Genetic Diversity Analysis and Construction of a Core Germplasm Resource Bank of Xinjiang’s Indigenous Cultivated Grapes

Abstract

This study systematically investigated the genetic characteristics and germplasm conservation strategies of Xinjiang Thompson Seedless grapes (Vitis vinifera Thompson Seedless) and Munake grapes (Vitis vinifera L. cv. Munake) using SSR molecular markers and whole-genome resequencing technology. A genetic diversity analysis of 165 Thompson Seedless accessions with 16 SSR markers identified 442 alleles (27.63 alleles per locus on average), with the expected heterozygosity (He = 0.76) and observed heterozygosity (Ho = 0.83) indicating moderate-to-low genetic diversity. A molecular variance analysis (AMOVA) further revealed that 96% of the genetic variation originated within populations, with minimal inter-population differentiation (Fst = 0.04). Phylogenetic reconstruction using SSR markers demonstrated significant genealogical associations between the two cultivars, supporting the hypothesis that Thompson Seedless may have originated from a domesticated lineage of Munake. The selected SSR markers exhibited high discriminatory power (PIC = 0.92), enabling the precise differentiation of accessions with closely related genetic backgrounds. Whole-genome resequencing identified 20,074,046 and 69,214,080 high-quality SNPs in 100 Thompson Seedless and 141 Munake accessions, respectively. Core germplasm banks were subsequently established: the Thompson Seedless core collection (25 accessions) captured 94% of genetic variation, reflecting genetic homogenization driven by intensive clonal cultivation, while the Munake core collection (42 accessions) retained 95% of allelic diversity and resolved regional synonym issues through highly polymorphic SNP markers. A comparative analysis revealed that Munake maintains higher genetic diversity due to natural gene flow, whereas Thompson Seedless faces heightened risks of genetic erosion from prolonged asexual propagation. These findings provide a theoretical foundation and technical framework for precise conservation, varietal improvement, and sustainable utilization of grape germplasm resources in Xinjiang.

1. Introduction

Grapes (Vitis vinifera L., 2n = 38) belong to the genus Vitis of the Vitaceae family and are woody, climbing plants with a multitude of varieties distributed worldwide, and a genome size of around 500 Mbp [1]. In terms of their cultivation area and production volume, grapes are second only to citrus, making them the second most important fruit globally [2]. The northwestern region of China, particularly Xinjiang, has a geographical environment that is particularly suitable for grape cultivation. Xinjiang has a diverse range of local grape varieties [3]. These include Thompson Seedless and Munake, which have a long history of cultivation and exhibit large planting areas and yields in Xinjiang.

Thompson Seedless grapes were disseminated to West Asia and Central Asia via the Silk Road and were subsequently introduced to the Xinjiang region in China. Their cultivation dates back to the 3rd century AD, and they have been extensively adopted in the Hetian area of Southern Xinjiang [4,5]. Turpan, Xinjiang, is the primary production area for Thompson Seedless grapes in China. The total grape planting area in China spans 56.74 million mu (approximately 3.78 million hectares), with approximately 90% of the total output dedicated to drying [6]. However, the prolonged and intensive cultivation of Thompson Seedless grapes, along with their asexual reproduction, have significantly reduced the genetic diversity of this cultivar, jeopardizing germplasm resources. This reduction affects both their yields and quality, as well as weakens pest and disease resistance, threatening the sustainable development of the grape industry [7].

Munake is an ancient grape variety that has been grown for thousands of years in Xinjiang, China, and is renowned for its exceptional fruit characteristics. The origin of the Munake grape, an ancient variety endemic to southern Xinjiang, combines local natural selection with historical cultivation practice [8,9,10]. They have a narrow distribution, adapted only to Xinjiang’s arid areas, which feature high accumulated temperatures, and are mainly cultivated in southern Xinjiang such as in Atushi and partly along Gansu’s Hexi Corridor [11]. Attempts to grow Munake grapes in other Chinese provinces have failed due to the crop’s strict climatic requirements (220+ frost-free days, <120 mm annual rainfall, 12–13 °C average temperature, and 4200+ °C accumulated temperature) [12]. With its desirable qualities but strict habitat requirements, Munake is specifically endemic to Xinjiang, which poses transportation and storage difficulties [13]. Its strict ecological adaptation makes it a typical case for studying crop domestication in arid zones, while genetic diversity analysis provides a key basis for subsequent resource conservation and molecular breeding.

Molecular markers have revolutionized germplasm characterization through applications in DNA fingerprinting, genetic diversity assessment, and core collection development [14,15,16,17,18]. Based on the above technical advantages, the complementary combination of SNP and SSR markers was selected in this study to address the limitations of traditional methods in assessing the genetic diversity of grapevines in Xinjiang [19]. SNPs provide genome-wide coverage with high reproducibility for population structure analysis [20,21,22,23,24]. SSRs offer cost-effective polymorphism detection validated in grape varietal identification [25,26,27,28,29,30,31,32,33,34]. While previous studies have extensively applied these markers to Eurasian grapevines [35,36,37], their utilization in characterizing Xinjiang’s endemic cultivars remains limited. Our integrated approach addresses this gap by combining SNP-based genome scanning with SSR fingerprinting tailored to local germplasm traits.

