Genetic Diversity and Association Analysis of Dioscorea polystachya Germplasm Resources Based on Phenotypic Traits and SSR Markers

Abstract

Dioscorea polystachya (Chinese yam) is a crop valued for both medicinal and edible purposes, and exhibits rich genetic diversity. However, research into its germplasm resources remains understudied, and molecular breeding efforts lag behind. To bridge this gap, this study employed an integrated approach, combining the analysis of 23 phenotypic traits (17 qualitative and 6 quantitative) with genotyping using 19 polymorphic SSR markers. This combined strategy was applied to 53 accessions collected across 16 Chinese provinces to assess genetic diversity, population structure, and marker–trait associations. Phenotypic analysis revealed high diversity, with the Shannon diversity index (I) ranging from 0.09 to 1.15 for qualitative traits and from 1.45 to 1.79 for quantitative traits. Tuber traits exhibited the highest variability (with a CV up to 71.45%), indicating significant potential for yield improvement. Principal component analysis distilled phenotypic variation into eight principal components (accounting for 73.13% of the cumulative variance), and elite germplasm (e.g., DP24, DP52) was selected for breeding based on this analysis. Stepwise regression prioritized eight core evaluation traits (e.g., flowering rate, tuber length). SSR markers amplified 80 alleles (mean 4.211/locus), showing moderate genetic diversity (He = 0.529, PIC = 0.585). Population structure analysis divided accessions into two subpopulations, correlated with geographic origins: Group 1 (northern/southwestern China) and Group 2 (central/eastern China), reflecting adaptation to local climates and human selection. Association analysis identified 10 SSR loci significantly linked (p < 0.01) to key traits, including YM07_2 (flowering, R² = 13.94%), YM37_2 (leaf margin color, R² = 19.03%), and YM19_3 (leaf width, R² = 19.34%). This study establishes a comprehensive genetic framework for Chinese yam, offering molecular tools for marker-assisted breeding and strategies to conserve high-diversity germplasm, thereby enhancing the utilization of this orphan crop.

1. Introduction

Yam (Dioscorea spp.) is an annual or perennial entwined vine plant of the Dioscorea genus in the Dioscoreaceae family [1]. It is a significant medicinal and edible crop with considerable nutritional and economic value. Its tuber is rich in starch, protein, choline, saponin, mucopolysaccharide, and other medicinal ingredients and minerals [2,3]. The origin of yams can be traced back to the Late Cretaceous period, with more than 650 species currently recognized within nine genera of the Dioscoreaceae family [4]. Yam (Dioscorea spp.) originated in Asia, Africa, and the Americas, where prolonged domestication established three distinct cultivation centers [5,6]. Among these, Asia serves as the primary distribution center, cultivating species including Chinese yam, D. alata, D. esculenta, D. japonica, D. bulbifera, D. hispida, and D. quinqueloba [7]. As a primary center of yam origin and domestication, China has documented Dioscorea species since antiquity—most notably in the pre-Qin era Classic of Mountains and Seas (Shanhaijing) [8]. The country harbors 93 native species, with Chinese yam, D. alata, D. esculenta, D. fordii, and D. persimilis being the most widely cultivated [9]. Chinese yam is endemic to China and uniquely adapted to temperate climates, exhibiting greater cold tolerance than other Dioscorea species [10]. Commonly known as ‘Huai Shan Yao’, it has been used in traditional medicine for over 2000 years to treat diarrhea, diabetes, and asthma. Recognized in the Chinese Pharmacopoeia, it remains a vital medicinal resource [11]. Chinese yam is extensively cultivated across China (excluding Qinghai and Xizang), with major production in Henan, Shandong, Jiangsu, Guangxi, Hunan, and Shanxi provinces [12]. According to statistics presented at the First National Yam Industry Development Conference (2023), the total cultivation area reached approximately 32,600 ha, yielding 450,000 t with an output value of CNY 2.7 billion.

