in group B that were not found in North America (Table 3). No private alleles were found in group A.

Genotypic diversity was high in the northeastern and southeastern USA, where almost every isolate had a unique multilocus haplotype (Table 1). Genotypic diversity of E. necator from V. rotundifolia was intermediate, with five haplotypes among 10 isolates (larger samples could not be obtained because of the rare occurrence of powdery mildew on V. rotundifolia). Genotypic diversity in groups A and B was very low. Only one haplotype was observed in group A and only two in group B for the eight loci used for all samples; five haplotypes were found in group B when all 11 loci were assayed.

Bayesian analysis of data from eight loci (EnMS2EnMS9) assigned 129 E. necator individuals to three populations (= 3; Fig. 1a). The three populations corresponded closely to the eastern USA (all hosts and regions), group A and group B. Nine isolates from the southeastern USA had assignment probabilities >0·5 to the group A population. Six of these isolates (vir1, BlMtntS2, GAmer1, GAmer3, NCaes1 and RoACL2) had the same multilocus sequence haplotype as group A (Brewer & Milgroom, 2010); multilocus sequence data are not available for the other three isolates (NCaes9, NCaes8 and NClab4). Isolates with partial genetic affinity to group A are referred to as being A-like. Isolate NClab4 had an assignment probability of 0·98 to the group A population. NClab4 has alleles identical to the group A haplotype at six loci, but data were missing for two loci. Other A-like isolates that shared alleles with group A at five or six of the eight loci had smaller assignment probabilities. Three additional isolates (Biltcy2, NCrip1 and NClab1) had the same multilocus sequence haplotype as group A but had low assignment probabilities (<0·15) to the group A population based on microsatellite data. Only one group B isolate (FA18) showed any detectable assignment probability to the group A population (Fig. 1a).

Figure 1.

 Bayesian assignment of Erysiphe necator individuals to populations inferred using structure 2·2·3 (Pritchard et al., 2000) on multilocus haplotypes defined by eight microsatellite markers (EnMS2–EnMS9). (a) Assignment of 129 individuals from the eastern USA and Europe. (b) Assignment of 70 individuals from the eastern USA only. Each individual is represented by a single vertical bar divided into segments of different colours (or shades) corresponding to the inferred assignment probability to the populations determined by the analysis. Individuals are arranged randomly within the population from which they were sampled: northeastern USA, southeastern USA, Vitis rotundifolia and European groups A and B. The most probable number of populations (k) was estimated using the method of Evanno et al. (2005). Black arrows indicate A-like isolates from the southeastern USA that were assigned with high probability to the group A population; *Indicates isolate NCrip1, which has some affinity to the A-like group (see text).

The Bayesian assignments were repeated on North American isolates, excluding all isolates from Europe to avoid any disproportionate effect of analysing multiple individuals with the same few haplotypes. The largest posterior probability supported five populations (= 5; Fig. 1b). When = 5, isolates from V. rotundifolia were assigned to a distinct population. Ten isolates (the nine identified as A-like in the previous analysis plus NCrip1, marked with an asterisk in Fig. 1) were assigned with high probability (>0·6) to a second population. The rest of the isolates were assigned to three additional populations, all with intermediate probabilities (Fig. 1b).

Additional population structure was detected in the eastern USA based on geographic regions and hosts (Table 4). Genetic differentiation between northeastern and southeastern USA (excluding isolates from V. rotundifolia) populations was significant for two loci (EnMS3 and EnMS7) and for the overall comparison with eight loci (ΦST = 0·158, = 0·01, Table 4). This comparison was repeated excluding the 10 A-like isolates from the southeast because they were identified as a separate population in the Bayesian analysis. Differentiation was significant at one locus (EnMS3, = 0·02), but overall was low (ΦST = 0·019) and not significant (= 0·23) for eight loci combined. The population from V. rotundifolia was significantly differentiated from populations from other Vitis hosts for three loci (EnMS4, EnMS5 and EnMS7) and for the overall comparison with eight loci (ΦST = 0·079, = 0·03). Significant differentiation was detected at only one locus between isolates from cultivated V. vinifera and from wild hosts (EnMS9, = 0·02), but the overall ΦST estimate was not significantly greater than zero (= 0·22). Seven and eight private alleles were found in the northeastern and southeastern samples, respectively (Table 1), indicating some restriction of gene flow between these two regions.