The diversity of grape germplasm resources is essential for sustainable agricultural development. European and American countries are more advanced in collecting and using grape germplasms, while China is lagging behind, relying mainly on traditional breeding techniques [38,39,40]. Xinjiang is located in the western region of China and is a primary production area for high-quality grapes. A clear analysis of the population’s genetic structure is a primary prerequisite for germplasm safety and utilization, forming the foundation of germplasm innovation and serving as an essential guarantee for the sustainable development of both germplasm resources and the industry [41,42]. The core germplasm is an integral part of crop conservation and innovation, covering a wide range of morphological traits; reflecting the geographical, genetic, and genotypic details of the species; and playing a crucial role in the sharing, application, and maintenance of gene banks, which have both academic and practical value.

This study integrates whole-genome resequencing and molecular marker techniques to systematically evaluate the genetic diversity levels and genetic structure characteristics of Xinjiang’s widely cultivated Thompson Seedless grape and its local endemic counterpart, Munake, while establishing a core germplasm resource bank. By considering the dynamic changes in germplasm resources under long-term cultivation practices, we further explore the genetic relationships and evolutionary patterns between the two varieties. The findings aim to provide scientific evidence for the optimization of conservation strategies and the precise identification and genetic improvement of Xinjiang grape germplasms, ultimately supporting the sustainable development and utilization of local grape resources.

2. Materials and Methods

2.1. Plant Material and Sampling

This study focused on two grape cultivars: Thompson Seedless and Munake. The Thompson Seedless samples were systematically collected from three distinct plantation types in the Turpan Basin: (1) Turpan Desert Botanical Garden (ex situ conservation plot), (2) Turpan Grape Valley (traditional trellis cultivation vineyard), and (3) Shanshan region (farmers’ cultivation). A total of 165 Thompson Seedless grape samples were collected from three regions for SSR-based genetic structure and genetic diversity analysis. Subsequently, 100 high-quality samples were selected from the Thompson Seedless grape collection based on preservation quality for resequencing analysis. The Munake SSR analysis data were obtained from the research team’s previous study data [43]. A total of 141 Munake grape accessions were collected for resequencing analysis, comprising 134 accessions from Xinjiang, China, 6 accessions from Baku, Republic of Azerbaijan, and 1 accession from Henan Province, China (Figure 1 and Supplementary File S1). The primary sampling sites for Munake grapes were located in Artux, Kashgar, and Hetian in southern Xinjiang, China, supplemented by samples from the aforementioned international and domestic regions. These vines were free of pests and diseases and were cultivated under ideal growth conditions. At least 3 leaves were collected from each plant. We implemented a stratified random sampling design, with vineyard parcels stratified by 10–20 m inter-parcel distances, to collect representative accessions. It should be noted that the term “population” in this study refers to accessions grouped by geographic planting sites for analytical purposes, rather than naturally reproducing biological populations. This grouping strategy facilitates the application of population genetics models to cultivated germplasms.

2.2. Genetic Diversity Analysis Based on SSR Molecular Markers

2.2.1. Genotyping and DNA Extraction

The collected leaves were dried with silica gel and then ground in liquid nitrogen. Subsequently, DNA was extracted and used as a template for polymerase chain reaction (PCR) amplification. The 16 pairs of SSR primers selected for this study (

Table 1. The 16 grape SSR primer pairs and base sequences.

Num	Name	Primer Sequence (F) (5′-3′)	Primer Sequence (R) (5′-3′)	MW (Dalton)	TM (°C)
1	VRZAG67	ACCTGGCCCGACTCCTCTTGTATGC	TCCTGCCGGCGATAACCAAGCTATG	7627	65
2	VCHR13a	TGGCAGAGCAAATGAATCAA	TTGGATGGATTGGAATGACC	6212	55
3	VVIP31	TATCCAAGAGACAAATTCCCAC	TTCTCTTGTTTCCTGCAAATGG	6682	56
4	VVMD7	AGAGTTGCGGAGAACAGGAT	CGAACCTTCACACGCTTGAT	6037	57
5	VMC4F3-1	AAAGCACTATGGTGGGTGTAAA	TAACCAATACATGCATCAAGGA	6720	56
6	Vchr8a	ACCCACTGCCACTCTCTCAT	AAATCTCCGGGATCCTTTTG	6083	59
7	VrZAG29	ATAACCAGGACAAGTTATTCAAGCC	ACCCAATTGACCATCTTTTATGCTG	7568	56
8	VrZAG25	CTCCACTTCACATCACATGGCATGC	CGGCCAACATTTACTCATCTCTCCC	7474	60
9	VVIb66	CCACTAGTGGTCAGAAAAGAAG	TTGTATTGTGTGCCTCTTCTCA	6674	53
10	VVIb01	TGACCCTCGACCTTAAATCTT	TGGTGAGTGCAATGATAGTAGA	6879	52
11	VVMD36	TAAAATAATAATAGGGGGACACGGG	CAACTGTAAAGGTAAGACACAGTCC	7769	56
12	VVS4	CCATCAGTGATAAAACCTAATGCC	CCCACCTTGCCCTTAGATGTTA	6622	55
13	VMC4C6	CTCCATCCCTATCTCATCAG	CTCTAACACCCAATCTCACA	5951	50
14	VVIi51	ATCCCAAGAGAACCAAGAAACT	GCTGATCTCAGTGCATATGTTG	6757	53
15	UDV-015	TGCACATTTCCCTCCTTAG	CGGGTTACTGGGAAGGGTAT	6254	54
16	VMC4H6	GTATAGAACCACGCATCCAACA	CCCTTAGTTTCCTCGTGCTTTT	6610	54

Note: The table provides the primer names and sequences for both the forward (F) and reverse (R) primers, as well as the temperature conditions and molecular weight sizes.

