Optimal value of K for the full germplasm set of 1197 accesions. Fifty data sets were obtained by setting the number of possible clusters (K) from 1 to 10 with five replications each. The log probability of data [LnP(D)] value for each given K increased continuously, but did not show an abrupt change; the optimal K value could not be inferred (A). But after applying the Evanno’s correction method (Evanno et al., 2005), there was a clear peak of ΔK at K = 2, indicating two major populations (Pop I and Pop II) in this panel of collections (B).
Model-based cluster membership of 1197 Citrullus accessions in two main populations (Pop I and Pop II) and four subpopulations (Pop IA, Pop IB, Pop IIA, and Pop IIB). Pop IA was consisted of 238 accessions, a mix of American and East Asian ecotypes. Pop IB had 101 accessions, all from East Asian ecotype. Pop IIA was a typical American ecotype panel and contained 368 accessions. Pop IIB contained 71 accessions, which was a wild species mixed subgroup, with the accessions belonging to the six species.
A dendrogram of 71 accessions in Pop IIB, displays three major clusters. The groups represent Citrullus naudinianus group [CXG, PI618817, and PI 671961 (black)], Citrullus colocynthis group (green) that also included the two Citrullus rehmii accessions [Grif 16376 and PI 632755 (pink)], Citrullus amarus group (yellow) that also includes the two accession of Citrullus ecirrhosus [PI 632751 and Grif 16945 (blue)], and the four accession of Citrullus lanatus subsp. vulgaris group [PI 482288, PI 505586, Seychelles and Colorado Preserving (purple color)].
Distribution of Nei’s gene diversity index (Nei) and Shannon’s information index (SW) among 23 SSR markers in the core set (C-Nei; C-SW) and the entire collection (E-Nei; E-SW), compared with 84 American core set (AC-Nei; AC-SW). Nei’s gene diversity index and SW were calculated using the POPGENE v1.32 software (Yeh et al., 1997). The entire collection included 1197 accessions, a core set of 130 accessions was developed from the entire collection, and 84 watermelon from the U.S. Department of Agriculture core collection we analyzed.
Click on author name to view affiliation information
Watermelon belongs to the genus Citrullus. There have been continuing interests in breeding of watermelon for economic benefits, but information on the scope and utilization of genetic variations in Citrullus is still limited. The present study was conducted in 2012–13, to evaluate the genetic diversity and population structure of the 1197 line watermelon collection maintained by the Beijing Vegetable Research Center (BVRC), which belongs to seven Citrullus species including Citrullus naudinianus, Citrullus colocynthis, Citrullus rehmii, Citrullus ecirrhosus, Citrullus amarus, Citrullus mucosospermus, and Cirullus lanatus subsp. vulgaris. Twenty-three highly informative microsatellite markers evenly distributed in the watermelon genome were used to assess genetic diversity in this collection. The markers detected on an average of 6.05 alleles per locus with the average value of polymorphism information content (PIC) at 0.49. A high level of gene diversity [Nei’s gene diversity index (Nei) = 0.56] and a low observed heterozygosity (Ho = 0.10) were revealed within the collection. Structure analysis grouped the 1197 accessions into two main populations (Pop I and Pop II) and an admixture group. Pop I contained 450 accessions from C. lanatus subsp. vulgaris (446) and C. mucosospermus (4). Pop II comprised 465 accessions, 379 of which belonged to C. lanatus subsp. vulgaris and 86 to C. naudinianus (3), C. ecirrhosus (2), C. rehmii (2), C. colocynthis (11), C. amarus (58), and C. mucosospermus (10). The remaining 282 accessions were classified as an admixture group. The two main populations were further subdivided into four subgroups. The groupings were consistent with the estimation of F statistics (Fst) and Nei’s genetic distances in collections. We confirmed the distinct genetic backgrounds between American and East Asian ecotypes. Subsequently, we defined a core set consisting of 130 accessions including 47 from Pop I, 68 from Pop II, and 15 from the Admixture group. This core set was able to capture all 133 alleles detected by 23 simple sequence repeats (SSRs) in 1197 accessions. These results will facilitate efficient use of genetic variations in Citrullus in watermelon breeding and help optimization of accessions in genomewide association studies.