Table 4.   Genetic differentiation of Erysiphe necator between regions in the eastern USA and between hosts of origin, based on eight microsatellite markers
PopulationsLocusMean ΦST
EnMS2EnMS3EnMS4EnMS5EnMS6EnMS7EnMS8EnMS9
  1. * 0·05.

  2. aA-like isolates include nine isolates that were clustered at high probability with isolates from European group A based on the results described in Fig. 1a. Assignment to most probable cluster was based on structure v2·2·3 (Pritchard et al., 2000).

  3. bWild Vitis hosts include isolates from V. rotundifolia (= 10), V. labrusca (= 6), V. riparia (= 14) and V. aestivalis (= 17).

  4. cOther Vitis hosts include isolates from V. vinifera (= 22), V. labrusca (= 6), V. riparia (= 14) and V. aestivalis (= 17).

Northeastern USA (= 27) vs. southeastern USA (= 33)
 ΦST−0·0270·425−0·0410·097−0·0100·2210·0120·0060·158
 P-value0·6800·010*0·8400·0700·4100·010*0·1800·3400·010*
Northeastern USA (= 27) vs. southeastern USA (= 24) without A-like isolatesa
 ΦST−0·0260·1820,000−0·015−0·0450·0020·050−0·0390·019
 P-value0·7100·020*0·4000·5000·8700·3400·120·8300·230
Vitis vinifera (= 22) vs. wild Vitis hosts (= 47) in eastern USAb
 ΦST0·043−0·0210·073−0·001−0·035−0·025−0·0270·1770·013
 P-value0·1000·5600·1400·3300·8300·5400·7900·010*0·220
V. rotundifolia (= 10) vs. other Vitis hosts in eastern USA (= 60)c
 ΦST−0·060−0·0580·1560·2640·122−0·053−0·0490·1600·079
 P-value1·0000·8600·020*0·010*0·0900·8400·6800·050*0·030*

Principal coordinates analysis was performed on the multilocus haplotypes defined by eight loci (EnMS2–EnMS9; Fig. 2). The first two axes explained a total of 74·3% of the variation. This analysis did not reveal any distinct patterns of differentiation, i.e. most haplotypes clustered into one large continuous group. Haplotypes from the northeastern and southeastern USA overlapped considerably, but not entirely. In particular, consistent with the Bayesian analyses, a separate population was evident that included the 10 A-like isolates from the southeastern USA and group A isolates from Europe. The group B haplotype that represented 26 of the 27 group B isolates clustered separately (Fig. 2). The haplotype for the remaining group B isolate (FA18 from Faugères, France) clustered with the large group from the eastern USA. One isolate (BlMtn2) from the southeastern USA was also an outlier from the main group as it had a unique allele (320) at locus EnMS4, without which this haplotype would have been placed in the main USA group.

Figure 2.

 Principal coordinates analysis of pairwise genetic distances among multilocus haplotypes of 129 Erysiphe necator isolates sampled from five different sources. Haplotypes are defined by alleles at eight microsatellite markers (EnMS2–EnMS9). Isolates in the USA were classified by the population from which they were sampled: northeastern USA, southeastern USA, and Vitis rotundifolia, whereas those from Europe were previously classified as groups A and B. Locations of the haplotype for group A and the most common haplotype for group B are indicated with arrows. Locations are also indicated for isolate G14 (transcriptome sequence used to discover microsatellite motifs), isolate FA18 (a group B isolate similar to haplotypes in the USA) and isolate BlMtn2 (outlying haplotype because of a unique allele at one locus). Solid black circle encloses the European group A haplotype and A-like haplotypes of isolates from the southeastern USA that were assigned with high probability to group A using structure 2·2·3 (Fig. 1a). Dashed black ellipse encloses all isolates from southeastern USA, except for isolate BlMtn2. Dotted grey ellipse encloses all but one of the isolates from the northeastern USA.