) were chosen based on their widespread applicability and high polymorphism in grape genetic diversity studies. These markers have been extensively validated in prior studies for applications such as a population structure analysis, germplasm characterization, and phylogenetic investigations [44,45,46,47,48]. All primers were experimentally optimized for annealing conditions and polymorphism parameters to ensure compatibility with Xinjiang grape germplasms. Both the forward and reverse primers were commercially synthesized (Shanghai General Biotechnology, Shanghai, China). PCR amplification was performed in a 25 µL reaction mixture. The mixture contained 1 µL (20 ng/µL) of template DNA, 12.5 µL (5 U/µL) of Premix Taq™, 1 µL (10 µmol/L) of each forward and reverse primer, and 9.5 µL of dd H₂O. An Applied Biosystems Verit96-Well thermal cycler was used to perform the PCR (Applied Biosystems Verit^TM96-Well thermal cycler, Thermo Fisher Scientific, Waltham, MA, USA). For further details, refer to Table 1.

PCR was performed in a thermocycler. First, a single denaturation was performed at 93 °C for 5 min. This was followed by 30 cycles, each cycle including denaturation at 94 °C for 30 s, annealing for 30 s at temperatures determined by the specific Tm values of each primer set, and extension at 65 °C for 90 s. The PCR products were detected with 6-carboxyfluorescein (FAM) and an automated fluorescence sequencing analyzer (Applied Biosystems^® 3130 Genetic Analyzers, Thermo Fisher Scientific, Waltham, MA, USA). The Genemapper version 3.0 software package was used to analyze the amplified fragment sizes of different samples at each SSR locus.

2.2.2. Genetic Structure and Genetic Diversity Analysis

We calculated genetic diversity indices such as the number of alleles (Na), effective number of alleles (Ne), Shannon’s information index (I), observed heterozygosity (Ho), and expected heterozygosity (He) using the GenAlEx 6.5 software package [49]. The polymorphic information content (PIC) was computed via the PowerMarker v3.25 [50] software program, and Nei’s genetic diversity index was determined using Popgene 32 [51]. A principal coordinate analysis (PCoA) was also conducted using GenAlEx 6.5. The genetic distance (GD) matrix for populations was generated through the distance-based module in GenAlEx 6.5. Subsequently, a molecular analysis of variance (AMOVA) was employed to assess the contributions of genetic variation both among and within populations. Bayesian assignment testing was performed using the STRUCTURE 2.3.4 software package [52]. Additionally, a cluster analysis was conducted via the unweighted pair group method with arithmetic mean (UPGMA) using the Mega 11 software program [53]. For the analysis in STRUCTURE 2.3.4, parameters were set with K values ranging from 1 to 10, and each K value was replicated 10 times. The burn-in period was set to 1 × 10⁴, the number of Markov Chain Monte Carlo (MCMC) replicates after the burn-in was 1 × 10⁵, and a mixed model was applied. Structure Harvester was utilized to identify the optimal ∆K value from the numerical results corresponding to each run (saved in the Results folder).

2.3. Construction of a Core Germplasm Bank Based on Whole-Genome Resequencing

2.3.1. Sequencing and Sequencing Quality Control

The DNA secure Plant Kit was employed to extract DNA from plant tissues (Tiangen Biotech (Beijing) Co., Ltd., Beijing, China, https://www.tiangen.com/content/details_40_21522.html (accessed on 10 November 2024)). The amplified products were separated by agarose gel electrophoresis, purified by using Universal DNA Purification Kit (Tiangen, https://www.tiangen.com/content/details_40_21348.html (accessed on 28 January 2025)). The qualified DNA samples were randomly fragmented into 350 bp segments using a Covaris crusher. We used either the TruSeq Library Construction Kit (Illumina, San Diego, CA, USA) or the MGI Easy FS DNA Prep kit (MGI Tech Co., Ltd., Guangdong, China) to construct the library. After library construction, we first conducted a preliminary quantification using Qubit2.0 [54]. The library was then diluted to a concentration of 1 ng/µL, and the insert size of the library was detected with an Agilent 2100. After the library passed the quality check, different libraries were pooled according to the requirements of effective concentration and target data volume. Finally, we performed sequencing in the PE150 mode using either the Illumina NovaSeq6000 or DNBSEQT7 sequencing platform [55]. The sequencing data were aligned to the respective reference genomes of the two grape cultivars for variant detection (Thompson Seedless: https://db.cngb.org/search/project/CNP0004225/ (accessed on 24 May 2024); Munake: http://ftp.ensemblgenomes.org/pub/plants/release-25/fasta/vitis_vinifera/ (accessed on 22 April 2022)).

2.3.2. SNP/InDel Detection and Annotation

The SNP and InDel detection was performed using Sentieon (https://pubmed.ncbi.nlm.nih.gov/32698196/ (accessed on 24 May 2024) [56]. Subsequently, the SNP and InDel sites were annotated with the ANNOVAR (http://www.openbioinformatics.org/annovar/ (accessed on 24 May 2024)) [57] software to determine the genomic region of variant sites and their mutation types. To reduce the error rate of SNP/InDel detection, the following criteria were selected for filtering: (1) the number of reads supported by SNPs/InDels was not less than 4; (2) the RMS mapping quality (MQ) of site alignment must not be less than 40; and (3) the genotype quality (GQ) value should not be lower than 5. A gene ontology (GO) enrichment analysis was performed using the ShinyGO v0.61 tool. The frequency of query genes was compared with the complete reference genome for V. vinifera, searching for possible enrichment in biological processes [58].