Watermelon is a major vegetable crop throughout the world. It is an important source of income for small-scale farmers of China. There is an ongoing need to improve watermelon, particularly for higher disease and pest resistances and yield. However, most breeding programs have relied on the use of established cultivars or elite breeding lines, which resulted in lower genetic diversity within cultivated watermelon, impeding progress of watermelon breeding efforts (Levi et al., 2004). It has been shown that the narrow genetic base in watermelon can be averted with the use of wild materials (Levi et al., 2013). Since publicly accessible passport data on features and variations for the collections are limited (Munisse et al., 2013), it would take considerable time and effort to develop adapted, high-performing lines from crosses between elite and wild or exotic genotypes (Kottapalli et al., 2007). As such, commercial breeders are often reluctant to use the wild or exotic relatives directly to increase the genetic diversity of elite germplasm. Information on genetic differentiation and relationships of germplasm will be beneficial in efficient use of genetic resources.
Watermelon belongs to the genus Citrullus, but the taxonomic status of species, subspecies, and varieties in this genus is evolving. Previously, Citrullus included four species including C. lanatus (Thunb.) Matsum. Et Nakai, C. colocynthis (L.) Schrad., C. eccirrhosus Cogn., and C. rehmii de Winter. Cultivated watermelon belongs to C. lanatus (Jarret and Newman, 2000). Three subspecies of C. lanatus were recognized, including subsp. vulgaris, representing the common sweet watermelon group, subsp. lanatus, comprising the citron and “tsamma” types, and subsp. mucosospermus, encompassing the “egusi” seed watermelons. C. lanatus is also frequently divided into two botanical varieties: var. lanatus and var. citroides with the former including the sweet type (subsp. vulgaris) and seed type (subsp. mucosospermus) (Laghetti and Hammer, 2007). Recent molecular phylogenetic analysis found that the three subspecies of C. lanatus were in fact unrelated species (Chomicki and Renner, 2015). As such, it was proposed that the genus Citrullus should include seven species: 1) C. lanatus with the sweet watermelon group as C. lanatus subsp. vulgaris, 2) C. amarus, also known as C. lanatus var. caffrorum or C. lanatus var. citroides, 3) C. mucosospermus, the “egusi” melon, previously treated as a subspecies of C. lanatus, 4) C. colocynthis, which is perennial and growing in northern Africa and adjacent Asia, 5) C. ecirrhosus, a tendril-less South African endemic, another perennial wild species, 6) C. rehmii, an annual wild species, and 7) C. naudinianus from the Namib–Kalahari regions (Achigan-Dako et al., 2015; Chomicki and Renner 2015).
This new taxonomic classification likely improve the functionality of Citrullus genetic resources for watermelon breeding. Molecular characterization (Dane and Liu, 2006; Levi et al., 2013; Minsart et al., 2011) revealed a high level of genotypic diversity between C. colocynthis and C. lanatus subsp. vulgaris or C. mucosospermus, and low genetic differentiation within C. lanatus subsp. vulgaris as compared with C. lanatus var. citroides (C. amarus). Obviously, additional work is needed to investigate the phylogenetic relationships among Citrullus and explore their potential in watermelon breeding (Chomicki and Renner, 2015; Levi et al., 2013; Mujaju et al., 2012; Nantoumé et al., 2013).
There are abundant collections of Citrullus germplasm deposited in various gene banks worldwide, which offer considerable opportunities for trait improvement in watermelon breeding. However, the large size, heterogeneous structure, and unavailability of information on trait diversity hamper the successful utilization of the genetic potential of these collections. At present, it is not realistic to characterize all available collections due to associated cost in labor, space, and time. Also, regeneration is a major problem, due to the numerous practical precautions required by the insect-mediated cross-pollination mating system of Citrullus. For the convenience of management, research, and application, Frankel and Brown proposed the concept of core collections (Frankel and Brown, 1984). A core collection is the maximum possible genetic diversity contained in the entire collection with a minimum of repetitiveness. To date, core collections have been established for many species, including tomato (Solanum pimpinellifolium L.) (Rao et al., 2012), cucumber (Cucumis sativus L.) (Lv et al., 2012), melon (Cucumis melo L.) (Hu et al., 2014), apple (Malus ×domestica Borkh.) (Liang et al., 2015), common wheat (Triticum aestivum L.) (Wingen et al., 2014), and peanut (Arachis hypogaea L.) (Upadhyaya et al., 2003). There is also a core set for watermelon (251 accessions) which was developed based on morphological descriptors and random selection by the USDA (http://www.ars-grin.gov/cgi-bin/npgs/), and the 84 watermelon from the USDA core collection we analyzed; however, a core collection under a genome-wide molecular characterization has not been performed.