To test whether populations identified by Bayesian analyses (Fig. 1) had any discernible biological significance, aggressiveness was compared among USA isolates for which it had been estimated previously. Frenkel et al. (2010) measured latent period and lesion size of isolates from the eastern USA on V. labrusca hybrid cv. Niagara, which is moderately resistant, and on V. vinifera cv. Cabernet Sauvignon, which is highly susceptible. A-like isolates (vir1, BlMtntS2, GAmer1, GAmer3, NCaes1, NCaes8, NCaes9, NClab4, RoACL2 and NCrip1) had a significantly (< 0·001) shorter mean latent period (5·9 days) on V. labrusca cv. Niagara than the group of 39 isolates from the eastern USA (6·6 days), for which data were available, and the 10 isolates from V. rotundifolia (6·5 days) (Tables 5 and S3). A-like isolates also had a significantly shorter latent period on cv. Niagara than other isolates in the southeastern USA (6·4 days), excluding those from V. rotundifolia. No differences were detected among these three populations for lesion size.

Table 5.   Aggressiveness of Erysiphe necator, measured as latent period and lesion size, in populations in the eastern USA when inoculated on two hosts, Vitis vinifera cv. Cabernet Sauvignon and V. labrusca hybrid cv. Niagara [data from Frenkel et al. (2010)]
PopulationaNumber of isolatesInoculated hostb
V. labrusca hybrid cv. NiagaraV. vinifera cv. Cabernet Sauvignon
Latent period (days)Lesion size (mm2)Latent period (days)Lesion size (mm2)
  1. aPopulations were determined using structure v2·2·3 (Pritchard et al., 2000) (Fig. 1b). A-like isolates include: vir1, BlMtntS2, GAmer1, GAmer3, NCaes1, NCaes8, NCaes9, NClab4, RoACL2 and NCrip1. The main population in the eastern USA included all other isolates besides the A-like isolates and those from V. rotundifolia.

  2. bAs the interaction ‘inoculated host × population’ was significant in an anova for both latent period and lesion size the results are presented for each inoculated host separately (Table S3).

  3. cValues in the same column followed by the same letter are not statistically different (= 0·05) according to Tukey–Kramer Honestly Significant Difference test.

Main eastern USA population396·63 B97·0 A6·12 A102·9 A
A-like isolates105·87 Ac102·2 A6·15 A101·8 A
Isolates from V. rotundifolia106·48 B97·5 A5·95 A104·1 A

IA was not significantly different from zero in the northeastern USA (IA = −0·085, = 0·797, = 27) or southeastern USA (IA = 0·109, = 0·209, = 24) when the A-like and V. rotundifolia isolates were excluded. None of the 28 locus pairs (among eight loci) were in linkage disequilibrium in the northeastern USA, but three pairs were significantly associated in the southeastern USA. Linkage disequilibrium was not estimated for the A-like and V. rotundifolia isolates because of small sample sizes and/or the small number of polymorphic loci for these groups.

Discussion

Eleven microsatellite markers were developed for E. necator by mining transcriptome sequences. This approach succeeded better than previous efforts to find microsatellites in E. necator by enriching plasmid libraries for repeated motifs (Dutech et al., 2007). Out of 96 enriched clones with repeat motifs, Dutech et al. (2007) reported 76 with unique sequences, but only 18 sequences (19%) were suitable for primer design and three (3%) were polymorphic. Mining the transcriptome found 116 candidate sequences with repeat motifs, of which 31 (27%) were suitable for primer design; nine (8%) were developed as reliable and reproducible markers for all populations (EnMS1EnMS9) and two additional markers (EnMS10 and EnMS11) were developed for groups A and B found in Europe, giving a total of 11 markers. It is suggested that when studying European populations, markers EnMS10 and EnMS11 should replace markers EnMS5 and EnMS9, which are not polymorphic between groups A and B. Although markers EnMS10 and EnMS11 were not reliable for North American isolates in the initial screening, subsequent genotyping with these markers has shown that amplification is possible for North American isolates, depending on the quality and quantity of DNA template used (unpublished data). No attempt was made to redesign or optimize primers and/or PCR conditions for any of the 13 candidate sequences that did not amplify PCR products in the first attempts. However, we speculate that such attempts would potentially yield additional markers. Moreover, additional transcriptome sequence is being assembled (L. Cadle-Davidson & M.G. Milgroom, unpublished data) and has the potential to yield additional sequences for mining.