Despite its significant nutritional and medicinal value, yam receives limited research attention and investment, resulting in its longstanding classification as an “orphan crop”—particularly exemplified by Chinese yam, which is well-known in China yet critically understudied globally [10]. Presently, research on Chinese yam is primarily concentrated on its medicinal value and the advancement of medicinal components, with the relatively limited investigation into resource types, evaluation, and genetic diversity [13,14,15]. Germplasm resources function as both repositories of genetic information and fundamental substrates for genetic improvement, while also serving as the essential basis for novel material development and cultivar breeding [16]. Genetic diversity research serves as a pivotal tool for assessing and leveraging germplasm resources [17]. Conducting genetic diversity analysis and comprehensive evaluation of Chinese yam germplasm resources enables enhanced conservation and characterization, reveals germplasm richness and genetic relationships, and provides elite materials to improve breeding quality and efficiency [18]. This constitutes a critical step toward advancing scientific innovation in Chinese yam germplasm utilization.

Phenotypic traits, isoenzymes, karyotype analysis, and DNA diversity have been used to study the genetic diversity of yam germplasm [9,19,20,21]. Phenotypic diversity represents the external expression of genetic diversity and constitutes the most fundamental methodology for the selection of germplasm resources and genetic background research [18,19,22]. The study of phenotypic genetic diversity in Chinese yam is primarily concerned with the examination of morphological traits, including leaves, stems, tubers, seeds, flowers, and roots [23,24,25,26,27,28]. Phenotypic traits are relatively straightforward to quantify; however, they are constrained by numerous factors and are particularly susceptible to environmental influences. Molecular markers are not susceptible to environmental influences and have been extensively employed in the fields of plant classification, breeding, and germplasm resource assessment [29]. Molecular markers such as Random Amplified Polymorphic DNA (RAPD), Inter-Simple Sequence Repeat (ISSR), Sequence-Related Amplified Polymorphism (SRAP), Amplified Fragment Length Polymorphism (AFLP), and Simple Sequence Repeats (SSR) have been extensively used in the investigation of the genetic diversity of yam [30,31,32]. However, most relevant studies have focused on D. alata, with scarce research reported on Chinese yam [18,30,31]. SSR molecular markers have the advantages of co-dominance, reproducibility, and high polymorphism, and have been widely used for the identification of germplasm resources and analysis of genetic diversity [32]. They are also one of the ideal markers for association analysis [33,34]. Association analysis leverages linkage disequilibrium to statistically assess associations between target traits and genetic markers in natural populations [35]. This approach has been widely adopted in plant genetics and breeding to identify genes and marker loci linked to agronomically important traits [36].

Chinese yam is dioecious and rarely flowers, and its seeds rarely develop to maturity, which makes cross-breeding for specific traits challenging [37]. Furthermore, it has long relied on asexual reproduction using tubers as planting material, leading to extensive viral accumulation, degradation of tuber quality, and significant reductions in both yield and quality. These are urgent issues that need to be addressed in Chinese yam agricultural production. However, studying the quantitative trait loci (QTL) associated with Chinese yam traits via linkage mapping is highly difficult and time consuming. Additionally, due to the late start of related research, the genome of Chinese yam has not yet been published, and the development of SNP molecular markers requires large-scale sequencing and incurs high costs. Therefore, using SSR molecular markers in combination with phenotypic traits to identify molecular markers associated with important phenotypic traits represents an effective approach to assist in the molecular breeding of Chinese yam. However, few such studies have been reported.

To clarify the genetic diversity of Chinese yam, identify molecular markers associated with important phenotypic traits, and provide a theoretical basis and material support for germplasm identification, innovation, and variety breeding, 53 germplasm resources of Chinese yam were collected from 16 provinces in the current study. Their genetic diversity and population structure were comprehensively characterized and evaluated by combining phenotypic traits with SSR molecular markers. Meanwhile, association analysis was performed to correlate quantitative traits with SSR molecular markers, aiming to identify marker loci associated with phenotypic traits. The findings of this study can provide a theoretical foundation and marker resources for molecular marker-assisted breeding in Chinese yam.

2. Materials and Methods

2.1. Plant Materials

A total of 53 Chinese yam germplasm resources were collected from major production regions across 16 provinces in China, encompassing both landraces and wild accessions (Table S1 and Figure 1A). All samples were planted in the Yam Germplasm Resource Garden of Jiangxi Agricultural University (Nanchang City, Jiangxi Province) in 2019. Tubers (80–120 g in weight) were used as seedings, with a row spacing of 20 cm × 1.2 m. Each plant was supported by a 2 m high bamboo pole for twining and was regularly managed with standard fertilization, weeding, irrigation, and pest control (Figure 1B). Each accession was planted with 10 individual plants. To ensure representativeness, 6 plants were randomly selected from the 10 for phenotypic trait determination, with three replicates conducted for each trait.