The power of molecular markers in assessing the genetic diversity and helping the management of plant genetic resources (e.g., development of a core collection) has been well demonstrated (Kong et al., 2014). From a plant breeding perspective, molecular analysis can help to define the lineage relationships among genetic materials and to eliminate the phenomenon of homonyms as well as aiding identification of breeding lines carrying genes of interest (Hu et al., 2014). In watermelon, molecular markers have been used for fingerprinting, identifying and characterizing watermelon genotypes, and examining population structure (Levi et al., 2013; Mujaju et al., 2010; Nantoumé et al., 2013; Nimmakayala et al., 2014; Zhang et al., 2012). Nevertheless, a broader understanding of the diversity and differentiation among and within Citrullus accessions is still lacking; also, there seems to be no information available on the overall genetic diversity and population structure of a wide spectrum of watermelon accessions.
The BVRC has maintained 1197 accessions representing seven species of Citrullus (C. naudinianus, C. colocynthis, C. rehmii, C. ecirrhosus, C. amarus, C. mucosospermus, and C. lanatus subsp. vulgaris). From the watermelon draft genome sequences, we previously identified 23 SSR markers that are highly polymorphic and evenly distributed across the watermelon genome (Zhang et al., 2012). The objectives of the present study are to use these SSR markers to examine the genetic diversity and the population structure of the BVRC collection and to construct a core set from this collection with our long-term goal to explore the diversity in watermelon resources for modern watermelon breeding.
The present study was conducted in 2012–13. The 1197 watermelon accessions maintained by BVRC belong to seven species: C. naudinianus (3), C. ecirrhosus (2), C. rehmii (2), C. colocynthis (11), C. amarus (58), C. mucosospermus (25), and C. lanatus subsp. vulgaris (1096). Details of these accessions used in this study are presented in Supplemental Table 1. Among them, 760 were introduced from the Southern Regional Plant Introduction Station in Griffin, GA, and the remaining 437 were diploid cultivars or inbred lines from different watermelon breeding programs and seed companies in China.
For each accession (10 plants were used for each accession), genomic DNA was extracted from 50 mg of young freeze-dried leaves following the hexadecyltrimethylammonium bromide (CTAB) protocol (Levi and Thomas, 1999); quantified with a Nanodrop™ ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE) and diluted to 10 ng/μL for subsequent use.
The 23 highly polymorphic watermelon SSRs have been described in our previous study (Zhang et al., 2012), which were used for genotyping the 1197 watermelon accessions. Each 15 µL polymerase chain reaction (PCR) reaction mixture contained 20 ng template DNA, 4 µm each of the left and right primers, 2.5 mm MgCl2, 2 mm each of deoxynucleotide triphosphates, and 0.5 U Taq DNA polymerase in 1× PCR buffer (Takara Company, China). The PCR reaction started with 94 °C for 5 min followed by 35 cycles of 94 °C for 20 s, 55 °C for 20 s, and 72 °C for 90 s with a final extension at 72 °C for 8 min. The PCR products were analyzed using 6% polyacrylamide gel electrophoresis in 1× Tris borate ethylenediaminetetraacetic acid buffer. The gel was stained with silver staining using the SILVER SEQUENCE DNA Sequencing System (Promega, Madison, WI).
Genetic parameters such as the major allele frequency, the average number of alleles and PIC were estimated using the PowerMarker v3.25 software (Liu and Muse, 2005). Shannon’s information index (SW), Ho, expected heterozygosity (He), Nei and estimation of pairwise Fst among groups were calculated using the POPGENE v1.32 software (Yeh et al., 1997).
A model-based Bayesian clustering method was applied to infer genetic structure and define the number of groups in the dataset using the software STRUCTURE v.2.3 (Pritchard et al., 2000). To identify the putative number of groups (K), we ran an admixture and related frequency model from the value of K = 1 to 10 (five runs at each K). Each run comprised a burn-in length of 10,000 followed by 1,000,000 MCMC (Monte Carlo Markov Chain) replicates. The choice of most likely K value was performed by calculating the estimated log probability of data [LnP(D)] and an ad-hoc statistic Δk based on the rate of change in LnP(D) between successive K values, as described by Evanno et al. (2005). Based on the maximum membership probability for each individual, all accessions were assigned to corresponding groups, as described by Remington et al. (2001).