Microsatellite genotyping of samples from southern France clearly distinguished groups A and B. Five of the 11 markers had fixed differences between the groups in France and two locations in Italy. One allele from each of the four polymorphic loci in group B was also found in group A. This pattern could arise by several processes, although it is not possible to distinguish among them with these data. For example, they could have arisen independently by parallel mutations within groups. Alternatively, they could have arisen only once (i.e. identical by descent) and be present in both groups because of shared ancestral polymorphisms in the source population or by recombination between groups after their introduction into Europe. Citing unpublished results, Montarry et al. (2009) found discrepancies among genetic markers at a low frequency in France, possible evidence for recombination between groups A and B. However, the present study found no more than two alleles common to group A in any putative recombinant haplotype, even though the group A haplotype and the most common group B haplotype differed at nine loci. This is weak evidence, at best, for recombination, and genotyping larger samples will be needed to address this question.

As in previous studies of E. necator with other markers (Délye et al., 1997; Péros et al., 2006; Brewer & Milgroom, 2010), more diversity was found in group B than in group A. However, for the same 53 isolates from southern France, Péros et al. (2005) used RAPD markers and found four and 11 multilocus haplotypes in groups A and B, respectively, compared to one and five haplotypes with nine polymorphic microsatellite markers. Both types of markers unambiguously identified isolates as belonging to group A or group B. RAPDs or AFLPs may be sufficient for identifying specific haplotypes (e.g. DNA fingerprinting), but are less useful for population genetic analyses where one must be certain of scoring alleles at individual loci. Moreover, the robustness and repeatability of microsatellites between laboratories makes them more valuable as markers than RAPDs. For example, microsatellite markers could potentially be used for genotyping powdery mildew colonies collected directly from the field, as done previously for genotyping a SNP (Montarry et al., 2008), a SCAR marker (Miazzi et al., 2008; Montarry et al., 2008) and a microsatellite marker (Péros et al., 2006). Such an analysis would be impractical with RAPDs, AFLPs or any other marker that could amplify DNA of other organisms found on grape leaves, whereas SNP, SCAR and microsatellite primers are specific to E. necator. The challenge for any genotyping directly from colonies on leaves in the field, however, is to obtain sufficient quantities of high quality DNA for PCR.

Eastern USA populations were analysed with eight of the nine polymorphic markers. Marker EnMS1 was excluded because it was too variable; with 19 alleles among 70 USA isolates, the PCR products of the same size might not have been identical by descent, violating a crucial assumption of most population genetic analyses. However, this marker might be beneficial for studying populations where it is not hypervariable (e.g. Europe). With the other eight markers, 56 haplotypes were found among 60 isolates from the eastern USA (excluding 10 isolates from V. rotundifolia); no haplotype occurred more than twice. This degree of diversity was biased by the fact that these markers were chosen because they were polymorphic among the test panel of 15 isolates (i.e. ascertainment bias). Not surprisingly, gene and genotypic diversity was much greater in the eastern USA than in Europe, a result found also with multilocus sequencing (Brewer & Milgroom, 2010) and is consistent with the hypothesis that populations of E. necator in Europe are derived from introductions from native populations in the eastern USA. Source populations are expected to be more diverse than introduced populations because of founder effects (Brewer & Milgroom, 2010).

Although no microsatellite haplotypes were found in the eastern USA that were identical to those of groups A or B in Europe, the Bayesian cluster analysis assigned nine isolates from the southeastern USA to cluster with group A with probabilities of >0·5. Six of these nine isolates had identical multilocus sequence haplotypes to group A (Brewer & Milgroom, 2010); the other three were not sequenced. Three other isolates had the same multilocus sequence haplotypes but had low assignment probabilities to group A. Although they may have inherited these sequenced genes from a group A or A-like ancestor, recombination might have resulted in inheritance of non-A-like alleles at enough microsatellite loci to reduce their overall relatedness to group A. These nine isolates (plus one other) also kept their separate genetic identity when analysis was conducted solely with USA isolates, excluding the possibility that too much weight was given to repeated genotypes (A and B) and spurious groups detected.

