Further analysis was based on 403 unique accessions: 296 from the Davis collections and 107 from the INRA Domaine de Vassal germplasm collection. Additional file 3: Table S3 presents the fingerprint profiles of the 403 unique accessions based on 19 SSR markers – one marker from each grape chromosome. Twenty-three accessions were removed from the study set. The number of alleles per marker and percent of missing data were calculated for the remaining 380 genotypes with 34 markers . Based on the collection records, the study set of 380 unique accessions consisted of 306 genotypes of V. vinifera subsp. sativa, 40 accessions of V. vinifera subsp. sylvestris, and 34 accessions of ViThis species from northern Pakistan, Afghanistan and China. A minimum of 9 and maximum of 44 alleles were observed with SSR markers VVIq52 and VVIv67, respectively. The average number of alleles for all markers was 22. There were 7 markers with 5% or more accessions that had missing data .Prior to phenotypic evaluation for powdery mildew resistance, the entire set of 403 accessions was genotyped for linkage with the powdery mildew resistance locus Ren1 at four SSR marker, which span 8.1 cM genetic block on chromosome 13. The resistance linked allele of 260 defined by marker VMCNg4e10.1 was observed in 47 accessions that included V. amurensis, V. romanetii, Muscadinia rotundifolia and wild V. vinifera subp. sylvestris accessions. The majority of the accessions with allele 260 for marker VMCNg4e10.1 did not have allele 216 at the marker UDV124. However, cut flower bucket a missing allele at either marker could have been due to a recombination event. Eleven accessions, including ‘Karadzhandal’ and ‘Kishmish vatkana’ had alleles that are in linkage with the Ren1 locus at the two distal markers, VMCNg4e10.1 and UDV124 that are in linkage with the Ren1 locus .

The germplasm set was then evaluated for alleles at sc47-18 and SC08-0071-014 that flank each side and co-segregates with the Ren1 locus. ‘Karadzhandal’ and ‘Kishmish vatkana’ fingerprint profiles were used to determine whether the allele of 249 defined by the marker sc47-18 and the allele of 143 defined by the marker SC08-0071-014 were linked to resistance. These are the two alleles that co-segregate with the Ren1. Allele 249 for marker sc47-18 was common: 80 of the 403 accessions shared it. Nearly all of the 47 accessions that carried the allele 260 for marker VMCNg4e10.1 also had allele 249 for marker sc47-18, confirming the tight linkage between these two markers . The allele 143 for marker SC08-0071-014 was rare; only 17 accessions in the entire data set carried it . Six accessions had the allele 143 for marker SC8-0071-014, but did not carry the resistance associated alleles at all tested markers . ‘Khalchili’ and ‘Khwangi’ had 143 flanked by 249 at sc47-18. Two of the V. vinifera subsp. sylvestris accessions had alleles 143 and 260 on the opposing flanks for marker VMCNg4e10.1; ‘Matrassa’ and a third sylvestris accession had the 143 allele and no resistance associated allele on the opposite flank. These six accessions are potential recombinants. Two Chinese species accessions, V. romanetii and V. yenshanensis had the 143 allele, but neither of them carried a resistance-associated allele at the other three markers. In the case of these two examples, we speculate that the presence of the 143 allele is either due to size homoplasy, or the alleles at other markers are lost due to recombinations. In 2009 and 2010, resistance to powdery mildew was evaluated on a 0 to 5 scale on all accessions from Davis that carried alleles linked to resistance at one or more markers . Four other accessions were evaluated in 2012 . The year effect was significant; disease pressure was more severe in 2010 . However, the susceptible controls were highly susceptible and resistant controls had no or minor symptoms in three test years . The location in the field test plot was not significant, indicating that the close-spaced field evaluation site with no spray was an efficient and cost effective way to screen for resistance .