The advanced M strategy, based on a modified heuristic algorithm implemented in the PowerCore software by Kim et al. (2007), was used to develop the core set. In the software, genotype data at all loci were transformed automatically into a frequency table that covered the whole collection.
All 23 SSR markers amplified clear, easily scoreable, and polymorphic bands in the 1197 accessions, confirming the usefulness of these markers in distinguishing different watermelon accessions. In addition, the band reproducibility and consistency for control genotypes (20 Citrullus genotypes with diverse genetic backgrounds and horticultural traits, which have been resequenced in the Xu’s laboratory, BVRC (Guo et al., 2013) assured that the SSRs used were reliable genetic markers for Citrullus diversity analysis.
A total of 133 alleles were detected by the 23 SSRs ranging from 4 to 9 with an average of 6.05 alleles per marker locus. The PIC value ranged from 0.35 to 0.70, with an average of 0.49. Among the 23 markers, BVWS00433 exhibited the highest PIC value (0.70) and H value (0.74), whereas BVWS00314 had the lowest (H = 0.38; PIC = 0.35), which was the same as observed in our previous study (Zhang et al., 2012). In general, the PIC values indicate the allele diversity and frequency (Guo and Elston, 1999). The average PIC value of 0.49 for 23 markers from this study was comparable with the average PIC of 0.50 by Kwon et al. (2010) for 49 sweet commercial watermelon cultivars, a PIC value of 0.53 by Joobeur et al. (2006) with eight C. lanatus var. lanatus and four C. lanatus var. citroides accessions. Mujaju et al. (2011) reported that a very high PIC value of 0.92 from a study with 22 accessions from Southern Africa whereas a low PIC value of 0.35 was reported by Nantoumé et al. (2013) in 134 watermelon landrace accessions from Mali and the average PIC of 0.34 by Munisse et al. (2013) for 96 Mozambique accessions. Different PIC values could be attributed to the number of accessions included in the above studies, as well as the polymorphism of SSR markers used. In general, those studies showing low PIC estimates only included two cultivated species, C. lanatus subsp. vulgaris and C. mucosospermus. Our results reported herein from seven Citrullus species may present the average of PIC values in this genus.
Based on the data from 23 SSR markers on 1197 watermelon accessions, the Nei’s gene diversity index ranged from 0.38 to 0.74 (mean = 0.56). Shannon’s information index over the whole panel was relatively high (mean = 0.99). Compared with the He value across all accessions (mean = 0.56), the Ho value was relatively low (mean = 0.10) suggesting certain degree of inbreeding of these materials.
Using a model-based approach of STRUCTURE, the population structure of 1197 Citrullus accessions was analyzed. Fifty data sets were obtained by setting the number of possible clusters (K) from 1 to 10 with five replications each. The LnP(D) value for each given K increased continuously, but did not show an abrupt change; the optimal K value could not be inferred (Fig. 1A). But after applying the Evanno’s correction method (Evanno et al., 2005), there was a clear peak of ΔK at K = 2, indicating two major populations (Pop I and Pop II) in this panel of collections (Fig. 1B). Pop I contained 450 accessions, which predominantly belonged to C. lanatus subsp. vulgaris (446) and a few (4) belonged to C. mucosospermus. Pop II comprised 465 accessions, 379 belonged to C. lanatus subsp. vulgaris, and the rest 86 accessions belonged to C. naudinianus (3), C. ecirrhosus (2), C. rehmii (2), C. colocynthis (11), C. amarus (58), and C. mucosospermus (10). The remaining 282 accessions had membership probabilities lower than 0.75 in any group and were classified into an admixture group, which belonged to C. lanatus subsp. vulgaris (271) and C. mucosospermus (11) (Table 1; Supplemental Table 1).The two main populations, Pop I and Pop II, could be further subdivided into four subgroups. The ΔK values suggested K = 2 to both Pop I and Pop II. Pop I was divided into two subgroups: Pop IA and Pop IB. Pop IA was consisted of 238 accessions, a mix of American and East Asian ecotypes from C. lanatus subsp. vulgaris. Pop IB had 101 accessions, all from C. lanatus subsp. vulgaris East Asian ecotype. Pop I admixture included 111 accessionsof C. lanatus subsp. vulgaris with membership probabilities less than 0.75 in any of the two groups. Pop II was also classified into two subgroups. Pop IIA was a typical C. lanatus subsp. vulgaris American ecotype panel and contained 368 accessions, of which 357 were of C. lanatus subsp. vulgaris, 7 of C. mucosospermus, and 4 of C. amarus. Pop IIB contained 71 accessions, which was a wild species mixed subgroup, with the accessions belonging to the six species, involving C. naudinianus (3), C. ecirrhosus (2), C. rehmii (2), C. colocynthis (9), C. amarus (51), C. lanatus subsp. vulgaris (4). Pop II admixture group was consisted 26 accessions, which were from C. lanatus subsp. vulgaris (18), C. colocynthis (2), C. amarus (3), and C. mucosospermus (3) (Table 1; Fig. 2). Accessions in the admixture group may have partial ancestry in more than one background. They probably had a complex history involving intercrossing or perhaps resulting from the gene flow between taxa.