2.2. Determination of Phenotypic Traits

Twenty-three phenotypic traits of leaves, stems, flowers, aerial tubers, tubers, and roots were investigated, including seventeen qualitative traits and six quantitative traits (

Table 1. Standardized description and classification of phenotypic traits in Chinese yam germplasm.

Traits	Classes (Codes)
	1	2	3	4	5
Flowering (FL)	Absent	Male	Female
Aerial tubers (ATs)	Absent	Present
Leaf shape (LS)	Heart triangle	Triangular ovate	Lanceolate	Round	Shape of halberd
Leaf color (LC)	Yellow-green	Greenish gray	Dark green
Leaf apex shape (LAX)	Obtuse	Acute
Distance between lobes (DBLs)	Intermediate	Very distant
Leaf margin color (LMC)	Green	Purple
Petiole color (PC)	Purple	Green	Greenish purple	Purplish red
Leaf vein color (LVC)	Yellow-green	Green	Purple
Leaf vein (LV)	Five	Seven	Nine
Stem color (SC)	Green	Green with purple	Brownish green	Purple
Tuber shape (TS)	Oval	Cylindrical	Irregular
Roots hair density (RHD)	Sparse	Dense
Place of roots on the tuber (PRT)	All	Upper and Middle
Tuber skin color (TSC)	Brown	Black	Gray
Tuber skin color under bark (TSCUB)	Beige	Purple
Flesh color (FC)	White	Yellow
Leaf length (LL)	Average leaf length of six mature leaves (cm)
Leaf width (LW)	Average leaf width of six mature leaves (cm)
Length to width ratio (LWR)	Average leaf length/average leaf width(cm/cm)
Tuber length (TL)	Average tuber length of six plants (cm)
Tuber diameter (TD)	Average tuber diameter of six plants (mm)
Tuber flesh weight (TFW)	Average yield of six plants (g)

). The investigation of leaf, stem, aerial tuber, and flower traits was conducted from June to September 2019 during the growing season, while that of tuber and root traits was conducted from October to November 2019 after harvest. The phenotyping methodology adhered to published standards [38,39,40]. All traits were evaluated using the descriptor-based grading system detailed in Table 1.

2.3. SSR Molecular Marker Analysis

Genomic DNA was extracted from leaves using the Takara MiniBest Plant Genomic DNA Extraction Kit (Takara, Beijing, China). The DNA concentration was measured with a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), and integrity was confirmed using 1.0% (w/v) ethidium bromide-stained agarose gels. Samples were diluted to 20 ng/μL and stored at −20 °C. Nineteen SSR markers, which exhibited polymorphic bands across all accessions, were selected from the initial 53 amplified markers for further analysis (Table S2). Primer synthesis was performed by Sangon Biotech Co., Ltd. (Shanghai, China). The PCR amplification reaction mixture and amplification protocol for SSR markers were conducted according to Cao et al. [41].

2.4. Data Analysis

Phenotypic trait data were organized and calculated using Microsoft Excel 2020. Statistical analyses, including principal component analysis, correlation analysis, stepwise regression analysis, and calculation of descriptive statistics (maximum, minimum, mean, standard deviation (SD), and coefficient of variation (CV)), were performed using SPSS 24.0. The formula for calculating the Shannon–Wiener diversity index (I) is: I = −∑ (p_i × ln p_i), where pi is the percentage of materials in the i-th level of a certain trait relative to the total number of materials [21]. Quantitative trait data were divided into 10 grades, ranging from the first grade [x_i < (x − 2σ)] to the tenth grade [x_i > (x + 2σ)] with an interval of 0.5σ for each grade, and the Shannon–Wiener diversity index (I) was calculated [42].

SSR bands were scored as present (1) or absent (0) based on electrophoretic banding patterns to construct a binary matrix. The number of alleles, allele frequencies, gene diversity, and polymorphism information content (PIC) index were calculated using PowerMarker V3.25 [43]. Observed Heterozygosity (Ho), Expected Heterozygosity (He), Gene Diversity (H), and Polymorphic Information Content (PIC) are defined as: Ho = H/N, where H is the number of individuals with heterozygous genotypes in the population, and N is the total number of individuals; He = 1 − ∑p_i², where p_i is the frequency of the i-th allele and k is the total number of alleles at the locus; H = 1/n∑(1 − ∑p_i²), where n is the number of loci; PIC = 1 − ∑p_i² − ∑∑2p_i²p_j², where p_i and p_j represent frequencies of distinct alleles.