‘Karadzhandal’, a known powdery mildew resistance accession was evaluated both years and had a powdery mildew resistance score of < 1. ‘Kishmish vatkana’ the other previously known powdery mildew resistant accession was under quarantine as part of the importation process and could not be evaluated. Accessions with Ren1-linked alleles at only one of the flanking markers exhibited no resistance to powdery mildew in the field test . The V. vinifera subsp. sativa cultivars: ‘Husseine’, ‘Khalchili’, ‘Late Vavilov’ and ‘Sochal’ were resistant. These four accessions had mean scores for leaf and cane PM symptoms ranging from 1.08 – 2.42 in 2009 and 0.83 – 2.42 in 2010 . In 2012, in a much smaller evaluation, two accessions from wild V. vinifera subsp. sylvestris were identified as resistant from field evaluations: O34-16, collected from Shiravan, Iran, had all four resistance-linked alleles; and DVIT3351.27, collected from Armenia, had the resistance allele at one of the flanking markers on each side. Mean leaf symptoms scores were 0.4 and 0.7 for DVIT3351.21 and O34-16, respectively . The powdery mildew resistant accession ‘Karadzhandal’ at Davis had similar marker profile to accession ‘Kara djandjal’ and the newly identified resistant accession ‘Husseine’ had identical marker profile to ‘Kandari noir’ inthe Vassal collection. Both of these accessions were found to be resistant in greenhouse screens carried out at Vassal. ‘Chirai oback’ and ‘Vassarga tchernaia’, which had all four Ren1-linked SSR marker alleles, were resistant based on greenhouse screening in France. The other two powdery mildew resistant accessions were ‘Soïaki’, which had resistance-linked alleles with four markers, and ‘Matrassa’, which had resistance-linked alleles with two markers on one side, have not been evaluated for disease symptoms. In total, this study identified and verified eight new accessions that are powdery mildew resistant. ‘Soïaki’ and ‘Matrassa’ were identified as potentially resistant based on marker analysis. Their disease resistance needs to be verified in a field or greenhouse screen.Probability of identity analysis found that nine markers were sufficient to identify unique accessions in the study set . The paternity exclusion probability for a single locus ranged from 10.6% to 72.6% . A cumulated probability of exclusion of 100% was reached using only 7 markers for paternity and 3 markers for a parent pair. The simulation for parentage analysis identified a LOD score threshold of 5.0 to assess a potential single parent and 4.0 to assess a parent pair with 34 SSR markers. Six newly identified resistant accessions ‘Husseine’, ‘Chirai obak’, DVIT3351.27, O34-16, ‘Soïaki’ and ‘Matrassa’ were not related to any other accession in the set; two of these are V. vinifera subsp. sylvestris.

The presence of powdery mildew resistance in unrelated genetic backgrounds is a very important result of this study that suggests that powdery mildew resistance in Central Asia is complex and potentially represent orthologous and paralogous homology for the Ren1 locus, first identified in ‘Karadzhandal’ and ‘Kishmish vatkana’. A second important inference from these results is that there maybe many more powdery mildew resistant accessions within V. vinifera subsp. sativa and Central Asian ViThis species and further exploration is needed. This resistance could be the result of intentional breeding efforts involving material collected and curated in the early 1900s at multiple institutes set up by the Russian geneticist Vavilov, or the result of unintentional breeding and selection of resistant material in an earlier period of domestication and selection over thousands of years. More importantly, identification of powdery mildew resistance in accessions of V. vinifera subsp. sylvestris indicates that the resistance is present in wild germplasm. There were four parent-progeny relationships identified in this study; two involved the previously published powdery mildew resistant accessions . ‘Vassarga tchernaia’ was conclusively identified as the female parent of ‘Kishmish vatkana’; it also shared a parent-progeny relationship with ‘Sochal’ although the direction of the cross is unknown. ‘Karadzhandal’ and ‘Late Vavilov’ share one or both alleles with all 42 SSR markers . The powdery mildew resistant accession ‘Khalchili’ was shown to have a first-degree relationship to ‘Yarghouti’ .A 620 bp region that includes the resistance associated 143 bp allele from