Citation: HortScience 51, 1; 10.21273/HORTSCI.51.1.23
Citation: HortScience 51, 1; 10.21273/HORTSCI.51.1.23
Overall, the population structure analysis identified two main populations and four subpopulations. Comparing Pop I and Pop II, molecular variance analysis revealed that only 14.0% of the total genetic variation was partitioned among groups, 70.0% within groups and 16.0% within accessions. Analysis with subgroups revealed that 24.0% of the variations occurred between subgroups and 60.0% within subgroups suggesting relatively moderate differentiation among subgroups.
The pairwise Fst values can reveal genetic distance populations (Table 2). It ranged from 0.074 between Pop IA (American and East Asian ecotype mixed) and Pop IIA (American ecotype) to 0.341 between Pop IB (East Asian ecotype) and Pop IIB (wild species mixed), suggesting that Pop IB and Pop IIB groups had the highest degree of differentiations, which was supported with genetic diversity estimates (Table 3). When compared with other subgroups, Pop IIB exhibited the highest average alleles per locus (5.64), H (0.56), and Ho (0.20) indicating highest genetic diversity within this subgroup. In contrast, Pop IB exhibited the least diversity in terms of average alleles per locus (2.50), H (0.30), and Ho (0.03). Since all accessions in Pop IB were cultivated species, this result clearly showed the high degree of genetic differentiation between wild and cultivated Citrullus species, which is in agreement with previous findings (Jarret et al., 1997; Joobeur et al., 2006; Levi et al., 2013; Mujaju et al., 2010).
It was interesting to note the distinct genetic backgrounds between American and East Asian ecotypes. The genetic parameters such as gene diversity, PIC, and heterozygosity were lowest in East Asian ecotypes (Pop IB), were higher in American and East Asian ecotypes mixed (Pop IA), and highest in American ecotypes (Pop IIA). However, this observation was inconsistent from results obtained by Nimmakayala et al. (2014) who placed these two ecotypes into a single group. The discrepancy may be due to the difference in sampling sizes used by the two studies. In our diversity panel, over 300 accessions were from China and the United Staes (Supplemental Table 1) including several intermediate types between East Asian types and American types, whereas Nimmakayala et al. (2014) used 130 cultivated forms of watermelon.
Among the three ecotypes of East Asian, American, American and East Asian ecotype mixed, the East Asian ecotype showed the narrowest genetic base. One explanation for this is the breeding effect. More breeding programs in China focus on limited watermelon varieties as breeding parents, such as ‘Charleston Gray’, ‘Jubilee’, ‘Crimson Sweet’, ‘Sugar Baby’, ‘Sugarlee’ etc., which caused the narrow genetic base.