2.3.3. Genetic Diversity and Population Structure Analysis

Cross-validated unsupervised clustering was further conducted using ADMIXTURE v1.3.0 [59]. The consistency of genetic groupings was validated through a Discriminant Analysis of Principal Components (DAPC) implemented in the R package “adegenet” [60]. A principal component analysis (PCA) was performed on numerically encoded SNP data (e.g., T/C as 0/1) using TASSEL v5 [61]. Phylogenetic relationships were reconstructed via a neighbor-joining tree with 1000 bootstrap replicates in MEGA, visualized and annotated in iTOL v4. Finally, a core germplasm collection was established with Core Hunter II. The collection was validated for genetic diversity (Shannon’s I, Nei’s index) and uniqueness through QR-encoded SNP fingerprints derived from whole-genome SNP data.

2.3.4. Core Germplasm Collection of Grapes

Based on the SNP markers’ co-consensuses [62], as well as their distribution across chromosomes, we selected a high-quality and discriminatory set of core SNP markers based on the even distribution of SNPs per chromosome and the principle of a minimum number of SNPs representing the maximum genetic diversity [63], and a total of 353 SNP markers were screened and used for the subsequent analysis. To encode all the genotypic data and supply a 2D barcode fingerprint for every core, we used the online software Caoliaoerweima (http://cli.im/ (accessed on 20 May 2024)). Each 2D fingerprint can supply a variety of information, including the variety name, type, and genotypical data, and can be accessed on both computers and mobile devices. Based on several genetic variables, we used the Core Hunter II software (http://www.corehunter.org (accessed on 20 October 2024)) to create a minimally repetitive core germplasm collection which is representative of the entire accession base.

3. Results

3.1. Mutation Detection in Thompson Seedless Grapes

We performed whole-genome resequencing on 100 Thompson seedless grape samples collected from three different regions. After thorough filtering of the sequencing data, we obtained high-quality clean data, covering various metrics such as sequencing data yield, sequencing error rate, Q20 and Q30 content, GC content, and others. This sequencing project, involving 100 samples, produced a significant amount of high-quality data, amounting to 454 Gb in total. The average effectiveness rate was 98.66%. The average percentage of bases with mass values ≥ 20 (Q20) was 97.99%, and the average percentage of bases with mass values ≥ 30 (Q30) was 94.83%. The average GC content was also determined to be 36.01% (for detailed data, see Supplementary Table S1). Regarding sequencing depth and coverage, the average depth across the 100 samples was 12.32. The average coverage at 1× was 96.94%, and the average coverage at 4× was 88.36%. It is worth noting that over 96.82% of the sequenced reads from the grape samples matched the reference genome (for more details, refer to Supplementary Table S2).

Our investigation into SNP detection identified a total of 20,074,046 SNP variant sites. Specifically, 811,303 SNPs were situated upstream, 744,937 SNPs were downstream, and 13,712,792 SNPs were intergenic. Moreover, 73,157 SNPs were detected within the 1 Kb regions flanking genes (both upstream and downstream). Within the exonic regions, there were 555,982 variants, comprising 10,355 stop gain, 1074 stop loss, 223,171 synonymous, and 321,382 non-synonymous variants. Additionally, 3,573,297 SNPs were intronic, 3889 were splicing-related, and 13,712,792 were intergenic. There were 13,892,353 SNP variant loci of the transitions type and 6,181,693 SNP variant loci of the transversions type (for detailed data, see Supplementary Table S3). The distribution of these SNP sites across the chromosomes is depicted in Figure 2A and Supplementary Figure S1, with denser regions shaded darker and less dense regions shaded lighter.

Our analysis of InDel detection uncovered a total of 3,043,093 InDel variant sites. Specifically, 203,638 InDels were identified in upstream regions, 175,996 in downstream regions, and 1,832,021 in intergenic regions. In addition, 17,757 InDels were detected within the 1 Kb flanking regions of genes (both upstream and downstream). Within the exonic regions, there were 533,460 variants, including 1437 stop gain, 99 stop loss, 12,244 frameshift deletions, and 5457 non-frameshift deletions. Furthermore, 576,103 InDels were found in intronic regions and 1036 in splicing regions (for detailed data, see Supplementary Table S4). The distribution of these InDel sites across the chromosomes is illustrated in Figure 2B and Supplementary Figure S2, with denser regions shaded darker and less dense regions shaded lighter. The findings indicate that the majority of InDels are concentrated in the upstream and downstream regions of genes, as well as in intronic regions.

To comprehensively understand gene functions, we performed a gene ontology (GO) analysis, categorizing genes according to three main aspects: biological processes, cellular components, and molecular functions (Figure 3). For biological processes, the GO analysis revealed that genes are engaged in a wide range of crucial biological activities. This encompasses basic cellular functions like metabolism, cell cycle control, and signal transduction, along with more specialized functions such as the immune response, development, and neuronal communication. Regarding cellular components, the GO analysis offered valuable insights into the location and arrangement of gene-derived products within cells. In terms of molecular function, the genes related to nucleoside-triphosphatase activity, hydrolase activity, and those acting on ester bonds are more frequent. In the cellular component, genes related to membrane-bounded organelle and the cytoplasmic part are more frequent. In the biological process, the genes related to the response to stimulus, small molecule metabolic process are more frequent.

To further optimize germplasm resource management and provide enhanced data support for grape breeding, we conducted core germplasm screening for Thompson Seedless grapes based on resequencing data, establishing a core germplasm bank for this cultivar. Five rounds of gradient sampling were performed randomly from the population, and the coverage value (CV) was calculated for each round. The relationship between the core germplasm size and allelic coverage rate was plotted. Using a CV threshold of 94%, 25 accessions were ultimately selected as the core germplasm, forming the core germplasm bank (Figure 4).

