The results of the PCoA analysis with 34 markers also produced three groups. The species group was clearly separated from the other two groups. The distinction between the O34-16 group and the TSL group was less clear . The O34-16 group contained nearly all of the subsp. sylvestris accessions, however within that group, there was no clear distinction between the cultivated and wild forms . To differentiate the wild sylvestris accessions from cultivated sativa forms, further analyses focused on only the O34-16 group. The Ward and UPGMA hierarchical clustering methods divided the 165 accessions of the O34-16 group into two clades. Seventy-six cultivated V. vinifera subsp. sativa accessions, including the powdery mildew resistant ‘Matrassa’ were in one group, and the second group consisted of 89 sativa and sylvestris accessions. When the Ward clustering method was applied to this second group of mixed accessions, there were again two clades. These four V. vinifera accessions are ancient cultivars and are likely transitional forms with the wild ancestor sylvestris. The powdery mildew resistant accession DVIT3351.27 was in this clade. The second clade had three less well-defined sub groups: the first contained six sylvestris accessions including the powdery mildew resistant accession O34-16; the second group contained accessions of V. vinifera subsp. sativa collected from Pakistan and Turkmenistan and one accession labeled V. jacquemontii; and the third group was a mix of wild sylvestris or feral types that were collected from Iran, Iraq, Turkmenistan, Pakistan and Russia, vertical farming system and five incorrectly identified accessions – three of which were labeled as V. jacquemontii, and the other two were collected from China.

Two distinct clades were revealed when clustering analysis was applied to the 40 wild sylvestris accessions . The first clade contained ten sylvestris accessions obtained from Turkmenistan, four accessions from Iran and two from Afghanistan. The Turkmenistan accessions were collected from the Kopet Dag mountain range, which defines the border between Turkmenistan and Iran on the east of the Caspian Sea. The powdery mildew resistant accession O34-16, collected near the town of Shirvan, Iran, was in this clade. Shirvan is near Mashhad, an important trade hub on the ancient silk route located on the other side of the Kopet Dag mountain range . Therefore, it was not surprising to see these accessions positioned in one clade. The second clade consisted of accessions collected from Georgia, Armenia and Iran. The powdery mildew resistant sylvestris accession DVIT3351.27, collected from Alaverdi, Armenia, was in this group. These results in conjunction with analyses that included the entire O34- 16 group suggest that the two wild sylvestris accessions could have acquired powdery mildew resistance from different genetic backgrounds. Gene diversity indices for each group are shown in Table 2. The average FIS value for the Species group was higher than the other two groups. This group consisted of only 29 accessions as compared to 165 accessions in the O34-16 group and 186 accessions in the TSL group. The average FIS for the O34-16 group was 0.10; expected heterozygosity was higher than observed for all but one of the 34 markers. The average near zero FIS value for the TSL group suggests that they are a panmictic population in Hardy-Weinberg equilibrium. Values for the differentiation index among the three groups were very low . These results indicate that there is no clear differentiation among these groups due to the presence of transitional forms that bridge the groups and indicate active gene flow among them, implying that domestication and selection is underway.Because two of the newly discovered powdery mildew resistant accessions were collected as sylvestris, it was important to confirm their true type based on morphological traits. Flower sex phenotype and seed morphology are two key criteria used to differentiate subsp. sylvestris from cultivated sativa forms .