Further clustering analysis of the 71 accessions in Pop IIB resulted in a multibranched dendrogram (Fig. 3). It displayed three major clusters that represented the C. naudinianus group (black group, CXG, PI618817, and PI 671961), the C. colocynthis group (green color) that also included the two C. rehmii accessions (Grif 16376 and PI632755, in pink color), C. amarus group (yellow color) including two accessions of C. ecirrhosus (PI632751 and Grif 16945, in blue color) and four accessions of C. lanatus subsp. vulgaris (PI 482288, PI505586, Seychelles and Colorado Preserving, in purple color) (Supplemental Table 1). From the dendrogram, it was clear that C. naudinianus (in black color) was distant from the other species, while C. rehmii (pink color) showed genetic similarity to the C. colocynthis group (green color) and C. ecirrhosus (blue color) was closer to the C. amarus group (yellow color). In Pop IIB population, the C. lanatus subsp. vulgaris group was composed of four accessions, which were phylogenetically closer to the C. amarus group than to the C. colocynthis group. These results were largely in agreement with the findings of Chomicki and Renner (2015) based on plastid and nuclear DNA data showing that the C. amarus and C. ecirrhosus may belong to the same group, while C. naudinianus was far from the other Citrullus species.
Citation: HortScience 51, 1; 10.21273/HORTSCI.51.1.23
Citation: HortScience 51, 1; 10.21273/HORTSCI.51.1.23
The phylogenetic relationships can also be reflected by the sexual compatibility in interspecific hybridization. Several reports have noted that crosses among the three Citrullus species (C. lanatus, C. colocynthis, and C. ecirrhosus) produced viable F1 hybrids while crosses between Praecitrullus fistulosus (2n = 2x = 24) and Citrullus species did not yield fruits (Navot and Zamir, 1987). So far, there is practically no information about the magnitude of the gene flow among species and accessions of Citrullus. Our attempts to cross C. naudinianus with other species were unsuccessful confirming that C. naudinianus is genetically distant from other Citrullus species. Thus, it might be the time to reconsider the sister species status of C. naudinianus with other species in Citrullus.
Based on morphological descriptors and random selection, the USDA and Germplasm Resources Information Network had developed 251 accessions as core set (<http://www.ars-grin.gov/cgi-bin/npgs/>), and the 84 watermelon from the USDA core collection we analyzed. In this study, SSR genotypic data were used to construct a core set of 130 from the 1197 accessions by using the software PowerCore. These 130 accessions were from seven Citrullus species, and 11 of them were present in the 84-accession core set designated by the USDA. Although this core set only accounted for 10.9% of the BVRC watermelon collection, it had captured all alleles detected in the collection. The Nei’s gene diversity index (Nei = 0.65) and Shannon’s information index (SW = 1.25) in this core set were higher than the original 1197 set (H = 0.56; I = 0.99) or the USDA 84-collection core set (H = 0.48; I = 0.92).The USDA core set included four accessions from Pop I, 42 from Pop II, and 38 from the admixture group that were from three species: C. lanatus subsp. vulgaris (77), C. amarus (5), and C. mucosospermus (2), which seems skewed toward Pop II and the admixture groups. In contrast, the core set we constructed herein had a relatively more balanced representation including 47 accessions from Pop I, 68 from Pop II, and 15 from the admixture group, which covered all seven species in Citrullus indicating greater genetic diversity of the 130 core set (Fig. 4).
Liang et al. (2015) selected 55 apple genotypes as the core collection and noted that, even though the core collection resulted in the most suitable based on the M-strategy, different subpopulations of the same size was able to retain the same amount of genetic variation. Since molecular markers are largely neutral in selection, maximum marker allele diversity does not necessarily mean diversity for traits that are more useful for plant breeding. For this reason, additional criteria can be used for the definition of the most appropriate subset of accessions representative of the Citrullus germplasm, which may include phenotypic and resistance traits, as well as historical or economically important target traits. In our 130 line core collection, accessions from C. naudinianus, C. ecirrhosus, and C. rehmii are underrepresented. Hence, future explorations in these species can possibly add novel alleles for disease resistances or abiotic stress tolerances, which are important for watermelon breeding.
In summary, the core set identified in this study is very useful for watermelon breeders. It permits a better organization of the crop’s gene pool management, more efficient sampling of the available germplasm resources, and better access to useful genetic variation for breeders.
Optimal value of K for the full germplasm set of 1197 accesions. Fifty data sets were obtained by setting the number of possible clusters (K) from 1 to 10 with five replications each. The log probability of data [LnP(D)] value for each given K increased continuously, but did not show an abrupt change; the optimal K value could not be inferred (A). But after applying the Evanno’s correction method (Evanno et al., 2005), there was a clear peak of ΔK at K = 2, indicating two major populations (Pop I and Pop II) in this panel of collections (B).