Six of the 10 A-like isolates were collected from V. aestivalis and V. labrusca vines growing wild in relatively isolated areas in the Appalachian Mountains where commercial vineyards are not common. The presence of multiple isolates with the group A multilocus sequence haplotype in the southeastern USA (Brewer & Milgroom, 2010) raised the question of whether group A was a successful clone that had been introduced into Europe. However, the geographic isolation where the A-like isolates were collected minimizes the chances that they represent re-introductions from Europe. Moreover, microsatellites revealed considerable diversity among the A-like isolates. Most A-like isolates clearly have mixed ancestry with populations that comprise most of the eastern USA isolates, probably because of sexual reproduction. The reproductive biology and ecological requirements of the A-like group still need to be explored in North America because this group has maintained some degree of genetic identity, suggesting that it may be partially reproductively isolated from the rest of the population. In addition, A-like isolates were significantly more aggressive than the rest of the isolates from the eastern USA on the moderately resistant V. labrusca cv. Niagara. Greater aggressiveness might contribute to the maintenance of their genetic identity because it likely correlates with higher fitness and better competitive ability on the wild Vitis species in eastern USA, which are more resistant than V. vinifera (Cadle-Davidson et al., 2011). In Europe, group A is reproductively isolated from group B, has very low genetic diversity and several studies considered it clonal (Amrani & Corio-Costet, 2006; Montarry et al., 2009). Additional studies of the A-like individuals from the eastern USA might help to explain the reproductive isolation between groups A and B in Europe. Because A-like isolates were mostly sampled from wild V. aestivalis and V. labrusca and were more aggressive on cv. Niagara than the rest of the population in the eastern USA, they might also be good candidates to use for screening powdery mildew resistance in breeding programmes.

No evidence was found for the presence of group B haplotypes in eastern North America, even though they are present at high frequencies in Europe and Australia (Stummer & Scott, 2003; Péros et al., 2005). These results confirm those of Brewer & Milgroom (2010), who found group B multilocus sequence haplotypes in the western USA (California and Oregon), but not in the eastern USA, and speculated that they had been introduced there from Europe. The genetic differentiation of group B from any genotypes found in North America raises the question of whether it originated from outside of North America, or from areas of North America that have not yet been sampled. Sampling E. necator from wild and domesticated Vitis spp. in the western USA and conducting parallel complementary population genetic analyses might shed some light on this. Apparently, there are currently no data on the genotypes of E. necator from Asia, where some Vitis species are resistant to mildew (Wang et al., 1995; Cadle-Davidson et al., 2011). A population genetic analysis of E. necator in Asia will be needed to test the hypothesis that group B was introduced from Asia to Europe.

Evidence was found of population structure between the northeastern and southeastern USA and between V. rotundifolia and other Vitis hosts. Three analyses of population structure in this study (ΦST, structure and PCoA) and multilocus sequencing (Brewer & Milgroom, 2010), each based on different assumptions, gave consistent results. Almost all of the genetic differentiation between the northeastern and southeastern USA populations was attributable to the A-like subpopulation; many of the A-like isolates were determined to have the same multilocus haplotype as group A isolates in Europe and Australia (Brewer & Milgroom, 2010). Microsatellites provided a finer resolution than multilocus sequencing and revealed marked diversity. Genetic structure revealed by microsatellite genotyping is also consistent with ecological and epidemiological evidence. For example, Frenkel et al. (2010) found greater aggressiveness in the southeastern USA than in the northeastern USA. Most of this difference between regions is because the A-like isolates are more aggressive than other isolates and are found only in the southeastern USA. The role of climate, such as extreme winter and summer temperatures, as a selective force for ecological differences between subpopulations of E. necator in the two regions may be an interesting subject of future investigation to further explain regional differentiation.