The population structure of all Chinese yam accessions was analyzed using STRUCTURE Version 2.3 software based on a Bayesian model [44]. The number of populations (K) was set from 1 to 10. For each K value, five replicates were run with a burn-in period of 100,000 iterations followed by 100,000 MCMC iterations. The optimal K value was determined using the maximum likelihood method and ∆K value, which was taken as the number of population subgroups. A bar plot was generated based on the Q value (the probability that the genomic variation of the i-th accession originates from the k-th population). Association analysis was performed using TASSEL 5.0 with both the General Linear Model (GLM) and Mixed Linear Model (MLM). For GLM, the population structure Q-matrix of the 53 accessions was used as a covariate in regression analyses between 19 SSR markers and eight phenotypic traits (FL, AT, LAX, LMC, LL, LW, LWR, TL) to identify significant loci and calculate their explained phenotypic variation (R²). For MLM, combined covariates of the Q-matrix and Kinship matrix were applied to the same traits to determine associated loci and their R² contributions [45]. The regression equations for GLM and MLM were Y = μ + βx + Qα + ε and Y = μ + βx + Qα + u + ε, respectively, where Y is the phenotypic trait vector, μ represents the population phenotypic mean, β denotes the marker genotype effect value, x is the marker genotype vector, Q indicates the population structure matrix, α stands for the covariate coefficients of population structure, ε refers to the random residual error, and u represents the individual random effect vector modeled by the Kinship matrix. The Kinship matrix was calculated using the Identity by State (IBS) algorithm with the formula K = (1/L)∑(number of shared alleles/2), where L is the total number of SSR markers (19) and K denotes the kinship coefficient between accessions i and j. SSR loci significantly associated with target traits were identified using a threshold of p ≤ 0.01.

3. Results

3.1. Analysis of Phenotypic Genetic Diversity of Chinese Yam Germplasm Resources

Significant phenotypic variation was observed in tuber, leaf, stem, and flower characteristics across the Chinese yam accessions (Figure 1C–F). The 17 qualitative traits from 53 germplasm accessions were categorized, with frequency distributions and diversity indices calculated for each trait (

Table 2. Frequency distribution and Shannon diversity index (I) analysis of qualitative traits in Chinese yam germplasm.

Traits	Classes (Codes)					I
	1	2	3	4	5
Flowering (FL)	0.23	0.28	0.49			1.04
Aerial tubers (ATs)	0.25	0.76				0.56
Leaf shape (LS)	0.02	0.49	0.06	0.08	0.40	1.15
Leaf color (LC)	0.08	0.06	0.87			0.48
Leaf apex shape (LAX)	0.45	0.55				0.69
Distance between lobes (DBLs)	0.66	0.34				0.64
Leaf margin color (LMC)	0.26	0.74				0.58
Petiole color (PC)	0.19	0.19	0.04	0.59		1.07
Leaf vein color (LVC)	0.68	0.28	0.04			0.74
Leaf vein (LV)	0.04	0.94	0.02			0.25
Stem color (SC)	0.04	0.55	0.02	0.38		0.90
Tuber shape (TS)	0.93	0.02	0.06			0.31
Roots hair density (RHD)	0.79	0.21				0.51
Place of roots on the tuber (PRT)	0.89	0.11				0.35
Tuber skin color (TSC)	0.45	0.36	0.19			1.04
Tuber skin color under bark (TSCUB)	0.98	0.02				0.09
Flesh color (FC)	0.98	0.02				0.09
Mean						0.62

). The Shannon diversity index (I) for qualitative traits ranged from 0.09 to 1.15 (mean = 0.59). The genetic diversity indices for flowering, leaf shape, petiole color, and tuber skin color all exceeded 1. LS showed the highest diversity (I = 1.15), while the tuber secondary color of upper bulbils (TSCUB), and flesh color color (FC) demonstrated the lowest diversity (I = 0.09).

3.2. Variation Analysis of Quantitative Traits

The six quantitative traits across 53 Chinese yam accessions exhibited substantial variation, with Shannon diversity indices (I) ranging from 1.45 to 1.79 (all > 1;

Table 3. Variation statistics and genetic diversity analysis of quantitative traits in Chinese yam germplasm resources.