3.2. Genetic Diversity of Thompson Seedless

Using 16 SSR loci, the genetic variation in 165 grape samples was estimated. Ultimately, we identified 442 alleles across the 165 grape samples, with an average of 27.63 alleles per SSR marker.

The number of alleles (Na) ranged from 21 (for VVMD7, VMC4F3-1, and VVIb01) to 39 (for UDV-015). The average number of effective alleles (Ne) across the 16 SSR loci was 5.41, with the maximum value of 13.72 at the UDV-015F locus and a minimum of 1.95 at the VVMD7 locus. The observed heterozygosity (Ho) ranged from a maximum of 1 (for VVIP31 and Vchr8a) to a minimum of 0.24 (for VVS4), with an average of 0.83. The expected heterozygosity (He) ranged from a maximum of 0.93 (for UDV-015) to a minimum of 0.49 (for VVMD7), with an average of 0.76. Among the 16 primer pairs screened, the Ne, Ho, and He values showed good polymorphism. The polymorphic information content (PIC) varied between 0.83 and 0.98, with an average of 0.92. As many as 16 SSR markers had a PIC value greater than 0.50, indicating a high level of polymorphism. The Shannon information index (I) had a maximum value of 2.98 (for UDV-015F) and a minimum of 1.21 (for VVMD7F). The complete genetic diversity values are presented in

Table 2. Genetic data of 16 SSR loci from 165 Thompson Seedless individuals. This table encompasses the names, Na and Ne values, Shannon’s I, Ho, and other polymorphic information for 16 SSR loci.

Locus	Na	Ne	I	Ho	He	uHe	F	Fis	Fst	Nm
VVIb01	23	3.46	1.70	0.97	0.71	0.71	−0.36	−0.40	0.04	6.55
VVIb66	31	6.03	2.25	0.83	0.83	0.84	0.01	0.17	0.12	1.87
VVMD36	34	6.48	2.38	0.90	0.85	0.85	−0.07	−0.17	0.11	2.04
VVS4	31	2.83	1.63	0.24	0.65	0.65	0.62	0.41	0.24	0.78
VrZAG29	32	3.80	1.94	0.85	0.74	0.74	−0.16	−0.15	0.09	2.43
VrZAG25	25	6.29	2.29	0.95	0.84	0.85	−0.13	−0.17	0.09	2.59
VMC4C6	26	4.77	2.07	0.88	0.79	0.79	−0.11	−0.10	0.07	3.14
VVIi51	30	8.11	2.43	0.92	0.88	0.88	−0.05	−0.22	0.16	1.36
UDV-015	39	13.72	2.98	0.85	0.93	0.93	0.08	0.04	0.08	2.77
VMC4H6F	28	6.98	2.32	0.90	0.86	0.86	−0.05	−0.26	0.16	1.32
VRZAG67	19	2.58	1.29	0.77	0.61	0.61	−0.26	−0.31	0.03	8.68
VCHR13a	23	2.99	1.44	0.99	0.67	0.67	−0.49	−0.53	0.05	4.71
VVIP31	26	5.25	1.99	1.00	0.81	0.81	−0.24	−0.26	0.02	10.19
VVMD7	21	1.95	1.21	0.28	0.49	0.49	0.43	0.37	0.03	7.55
VMC4F3-1	22	3.50	1.67	0.99	0.71	0.72	−0.39	−0.43	0.07	15.39
Vchr8a	32	7.75	2.45	1.00	0.87	0.87	−0.15	−0.19	0.04	6.62
Mean	27.6	5.41	2.00	0.83	0.76	0.77	−0.08	−0.14	0.08	1.01

Note: Na = number of different alleles; Ne = effective number of alleles; Ho = observed heterozygosity; He = expected heterozygosity; uHe = unbiased expected heterozygosity; I = Shannon’s information index; F (Null) = null allele; Fis = inbreeding coefficient; and Fst = proportion of differentiation.

.

The effective number of alleles (Ne) averaged 3.77, 6.07, and 4.31 for the three populations. The population from TGV exhibited the highest Shannon information index (I), with an average of 1.92, indicating the richest genetic diversity. Conversely, the TDBG showed the lowest I value of 1.48, suggesting potentially lower genetic diversity. This information is instrumental for researchers in guiding future germplasm protection efforts. The observed heterozygosity (Ho) and expected heterozygosity (He) confirmed the higher genetic diversity of the TGV population, while the lower Ho and He values in the TDBG implied lower genetic diversity and a higher level of inbreeding. The observed negative fixation (F) indices were consistent across the three regions. The values of the genetic diversity parameters among the three populations are shown in

Table 3. Data on genetic diversity parameters for three populations.

Pop		N	Na	Ne	I	Ho	He	uHe	F
TDBG	Mean	65.56	10.43	3.77	1.48	0.84	0.68	0.68	−0.20
	SE	1.93	1.08	0.38	0.12	0.07	0.04	0.04	0.10
TGV	Mean	23.38	13.25	6.07	1.92	0.84	0.77	0.78	−0.13
	SE	1.56	1.79	0.89	0.18	0.05	0.04	0.04	0.09
SS	Mean	36	12.31	4.31	1.67	0.79	0.71	0.72	−0.11
	SE	2.27	1.22	0.61	0.12	0.06	0.03	0.03	0.08

Note: This table includes the names of three populations of Thompson Seedless grapes, along with their polymorphism data. Na = number of different alleles; N = number of individuals sampled; Ne = effective number of alleles; I = Shannon’s information index; and HO = observed heterozygosity. Pop stands for the name of the three populations: TDBG corresponds to the Turpan Desert Botanical Garden population, TGV to the Turpan Grape Valley population, and SS to the ShanShan population.