marker SC8-0071-014 was sequenced for the 12 powdery mildew resistant accessions and two susceptible V. vinifera subsp. sylvestris . Two accessions of Chinese species that had a 143 bp fragment at SC8-0071-014 were also sequenced. The sequences were nearly identical for all fourteen V. vinifera accessions except for occasional single nucleotide polymorphisms between both unrelated and genetically related accessions. The sequences of the two Chinese species, V. romanetii and V. yenshanensis were very different from one another and from the V. vinifera sequence, obvious examples of size homoplasy, where two alleles are identical in size but result from independent events .For the 12 powdery mildew resistant accessions, including the two previous known powdery mildew resistant accessions ‘Kishmish vatkana’ and ‘Karadzhandal’, genetic analysis was expanded to a 26 cM genomic block with six SSR markers including the Ren1 region . Six of the newly identified powdery mildew resistant accessions, including O34-16, a V. vinifera subsp. sylvestris, had similar alleles similar to ‘Kishmish vatkana’, flower display buckets and ‘Karadzhandal’ with six SSR markers . SSR marker allelic comparison of two other resistant accessions indicated that a recombination event had occurred between markers at different junctions. The wild subsp. sylvestris accession, DVIT3351.27 had complex allelic combination of markers surrounding the Ren1 region suggesting a different genetic origin of powdery mildew resistance.The genetic diversity of the core set of 380 Central Asian accessions was evaluated with hierarchical clustering , principle coordinate analysis , and a model-based clustering method implemented in the program STRUCTURE. All three analysis methods generated three groups with data from 19 or 34 markers. The delta K value calculated from the output of STRUCTURE was 45.0 at K = 3 compared to less than 5.0 at all other values of K. The three groups determined by PCoA were similar to those produced by STRUCTURE . The Q-values assigned by STRUCTURE for 380 accessions in three groups are displayed in Additional file 9: Table S9. Group A contained 29 ViThis species accessions, nearly all of which originated in China. The Q-value for membership in this group was 0.90 or above for 25 accessions. ViThis yenshanensis and B-166-016, an accession labeled as ViThis spp., both collected from China had Q-values split between the group A and B. A second V. yenshanensis accession, also collected from China, had Q values split between group A and C. ‘Khir Ghuluman’, collected by H.P. Olmo in Afghanistan, had Q-values split among all three groups indicating it was a possible hybrid of a local species and cultivated varieties. ‘Khir Ghuluman’ was labeled as V. vinifera, presumably because it is a cultivated variety in Afghanistan. None of the previous and newly identified powdery mildew resistant accessions were in this group. Group B contained 165 samples, a mix of both V. vinifera cultivars and wild V. vinifera subsp. Sylvestris accessions. Three new powdery mildew resistant accessions were placed in this group with Q-values of 0.80 and higher. O34-16 and DVIT3351.27 were collected as subsp. sylvestris, and ‘Matrassa’ was collected as subsp. sativa. For clarity, this group will be referred to as O34-16 from this point onward in the manuscript. Thirty nine accessions collected as subsp. sylvestris were in this group, 36 of which had Q-values of 0.80 or higher. Two Chinese ViThis species accessions, 588650.a and B-166-019 were in this group, and are most likely hybridized forms. All four accessions of V. jacquemontii collected from Pakistan were also in this group. The Q values of these accessions, one as high as 0.97, suggest that these accessions are not pure species and may be hybrid or mislabeled forms . Group C was named after ‘Thompson seedless’ to indicate that the group consists primarily of table grape cultivars. Two previously identified and seven of the new powdery mildew resistant accessions were in this group . The group consisted of 185 accessions labeled as V. vinifera subsp. sativa and one accession of subsp. sylvestris – O35-64 collected from Iran by H.P. Olmo. The Q values for this particular accession placed it in groups A and C. ‘Kala Khostan’ with a group C Q-value of 0.66 is the only other accession in this group with association to the species group. All but 10 of the remaining 184 accessions had group TSL Q-values of 0.70 or higher .