The flower phenotype of the subsp. sylvestris accessions collected from Armenia, Georgia and Turkmenistan could not be determined because they were young potted plants. Flower phenotype data for 15 wild V. vinifera accessions was obtained from GRIN, the National Germplasm Resource Information Network [31]. A combination of two DNA markers was used to differentiate male, hermaphrodite and female flower phenotype for the set of 380 accessions. Field phenotypic observations for the 95 accessions from the Vassal collection matched the flower phenotype predicted by DNA analysis with only one exception – ‘Yhsouh ali’ , which was recorded as a female, but DNA analysis indicated it was a hermaphrodite. These test results indicate that the combination of both markers is a reliable system to determine flower phenotype. DNA marker-based flower phenotyping of the 40 wild forms of V. vinifera subsp. sylvestris and all ten newly discovered powdery mildew resistant accessions are presented in Table 4; the results for all other accessions are presented in Additional file 11: Table S10. The flower phenotype was undetermined for 11 accessions due to ampliftication failure with one or both markers. Phenotypic observations differed from genotypic results for only three accessions. Two accessions in the species group, C-166-025 and DVIT1159.3, are recorded as male, but are hermaphrodite based on the DNA analysis. The third anomaly was the cultivar ‘Neeli’ , which was scored as a hermaphrodite, but is listed as a female plant in the GRIN database. DNA marker analysis of flower sex in the V. vinifera subsp. sativa group of cultivars found that 223 were hermaphrodite, 57 were female, and five were identified as male . One of the five males, ‘Kala Kostan’ is recorded as a female in GRIN; the flower phenotype could not be verified for the other four genotypically male cultivars. Eighteen of the 40 V. vinifera subsp. sylvestris accessions were male, including newly identified powdery mildew resistant accession, DVIT3351.27. Eight accessions were female, including resistant accession O34-16 . Fourteen others were hermaphrodite. Seeds were extracted from ten of the wild sylvestris that H.P. Olmo collected from Iran and Afghanistan .

The combined results from seed morphology and flower sex phenotyping, revealed that 14 accessions designated as sylvestris are likely not pure sylvestris, but instead hybridized forms of native wild species and cultivated varieties. interestingly, three accessions from H.P. Olmo’s O series, which are male, based on genotypic analysis, bear fruit . The flower phenotype of the accession O34-26 was scored differently in each of three years on GRIN. Similarly, observations of the flower phenotype for O34-55on GRIN varied from year-to-year between hermaphrodite and female. O35-47, the third genotypically male accession has been recorded as a hermaphrodite. The flower sex of the 12 powdery mildew resistant accessions was also determined: the sativa accessions ‘Husseine’ ‘Soïaki’, ‘Sochal’, and ‘Vassarga tchernaia’ are female vines; the other six including ‘Kishmish vatkana’ and ‘Karadzhandal’ are hermaphrodites . Two of the new powdery mildew resistant accessions are clearly V. vinifera subsp. sylvestris. O34-16 is a female vine with obvious wild type seed morphology . The accession DVIT3351.27 is a male flowered V. vinifera subsp. sylvestris.In this study, we exploited available genetic information on the powdery mildew resistance locus Ren1 to identify additional germplasm that shared a Ren1-like local haplotype, and then attempted to clarify the evolution of powdery mildew resistance and its domestication in cultivated V. vinifera subsp. sativa. Ten new powdery mildew resistant accessions were discovered that possess a Ren1-like local haplotype, which was earlier identified in ‘Kishmish vatkana’ and ‘Dzhandzhal kara’ from Central Asia. We discovered that powdery mildew resistance is present in two V. vinifera subsp. sylvestris accessions, a taxon considered to be the progenitor of the cultivated form sativa. Four of the resistant accessions ‘Vassarga tchernaia’, ‘Chirai ’, ‘Late Vavilov’ and ‘Khalchili’ are obscure varieties with few records in the ViThis International Variety Catalog or the European ViThis database. The first three accessions were obtained from germplasm collections in Uzbekistan, Tajikistan, and Turkmenistan, respectively; Harold P. Olmo collected ‘Khalchili’ from Afghanistan in 1948. The other four resistant sativa accessions are better known. ‘Husseine’ was also collected from Afghanistan and it is available worldwide with records in 20 germplasm collections with 61 synonyms. ‘Matrassa’ was acquired from the Azerbaijan collection, and is available