.

The quantitative analysis of genetic differentiation, conducted through fixation indices (FSTs) and gene flow (Nm) estimates, further illuminated the genetic connections between the populations. In particular, the high Nm values at the VMC4F3-1, VVIP31, VRZAG67, and VVIb01 loci indicated significant gene exchange, with these loci playing a pivotal role in genetic interchange.

Based on the genetic distance matrix, the principal coordinate analysis (PCoA) indicated that the first, second, and third axes accounted for 9.05%, 3.42%, and 3.08% of the SSR variation, respectively (Figure 5). An analysis of variance (ANOVA) of the molecular variance showed that 96% of the variance came from within populations and 4% from among populations. In the three Thompson Seedless populations in Xinjiang, it was observed that there was a similar genetic structure between populations 2 and 3, and there was a low level of similarity between the genetic structures of populations 1, 2, and 3.

3.3. Whole-Genome Resequencing Elucidates Genetic Diversity and Core Collection of Munake Grape Cultivars

For preliminary genome-wide SNP analysis, the 141 collected Munake grape accessions were resequenced. Many SNPs (69,214,080) were generated from the resequencing of 141 grape cultivar accessions, which were then subjected to stringent quality control to produce a set of clean reads. Quality scores were assigned to the data, with 89.03% of reads having a score of Q30, and 95.21% of reads having a score of Q20, with a base error rate of <1%. Overall, the clean data showed a normal GC distribution and a high sequencing quality.

An ADMIXTURE analysis showed that the cross-validation error rate was smallest at K = 4 (Figure 6A), indicating a tendency for the entire sample to cluster into four groups, independent of their origin or collection site (Figure 6B).

According to the distribution of SNP markers across the reference genome, we identified 353 SNP sites distributed. The minor allele frequency (MAF) of these SNPs ranged from 0.340 to 0.50 (average = 0.418), with observed heterozygosity (Ho) values between 0.459 and 0.621 (average = 0.568). Genetic diversity ranged from 0.454 to 0.502 (average = 0.485), and polymorphism information content (PIC) values between 0.350 and 0.375 accounted for over 75% of the loci (

Table 4. The evolution of genetic diversity based on the 353 core SNPs selected.

Index	Total	Min/Max (Ave)
Observed allele number	706.000	2.000–2.000(2.000)
Expected allele number	682.980	1.828–2.000(1.935)
Observed heterozygous number	200.389	0.459–0.621(0.568)
Expected heterozygous number	170.420	0.453–0.500(0.483)
Nei diversity index	171.068	0.454–0.502(0.485)
Shannon–Wiener index	238.542	0.645–0.693(0.676)
Polymorphism information content	129.249	0.350–0.375(0.366)

). Quality control confirmed that these SNP markers exhibited complete genotypic data, homozygosity dominance, and consistent identification across all 141 Munake accessions. Ultimately, 42 samples were identified for fingerprinting based on core germplasm selection (Supplementary Figure S4). A QR-encoded 2D barcode system was implemented to link genotypic data with individual accessions (Supplementary File S2).

3.4. Analysis of the Relationship Between Thompson Seedless and Munake Grapes

We performed a joint analysis using the existing SSR labeling data for Munake, as well as the newly generated data for Thompson Seedless, and we analyzed the genetic diversity between these two varieties. The Shannon information index (I) for the populations in 11 regions ranged from 1.03 to 1.96; the average value between the Munake populations was 1.56, and the average value between the Thompson Seedless populations was 1.52. The observed heterozygosity (Ho) in these 11 regions ranged from 0.77 to 0.88; the average value between the Munake populations was 0.80 and the average value between the Thompson Seedless populations was 0.85. This may indicate that they have similar richness in terms of genetic diversity and a close genetic relationship. In this study, the genetic differentiation of the populations was also tested through each locus’s fixation indices (FIS, FST), and the F-statistic results for each locus were as follows. The inbreeding coefficient (FIS) for six microsatellite loci ranged from −0.37 to 0.36. The fixation index (FST) ranged from 0.11 to 0.21. There was a certain degree of genetic differentiation between the populations in different regions, but the overall degree of differentiation was not particularly high. The gene flow (Nm) value was between 0.93 and 2.01. Regarding the magnitude of gene flow, VVIP31 showed the highest degree of gene exchange, with 83.3% of the gene loci Nm values being greater than 1. Overall, the average extent of gene flow between the sites was 1.28. An analysis of the genetic differentiation and gene flow among the 11 different populations showed that the Dest values ranged from 0.37 to 0.56 and the Gst values ranged from 0.08 to 0.18. These differences in genetic differentiation among the populations in different regions are of great significance for the protection and utilization of grape germplasm resources. They reflect the genetic diversity and adaptability of populations and can help to formulate effective germplasm conservation strategies. For detailed results, see

Table 5. Genetic parameters derived from 6 SSR loci in 144 samples of Munake grapes and 165 samples of Thompson Seedless grapes.