Traits	Maximum	Minimum	Range	Mean	SD	CV	I
Leaf length (cm)	12.82	5.88	6.93	8.48	1.39	16.41	1.50
Leaf width (cm)	8.98	2.85	6.13	5.96	1.21	20.28	1.62
Length-to-width ratio	2.20	1.10	1.10	1.46	0.23	15.80	1.57
Tuber length (cm)	77.50	17.50	60.00	40.46	12.58	31.10	1.77
Tuber diameter (mm)	15.38	1.75	13.63	6.65	2.95	44.36	1.79
Tuber flesh weight (g)	1316.00	61.50	1254.50	350.72	250.60	71.45	1.45

SD: std. deviation; CV: coefficient of variation (%); I: genetic diversity index.

). The variation in the six quantitative traits among the 53 accessions differed, with coefficients of variation ranging from 15.80% to 71.45%. The coefficients of variation, in descending order, were for tuber flesh weight, tuber diameter, tuber length, leaf width, leaf length, and length-to-width ratio. Specifically, the coefficients of variation for tuber length, tuber diameter, and tuber flesh weight were 31.10%, 44.36%, and 71.45%, respectively, indicating that tubers have great potential for genetic improvement. Furthermore, the frequency distributions of the six quantitative traits exhibited continuous normal distributions, which meet the phenotypic data requirements for association analysis (Figure 2).

3.3. Comprehensive Evaluation of Phenotypic Traits of Germplasm Resources

3.3.1. Principal Component Analysis of Phenotypic Traits

Principal component analysis of the 23 phenotypic traits revealed that eight new independent comprehensive indices with eigenvalues greater than 1 could be extracted, with a cumulative contribution rate of 73.125% (

Table 4. The proportion of the morphological variation and traits of Chinese yam germplasm contribution explained by the first eight principal components.

Trait	Principal Component
	1	2	3	4	5	6	7	8
FL	0.01	0.47	0.14	−0.42	0.18	0.02	0.14	−0.06
AT	0.00	0.08	−0.38	−0.55	0.49	0.12	−0.14	0.00
LS	−0.51	0.05	0.39	0.37	0.17	0.02	0.15	0.04
LC	0.57	0.48	0.14	0.28	−0.13	0.08	0.12	0.03
LAX	−0.57	−0.17	0.03	−0.06	0.04	0.13	0.08	0.41
DBL	−0.41	−0.12	−0.11	0.34	0.41	−0.28	0.45	−0.02
LMC	−0.45	0.40	0.22	0.24	−0.09	−0.24	0.37	−0.03
PC	0.38	−0.13	−0.02	0.07	−0.35	−0.39	0.35	0.32
LVC	−0.16	−0.14	−0.22	−0.18	0.27	0.36	0.48	0.38
LV	0.65	0.01	0.46	−0.22	0.31	0.17	0.18	−0.01
SC	−0.33	0.44	0.26	0.18	0.10	−0.11	−0.27	0.16
TS	0.29	−0.43	0.08	0.47	−0.09	0.15	0.27	−0.21
RHD	0.27	−0.06	−0.31	0.39	0.28	−0.05	0.03	−0.49
PRT	−0.04	−0.47	−0.41	−0.05	0.42	0.11	0.27	−0.14
TSC	−0.08	0.43	0.18	−0.36	−0.37	0.21	0.41	−0.04
TSCUB	0.39	−0.43	0.75	−0.04	0.24	0.01	−0.04	0.07
FC	0.39	−0.43	0.75	−0.04	0.24	0.01	−0.04	0.07
LL	0.59	0.09	−0.39	0.13	0.20	−0.43	−0.11	0.39
LW	0.65	0.50	−0.21	0.08	0.30	−0.24	0.06	0.18
LWR	−0.36	−0.78	−0.04	0.04	−0.10	−0.18	−0.24	0.21
TL	−0.44	0.45	0.29	0.07	0.38	−0.10	−0.18	−0.13
TD	0.32	0.07	−0.27	0.43	−0.19	0.64	−0.11	0.16
YPP	−0.13	0.33	0.03	0.55	0.31	0.37	−0.19	0.27
Eigenvalue	3.67	3.03	2.49	2.02	1.73	1.41	1.36	1.11
Contribution rate (%)	15.95	13.16	10.83	8.80	7.50	6.14	5.91	4.83
Accumulated contribution rate (%)	15.95	29.11	39.94	48.74	56.24	62.39	68.30	73.13