in 15 collections with 26 synonyms. ‘Soïaki’ is found in 10 collections with 3 synonyms. ‘Matrassa’ and ‘Soïaki’ are listed by Russian grape breeders as cultivars for high quality table, sparkling and dessert wines. The eighth resistant sativa accession ‘Sochal’ is only held at two collection sites in the USA. Plant inventory records indicate that cuttings of ‘Sochal’ were obtained in 1971 from the N. I. Vavilov institute of Plant industry, Leningrad. Eight of the newly identified accessions carrying Ren1-like local haplotypes were acquired from five neighboring countries of Central Asia and the Caucasus, all major junctions for trade on the ancient silk route for thousands of years. It is not hard to believe that selected grape germplasm, vertical farming racks favored for desirable fruit characteristics, was moved back and forth in the form of seeds and cuttings from one region to another, where they were likely crossed with local varieties in remote isolated valleys and villages in different regions. In addition to the identification of eight new powdery mildew resistant accessions, this study also gathered information on the genealogical relationships. A likelihood based method that determines potential parent progeny relationships without any prior knowledge revealed four first-degree relationships. We identified ‘Vassarga tchernaia’ as the female parent of ‘Kishmish vatkana’ and verified ‘Thompson seedless’ as the male parent. ‘Vassarga tchernaia’ and ‘Sochal’ shared a first-degree relationship, sharing at least one allele at 42 markers. Both are female vines with reflexed stamens and seeded fruit. It is difficult to determine the direction of the relationship between ‘Sochal’ and ‘Vassarga tchernaia’. Nevertheless, both of them are female vines, resistant to powdery mildew, and produce seeded fruit. ‘Sochal’, ‘Vassarga tchernaia’ and ‘Kismish vatkana’ are not found in historical collection/ breeding records and may have been disregarded due to undesirable fruit attributes, e.g. loose clusters, and small seeded berries, which did not saThisfy selection criteria for that particular region.

The other two first-degree relationships identified in this study were between ‘Late Vavilov’ and ‘Karadzhandal’, and between ‘Khalchili’ and the powdery mildew susceptible ‘Yarghouti’. All four arehermaphrodites and only ‘Karadzhandal’ is well known with a recorded history. One of the important findings of this study is that four of the new powdery mildew resistant cultivars ‘Chirai obak’, ‘Husseine’, ‘Matrassa’ and ‘Soiaki’ are not directly related to any other accession in this study or in the complete Vassal collection. This implies that the story of powdery mildew resistance in cultivated varieties is complex and what we have revealed in this study may not be the complete picture due to extinction or missing cultivars in our collections. It is likely that a thorough search of germplasm collections in Central Asia would unearth more resistant germplasm. With the exception of ‘Matrassa’, STRUCTURE placed all seven new subsp. sativa powdery mildew resistant cultivars in the TSL group, even though they were collected from different regions of Central Asia. These results suggest that selection and active flow of desirable plant material was common in this region of grape domestication and that multiple breeding efforts were underway to saThisfy the local tastes for quality grapes. The TSL group also indicates that breeding efforts were also directed at seedlessness as ‘Thompson seedless’, one of the ancient varieties, was a popular parent for a large number of table grape cultivars. Two of the resistant accessions in this study belong to the subsp. sylvestris, which prompts many questions. Are these two accessions truly wild sylvestris that have never been cultivated or are they hybrids between wild and cultivated forms? What is the direction of the gene flow for powdery mildew resistance – did these two sylvestris accessions acquire resistance from cultivated forms or did the resistance come from wild types to cultivated forms? O34-16 is a female vine with seeded fruit, and seed shape typical of the sylvestris type grapes – small round seeds and short beaks. The accession DVIT3351.27 is a male vine. Dioecy is one of the key traits distinguishing the wild sylvestris from the cultivated sativa. Additionally, the male flower phenotype is only associated with wild ViThis species. According to the model of Antcliftf , flower phenotype is controlled by a single major locus with three alleles: male dominant to hermaphrodite , which is dominant to the female . In the wild, one should find only male and female vines in the absence of gene flow from hermaphroditic cultivated varieties.