Model-based cluster membership of 1197 Citrullus accessions in two main populations (Pop I and Pop II) and four subpopulations (Pop IA, Pop IB, Pop IIA, and Pop IIB). Pop IA was consisted of 238 accessions, a mix of American and East Asian ecotypes. Pop IB had 101 accessions, all from East Asian ecotype. Pop IIA was a typical American ecotype panel and contained 368 accessions. Pop IIB contained 71 accessions, which was a wild species mixed subgroup, with the accessions belonging to the six species.
A dendrogram of 71 accessions in Pop IIB, displays three major clusters. The groups represent Citrullus naudinianus group [CXG, PI618817, and PI 671961 (black)], Citrullus colocynthis group (green) that also included the two Citrullus rehmii accessions [Grif 16376 and PI 632755 (pink)], Citrullus amarus group (yellow) that also includes the two accession of Citrullus ecirrhosus [PI 632751 and Grif 16945 (blue)], and the four accession of Citrullus lanatus subsp. vulgaris group [PI 482288, PI 505586, Seychelles and Colorado Preserving (purple color)].
Distribution of Nei’s gene diversity index (Nei) and Shannon’s information index (SW) among 23 SSR markers in the core set (C-Nei; C-SW) and the entire collection (E-Nei; E-SW), compared with 84 American core set (AC-Nei; AC-SW). Nei’s gene diversity index and SW were calculated using the POPGENE v1.32 software (Yeh et al., 1997). The entire collection included 1197 accessions, a core set of 130 accessions was developed from the entire collection, and 84 watermelon from the U.S. Department of Agriculture core collection we analyzed.
Contributor Notes
This study was supported by funds from the Ministry of Science and Technology of China (31171980, 31361140355, 31401893, and 2013BAD01B04); Ministry of Agriculture of China (CARS-26); Beijing Municipal Science and Technology Commission, China (Z121105002512037, Z121105002612013); Beijing Natural Science Foundation (6141001, 6144023); and Beijing Excellent Talents Programme (2014000021223TD03, 2013D002020000003).
Corresponding author. E-mail: xuyong@nercv.org.
Optimal value of K for the full germplasm set of 1197 accesions. Fifty data sets were obtained by setting the number of possible clusters (K) from 1 to 10 with five replications each. The log probability of data [LnP(D)] value for each given K increased continuously, but did not show an abrupt change; the optimal K value could not be inferred (A). But after applying the Evanno’s correction method (Evanno et al., 2005), there was a clear peak of ΔK at K = 2, indicating two major populations (Pop I and Pop II) in this panel of collections (B).
Model-based cluster membership of 1197 Citrullus accessions in two main populations (Pop I and Pop II) and four subpopulations (Pop IA, Pop IB, Pop IIA, and Pop IIB). Pop IA was consisted of 238 accessions, a mix of American and East Asian ecotypes. Pop IB had 101 accessions, all from East Asian ecotype. Pop IIA was a typical American ecotype panel and contained 368 accessions. Pop IIB contained 71 accessions, which was a wild species mixed subgroup, with the accessions belonging to the six species.
A dendrogram of 71 accessions in Pop IIB, displays three major clusters. The groups represent Citrullus naudinianus group [CXG, PI618817, and PI 671961 (black)], Citrullus colocynthis group (green) that also included the two Citrullus rehmii accessions [Grif 16376 and PI 632755 (pink)], Citrullus amarus group (yellow) that also includes the two accession of Citrullus ecirrhosus [PI 632751 and Grif 16945 (blue)], and the four accession of Citrullus lanatus subsp. vulgaris group [PI 482288, PI 505586, Seychelles and Colorado Preserving (purple color)].
Distribution of Nei’s gene diversity index (Nei) and Shannon’s information index (SW) among 23 SSR markers in the core set (C-Nei; C-SW) and the entire collection (E-Nei; E-SW), compared with 84 American core set (AC-Nei; AC-SW). Nei’s gene diversity index and SW were calculated using the POPGENE v1.32 software (Yeh et al., 1997). The entire collection included 1197 accessions, a core set of 130 accessions was developed from the entire collection, and 84 watermelon from the U.S. Department of Agriculture core collection we analyzed.