). The eigenvalues of these eight principal components are 3.67, 3.03, 2.49, 2.02, 1.73, 1.41, 1.36, and 1.11, respectively, with corresponding contribution rates of 15.95%, 13.16%, 10.83%, 8.80%, 7.50%, 6.14%, 5.91%, and 4.83%, respectively. For the first principal component, the eigenvector of LV has the highest absolute value, followed by those of LW, LL, LC, and LAX, indicating that the first principal component is a comprehensive index related to leaf factors. For the second principal component, the eigenvector of LWR has the highest absolute value, followed by those of LW, LC, PRT, and FL. For the third principal component, the eigenvectors of TSCUB and FC have the highest absolute values, both at 0.746. The highest absolute values of the eigenvectors for the fourth, fifth, sixth, seventh, and eighth principal components correspond to YPP, AT, TD, LVC, and RHD, respectively.

3.3.2. Comprehensive Evaluation of Germplasm Resources

The data for the 23 phenotypic traits of the 53 accessions were standardized, and the factor score coefficients of the eight principal components mentioned above were substituted to obtain principal component scores. The comprehensive scoring formula for accessions was derived based on the values of D1–D8 and the contribution rate weights of each principal component: D = 0.22D1 + 0.18D2 + 0.15D3 + 0.12D4 + 0.10D5 + 0.08D6 + 0.08D7 + 0.07D8. This formula was used to evaluate the phenotypic traits of the 53 Chinese yam germplasm accessions. The top ten accessions with the highest comprehensive scores were DP24, DP52, DP22, DP5, DP23, DP42, DP39, DP1, DP2, and DP47 (

Table 5. The ranking of Chinese yam germplasm resources by calculating quality-related traits.