Pop	Grape Varieties		N	Na	Ne	I	Ho	He
KOR	Munake	Mean	11.833	7.500	4.635	1.638	0.847	0.741
		SE	0.167	1.500	0.821	0.225	0.137	0.054
ATS	Munake	Mean	3.833	3.333	3.188	1.029	0.833	0.599
		SE	0.167	0.955	0.969	0.219	0.167	0.060
KSG	Munake	Mean	27.833	9.833	4.204	1.672	0.786	0.737
		SE	0.167	0.749	0.610	0.118	0.153	0.036
ATX	Munake	Mean	25.667	10.167	3.775	1.624	0.801	0.705
		SE	0.211	1.400	0.443	0.153	0.112	0.054
RQ	Munake	Mean	13.500	7.667	4.299	1.580	0.771	0.707
		SE	0.224	0.989	0.969	0.188	0.145	0.064
HT	Munake	Mean	46.667	15.667	4.946	1.960	0.815	0.768
		SE	0.333	0.955	0.713	0.139	0.095	0.042
QN	Munake	Mean	10.000	7.000	4.437	1.646	0.783	0.759
		SE	0.000	0.775	0.517	0.119	0.101	0.027
TTOU	Munake	Mean	3.000	4.167	3.683	1.330	0.778	0.704
		SE	0.000	0.401	0.421	0.119	0.111	0.045
TDBG	Thompson Seedless	Mean	67.333	9.667	3.336	1.304	0.806	0.629
		SE	2.654	2.753	0.711	0.240	0.147	0.070
TGV	Thompson Seedless	Mean	28.667	8.833	3.953	1.476	0.856	0.688
		SE	1.145	1.922	1.020	0.212	0.121	0.049
SS	Thompson Seedless	Mean	39.500	14.667	4.523	1.775	0.882	0.724
		SE	2.566	1.498	1.071	0.193	0.073	0.053

Note: Pop stands for Population Name. KOR—Korla population; ATS—Artoush population; KSG—Kashgar population; ATX—Artux population; RQ—Ruoqiang population; HT—Hetian population; QN—Qiemo population; TTOU—Utuprague population; TDBG—Turpan Desert Botanical Garden population; TGV—Turpan grape valley population; and SS—Shanshan area population. Na = number of different alleles; N = number of individuals sampled; Ne = effective number of alleles; I = Shannon’s information index; and HO = observed heterozygosity.

and

Table 6. Genetic distance data according to regional distribution.

All Pops.	Fis	Fst	Gis	Gst	Nm
VVIP31	−0.20	0.11	−0.15	0.08	2.01
VVMD7	−0.19	0.21	−0.14	0.18	0.93
Vchr8a	0.36	0.18	0.40	0.13	1.13
VMC4F3	−0.31	0.16	−0.26	0.13	1.37
VrZAG67	−0.18	0.17	−0.13	0.14	1.22
VCHR13a	−0.37	0.20	−0.33	0.18	1.01

Note: Fis = inbreeding coefficient within individuals; Fst = inbreeding coefficient within subpopulations; Gis = inbreeding coefficient within individuals; Gst = analog of Fst; and Nm = gene flow.

. However, the Munake populations Artux and Utuprague exhibited notably lower values for Na, Ne, I, and H. It is critical to acknowledge that the small sample sizes of these two populations (n = 4 and n = 3, respectively) may lead to an underestimation of genetic diversity and inflated statistical errors. For this reason, we performed a significant difference analysis to verify the effect of the sample size of the two populations on the results, which showed that Na is significantly affected by the number of samples, whereas Ne, I, and He show a non-significant effect (Supplementary Figure S5). Future studies with expanded sampling are required to validate these observations.

We performed a PCoA on the samples of the two varieties. In the PCoA diagram, it can clearly be seen that Thompson Seedless and Munake are divided into two large populations, and there is strong genetic differentiation between the Munake and the Thompson Seedless populations. In the PCoA plot, PC1 accounts for 14.18% and PC2 accounts for 5.26% of the variation (Figure 7A). The use of primers consistent with Munake can not only aid in the analysis of the genetic diversity of Thompson Seedless but also contribute to the identification of different varieties. In addition, the PCoA diagram shows that some individuals from the Munake grapes gathered near the Thompson Seedless grapes, which also reflected the correlation between them. Moreover, in the PCoA results, the Munake individuals showed a dispersed pattern, whereas the Thompson Seedless individuals were mostly clustered at three points, which may have resulted from the different planting styles implemented for the two grapes.

The clustering tree also explained the results of the PCoA. The developmental tree did not accurately separate the two varieties, and the Thompson Seedless individuals were more likely to serve as sub-branches of Munake (see the circled area in Figure 7B). This implies that there is a genetic relationship between the individuals in both grape varieties and that the Thompson Seedless grapes may be based on the Munake grapes bred for the purpose of this study.

Bayesian clustering revealed significant differences in the genetic structure between Thompson Seedless and Munake. ΔK indicated that the best classification result was at K = 4. The population structure analysis plots from K = 3 to K = 5 illustrate that Munake was the first to begin to show intra-population structural differentiation at K = 3, while Seedless White did not show intra-population structural differentiation until K = 5 (Figure 8A). The population structure analysis (Figure 8B) revealed the trends in the parameters corresponding to the K values and ΔK, showing a clear inflection point at K = 4, indicating that the sample individuals may have contained four genetic subgroups. When K = 4, populations 3, 5, 7, and 9 are clustered together. At K = 4, the genetic structure of the Munake populations shows slight differences between some regions, while the Thompson Seedless populations do not differ in their genetic structure according to the planting location (Figure 8C).