The lack of genetic differentiation between isolates from V. vinifera and wild Vitis host species might be explained by the fact that V. vinifera is extremely susceptible to powdery mildew, so there would be little barrier to gene flow of E. necator from wild hosts to V. vinifera. Similarly, some analyses detected genetic differentiation between isolates from V. rotundifolia and isolates from other Vitis species, even though the sample size from V. rotundifolia was small (= 10) and from a relatively limited geographic area. Differentiation based on microsatellites is consistent with the finding that isolates from V. rotundifolia had unique multilocus sequence haplotypes (Brewer & Milgroom, 2010) and were the only ones that could establish colonies on an accession of V. rotundifolia (Frenkel et al., 2010).

The Bayesian analysis using structure (Pritchard et al., 2000) identified a population of A-like isolates, a population of isolates from V. rotundifolia and three additional populations in the eastern USA (Fig. 1b). These latter three populations have no simple explanation because all isolates (except for A-like isolates and V. rotundifolia isolates) showed mixed assignments to all three. Analysis with a different Bayesian assignment algorithm (BAPS, Corander et al., 2008) did not provide any additional clarity (results not shown). Whether these populations are artefacts of the analyses or have biological significance may require additional sampling.

Two unsolved problems need to be addressed about the microsatellite markers developed in this study. The first, which applies to all the genetic markers of E. necator, is that it is not known whether the markers are linked. Unfortunately, the difficulty of obtaining a sufficient number of ascospore progeny from crosses (Stummer & Scott, 2003) severely limits the ability to determine genetic linkages among markers. However, linkage disequilibrium (LD) was not found between any loci in the sample from the northeastern USA, and only a few pairs were significant in the southeastern USA, strongly suggesting that genotypic diversity of E. necator in populations in the eastern USA is maintained by regular sexual reproduction, consistent with epidemiological evidence for overwintering as ascospores. Lack of LD might indicate either the lack of physical linkage or sufficient recombination within natural populations to break down LD. Alternatively, sample sizes may have been too small to reliably detect LD, in which case the failure to reject the null hypothesis is not sufficient evidence for lack of linkage. The second problem is that these microsatellites are in coding regions and may be under selection (Li et al., 2004). Microsatellites in coding regions have been found to have fewer alleles and lower gene diversity than anonymous genomic microsatellites for some species (Peleg et al., 2008), whereas no differences were found in others (Breuillin et al., 2006). Therefore, it will be interesting to compare the unpublished genomic microsatellites in E. necator mentioned by Montarry et al. (2009) and Dutech et al. (2007) with transcriptome-derived (EST-SSR) markers to test if the two types of markers affect interpretations of population structure.

Although it is not known whether the microsatellites described here are under selection, they otherwise fulfil the characteristics of ideal genetic markers by being reproducible, codominant and affordable (Brown, 1996), and will be useful in future studies of the population biology of E. necator. For example, it would be possible to address the extent to which the two genetic groups found in Europe are reproductively isolated. Montarry et al. (2009) cited interesting, but unpublished, results that suggested a low frequency of recombinants between groups A and B in France. With a larger number of markers available, it will be easier to detect recombination between genetic groups. Studies of the mating system of E. necator can also be conducted to test for LD within local populations, e.g. within a single vineyard, to avoid problems of admixture inherent in sampling from broad geographic areas. Large sample sizes necessary to achieve sufficient statistical power for testing hypotheses of LD can be genotyped more easily (and more economically) with microsatellites than with multilocus sequencing.

Acknowledgements

We thank Monica Miazzi for sharing DNA from E. necator isolates from southern Italy. We also thank anonymous reviewers for helpful comments on previous drafts. This research was supported by Vaadia-BARD Postdoctoral Fellowship Award No. FI-410-2008 from the United States – Israel Binational Agricultural Research and Development Fund (BARD) to OF, a Research Travel Grant from the Cornell University Graduate School to MTB, a USDA-Viticulture Consortium-East award to LC, and Hatch project NYC-153410 to MGM.

Ancillary