Accessions	D1	D2	D3	D4	D5	D6	D7	D8	D Value	Ranking
DP24	0.482	1.000	0.276	0.624	0.758	0.923	0.281	0.700	0.626	1
DP52	1.000	0.000	1.000	0.325	0.853	0.371	0.394	0.641	0.597	2
DP22	0.582	0.669	0.087	0.656	0.842	0.584	0.971	0.259	0.569	3
DP5	0.544	0.853	0.249	0.566	0.430	0.565	0.417	0.901	0.562	4
DP23	0.528	0.958	0.249	0.581	0.875	0.196	0.358	0.471	0.561	5
DP42	0.363	0.644	0.322	0.999	0.569	0.419	0.285	1.000	0.546	6
DP39	0.646	0.771	0.192	0.458	0.354	0.508	0.655	0.740	0.544	7
DP1	0.552	0.803	0.237	0.463	0.398	0.680	0.530	0.710	0.544	8
DP2	0.503	0.825	0.195	0.457	0.731	0.241	0.747	0.487	0.529	9
DP47	0.601	0.811	0.175	0.369	0.364	0.602	0.467	0.797	0.526	10
DP41	0.601	0.779	0.170	0.333	0.466	0.735	0.495	0.545	0.522	11
DP27	0.574	0.620	0.054	0.418	1.000	0.342	0.502	0.763	0.517	12
DP37	0.750	0.360	0.099	0.935	0.303	1.000	0.425	0.173	0.516	13
DP38	0.434	0.582	0.240	0.493	0.540	0.293	1.000	0.914	0.515	14
DP30	0.848	0.658	0.025	0.509	0.440	0.224	0.198	0.959	0.511	15
DP28	0.516	0.896	0.219	0.597	0.926	0.108	0.286	0.074	0.510	16
DP49	0.551	0.475	0.278	0.777	0.330	0.248	0.932	0.516	0.504	17
DP21	0.611	0.686	0.158	0.303	0.302	0.636	0.478	0.970	0.503	18
DP25	0.349	0.872	0.345	0.599	0.719	0.613	0.176	0.090	0.502	19
DP20	0.441	0.846	0.278	0.389	0.666	0.256	0.537	0.477	0.501	20
DP40	0.492	0.739	0.253	0.286	0.494	0.350	0.957	0.486	0.501	21
DP48	0.673	0.751	0.148	0.387	0.398	0.125	0.566	0.797	0.500	22
DP29	0.866	0.648	0.000	0.551	0.583	0.000	0.250	0.660	0.495	23
DP53	0.529	0.726	0.248	0.450	0.428	0.357	0.386	0.721	0.490	24
DP44	0.557	0.800	0.227	0.292	0.430	0.268	0.745	0.353	0.484	25
DP45	0.575	0.680	0.155	0.500	0.605	0.170	0.733	0.224	0.481	26
DP19	0.699	0.233	0.067	1.000	0.436	0.394	0.887	0.000	0.473	27
DP43	0.525	0.820	0.308	0.237	0.337	0.374	0.664	0.264	0.473	28
DP8	0.664	0.708	0.164	0.381	0.399	0.241	0.390	0.568	0.473	29
DP32	0.765	0.630	0.061	0.557	0.581	0.124	0.144	0.505	0.471	30
DP3	0.425	0.468	0.256	0.608	0.432	0.272	0.764	0.781	0.468	31
DP36	0.553	0.774	0.209	0.160	0.335	0.577	0.515	0.411	0.462	32
DP10	0.497	0.754	0.281	0.479	0.463	0.319	0.243	0.363	0.461	33
DP18	0.498	0.775	0.290	0.389	0.467	0.475	0.255	0.209	0.460	34
DP16	0.588	0.770	0.220	0.395	0.387	0.221	0.289	0.397	0.455	35
DP51	0.652	0.793	0.208	0.111	0.297	0.088	0.682	0.383	0.447	36
DP15	0.438	0.739	0.264	0.290	0.674	0.156	0.584	0.211	0.446	37
DP14	0.588	0.609	0.157	0.130	0.522	0.491	0.389	0.591	0.442	38
DP4	0.459	0.622	0.281	0.322	0.405	0.712	0.268	0.377	0.440	39
DP7	0.596	0.684	0.218	0.289	0.395	0.170	0.405	0.386	0.433	40
DP46	0.440	0.698	0.330	0.133	0.170	0.359	0.792	0.516	0.432	41
DP13	0.629	0.666	0.182	0.223	0.425	0.475	0.254	0.237	0.430	42
DP9	0.538	0.702	0.270	0.343	0.398	0.082	0.329	0.450	0.429	43
DP50	0.616	0.529	0.205	0.341	0.000	0.886	0.398	0.147	0.417	44
DP34	0.665	0.527	0.137	0.427	0.401	0.327	0.252	0.136	0.409	45
DP31	0.698	0.438	0.108	0.388	0.275	0.237	0.180	0.722	0.404	46
DP33	0.724	0.354	0.011	0.405	0.521	0.232	0.411	0.183	0.389	47
DP26	0.569	0.625	0.238	0.286	0.196	0.056	0.417	0.371	0.389	48
DP6	0.416	0.596	0.258	0.000	0.349	0.289	0.313	0.474	0.353	49
DP11	0.251	0.277	0.144	0.761	0.126	0.079	0.400	0.763	0.320	50
DP17	0.379	0.009	−0.144	−0.229	1.117	0.710	0.443	0.679	0.288	51
DP35	0.256	0.105	0.213	0.192	0.505	0.364	0.537	0.482	0.286	52
DP12	0.000	0.326	0.264	0.698	0.369	0.142	0.000	0.430	0.261	53

). Correlation analysis between the phenotypic trait D value and the 23 phenotypic traits is shown in Figure 3. The D value was highly significantly correlated with six phenotypic traits: it was negatively correlated with LWR and positively correlated with LC, LV, LW, TD, and YPP (Figure 3).

To screen the optimal evaluation indicators for Chinese yam germplasm resources, an optimal regression equation was constructed using 23 phenotypic traits and the comprehensive score D value. Stepwise linear regression was performed with the 23 phenotypic traits as independent variables and D values as dependent variables, yielding the optimal equation: y = 0.562 − 0.025x₁ − 0.028x₂ + 0.045x₅ + 0.040x₇ − 0.016x₁₇ + 0.02x₁₈ − 0.081x₁₉ + 0.001x₂₀ (r = 0.776, R² = 0.602, F = 2.064, p = 0.033), where x₁, x₂, x₅, x₇, x₁₇, x₁₈, x₁₉, and x₂₀ represent FL, AT, LAX, LMC, LL, LW, LWR, and TL, respectively. The correlation coefficient (r) and coefficient of determination (R²) of the equation were 0.776 and 0.602, respectively, indicating that the eight independent variables can explain 60.2% of the total variation in the D value. With an F value of 2.064 and a p value of 0.033, the equation reached a significant level, confirming its suitability for evaluating the comprehensive performance of Chinese yam germplasm. Thus, the phenotypic traits FL, AT, LAX, LMC, LL, LW, LWR, and TL were the main indicators for the comprehensive evaluation of Chinese yam germplasm resources.