4. Discussion

In crop breeding systems, the characterization and application of genetic diversity hold particular significance. Molecular markers capable of detecting DNA-level polymorphisms have been widely adopted as robust tools for diversity assessment [64,65]. Among these, simple sequence repeat (SSR) markers have emerged as particularly valuable for genetic fingerprinting due to their high abundance, polymorphism, codominant inheritance patterns, technical reproducibility, and adaptability to automated platforms [66,67,68,69,70]. Our analyses substantiated the efficacy of SSR markers, revealing substantial polymorphism across 16 loci [71]. Through genotyping 165 samples, we identified 442 alleles with an average of 27.63 alleles per locus. Genetic diversity parameters showed expected heterozygosity (He) = 0.76 and observed heterozygosity (Ho) = 0.83. An analysis of molecular variance (AMOVA) indicated that 96% of genetic variation occurred within populations, with limited inter-population differentiation (Fst = 0.04). A further population genetic analysis indicated that materials were not strictly grouped by geographic origin, findings consistent with the reproductive characteristics of Thompson Seedless. These findings collectively highlight the genetic vulnerability inherent to clonally propagated crops and underscore the imperative to develop dynamic conservation frameworks incorporating wild germplasms [72]. Phylogenetic reconstruction using SSR markers suggested that Thompson Seedless may represent a domesticated derivative of Munake grapes. Population structure analysis (K = 4) revealed subtle differentiation within Munake populations, likely attributable to gene flow from sympatric varieties. This observation aligns with Munake’s reproductive biology characterized by natural outcrossing and seed-based propagation. Conversely, Thompson Seedless exhibited genetic homogeneity across populations, reflecting constrained variation under intensive clonal cultivation. Notably, this study has limitations, including the potential underestimation of genetic diversity parameters in the Artux (n = 4) and Utuprague (n = 3) populations due to small sample sizes. A significant difference analysis confirmed that the Na (allele number) correlated significantly with sample size (Supplementary Figure S1). Future research should expand sampling, integrate epigenetics to quantify clonal variation from cultivation practices, and apply these insights to molecular breeding.

The establishment of a core collection will significantly enhance the efficiency of germplasm management and utilization [73,74]. Core collections built using molecular markers are less susceptible to environmental or external factors. In this study, using a high-throughput SNP genotyping platform, we selected and developed a set of core SNP markers from the initial SNP dataset of Thompson Seedless, achieving 94% variant coverage. A total of 25 accessions were identified as representative of the core germplasm bank for Thompson Seedless. The low variation level in Thompson Seedless is closely tied to its propagation traits, as clonal propagation restricts new variation generation. The Munake core germplasm bank established in this study (42 accessions, 353 SNP markers) captured 95% of allelic diversity. Its high polymorphism (PIC = 0.36–0.37) and homozygosity consistency effectively resolved synonym issues (e.g., regionally synonymous cultivars). The integration of KASP (Kompetitive Allele-Specific PCR) technology and a QR-encoded barcoding system enabled rapid genotype validation and portable data access, providing a model for digital germplasm management. As a vital cultivated crop, the genetic variations in grapes are primarily concentrated in structural variations (SVs). Performing an SNP (single-nucleotide polymorphism) analysis effectively complements the limitations of SSR (simple sequence repeat) studies, enabling the exploration of more extensive genetic variation. Munake grapes, representing small-scale farmer cultivation, exhibit higher genetic diversity compared to Thompson Seedless grapes. The latter, due to clonal propagation and intensive cultivation practices, suffer from severe genetic homogenization, resulting in significant loss of genetic information. Our core germplasm results demonstrate that to achieve the same variant coverage, the number of core accessions required for Munake grapes (29%) exceeds that for Thompson Seedless (25%). This validates the critical impact of different cultivation modes on grape genetic diversity.

The Xinjiang region plays a pivotal role in grape domestication and dissemination, and studies on the genetic diversity of its grape cultivars hold significant reference value for genetic breeding [5]. Among these, Thompson Seedless and Munake grapes, two major Eurasian cultivars cultivated in China for nearly a millennium, represent distinct agricultural practices: dispersed cultivation and intensive cultivation. This study marks the first investigation into the genetic diversity and phylogenetic relationships of these two cultivars using SSR and SNP molecular markers, contributing substantially to germplasm conservation, varietal identification, and marker-assisted breeding. Through screening their core germplasm collections, we explored the critical impact of different cultivation practices on grape genetic diversity, revealing that Thompson Seedless faces more urgent conservation challenges due to severe genetic homogenization caused by clonal propagation. Our findings demonstrate that introducing natural gene flow (e.g., seed propagation and hybridization with wild germplasms) into clonal propagation-dominated systems is a vital strategy to mitigate genetic erosion and enhance varietal adaptability. Future research must integrate functional genomics and phenotypic data to design targeted conservation and breeding strategies, balancing the sustainability and stress resilience of Xinjiang’s grape industry [9,75].

5. Conclusions

This study employed SSR molecular markers and whole-genome resequencing technologies to reveal significant genetic homogenization in Thompson Seedless grapes (96% of genetic variation originating within populations, Fst = 0.04) caused by long-term clonal propagation and intensive cultivation, sharply contrasting with Munake grapes, which maintained high genetic diversity through natural gene flow. Based on whole-genome data, 25 core germplasm accessions of Thompson Seedless (retaining 94% allelic diversity) and 42 core accessions of Munake (retaining 95% allelic diversity) were identified. A QR code-based genotypic fingerprint database was constructed using highly polymorphic SNP markers, effectively resolving synonym issues caused by homonymous naming. A phylogenetic analysis supported the origin of Thompson Seedless from a domesticated branch of Munake. For germplasm conservation, introducing gene flow is critical for Thompson Seedless to mitigate genetic erosion. Limitations in sample size and mechanisms underlying clonal variation warrant further investigation. Future research should integrate multi-omics approaches to optimize stress-resilient breeding strategies for Xinjiang’s viticulture industry.