3.4. Genetic Diversity Analysis of SSR Molecular Markers

3.4.1. Analysis of SSR Molecular Marker Polymorphism and Genetic Diversity

As shown in

Table 6. Genetic diversity of 19 SSR markers in 53 Chinese yam germplasm resources.

Markers	Na	Ne	I	Ho	He	H	PIC
YM02	5	2.297	1.033	0.302	0.571	0.565	0.703
YM03	2	1.935	0.676	0.531	0.488	0.483	0.318
YM06	7	6.531	1.970	0.725	0.858	0.847	0.927
YM07	6	4.681	1.646	0.558	0.796	0.786	0.907
YM09	7	6.179	1.864	0.619	0.848	0.838	0.942
YM12	2	1.679	0.594	0.521	0.409	0.404	0.402
YM13	4	1.288	0.493	0.244	0.226	0.224	0.392
YM17	2	1.913	0.670	0.787	0.483	0.477	0.294
YM19	4	2.720	1.131	0.935	0.639	0.632	0.533
YM21	2	1.641	0.579	0.532	0.395	0.390	0.374
YM24	2	1.939	0.677	0.600	0.490	0.484	0.473
YM30	2	1.110	0.205	0.063	0.100	0.099	0.150
YM32	8	5.563	1.848	0.773	0.830	0.820	0.942
YM33	3	1.190	0.325	0.135	0.161	0.160	0.259
YM35	5	3.458	1.337	0.891	0.719	0.711	0.819
YM37	7	4.080	1.641	0.886	0.764	0.755	0.904
YM41	2	1.156	0.261	0.146	0.137	0.135	0.218
Da1D08	5	2.741	1.201	0.739	0.642	0.635	0.843
SSR-17	5	1.977	0.927	0.575	0.500	0.494	0.717
Total	80	54.078	19.078	10.562	10.056	9.939	11.117
Mean	4.211	2.846	1.004	0.556	0.529	0.523	0.585

Na: observed number of alleles; Ne: effective number of alleles; I: Shannon’s diversity index; Ho: observed heterozygosity; He: expected heterozygosity; H: Nei’s gene diversity; PIC: polymor-phism information content index.

, 19 SSR markers amplified a total of 80 alleles in the 53 germplasm accessions, with an average of 4.211 alleles per marker. The primer YM32 yielded the highest number of alleles (eight). The average number of effective alleles was 2.846; the highest was observed for YM06 (6.531), and the lowest for YM30 (1.110). Shannon’s diversity index (I) ranged from 0.205 to 1.970, with an average of 1.004. The average observed heterozygosity (Ho), expected heterozygosity (He), and Nei’s gene diversity (H) were 0.556, 0.529, and 0.523, respectively. The polymorphism information content (PIC) index ranged from 0.15 to 0.942, with a mean of 0.585.

3.4.2. Analysis of Population Genetic Structure

The genetic structure of the Chinese yam population was analyzed using SSR data based on a Bayesian clustering model. The optimal number of groups (K) was estimated according to Evanno et al. [46], with the highest ΔK value for the 53 individuals observed at K = 2 (Figure 4A). At K = 2, the 53 germplasm accessions were divided into two groups: the first group (11 accessions) is represented in red, and the second group (42 accessions) is represented in green. These two groups are denoted by red and green bars in the bar plot, containing 11 and 42 germplasm accessions, respectively (Figure 4B). Additionally, Group 1 exhibited a lower leaf width ratio, shorter and thicker tubers, along with higher yield, whereas Group 2 showed contrasting characteristics but demonstrated greater phenotypic diversity (Table S3).

The phylogenetic tree constructed using the neighbor-joining method is shown in Figure 4C; the 53 germplasm accessions were clustered into two populations (red and green), consistent with the results of the population structure analysis. Specifically, accessions DP41, DP47, DP51, DP15, DP53, DP52, DP38, DP33, DP50, DP31, and DP34 were grouped together, with the remaining accessions forming the second group.

The results of principal component analysis (PCA) were consistent with those of the phylogenetic tree and population structure analysis, as the 53 germplasm accessions were also divided into two groups (Figure 4D).

3.5. Linkage Disequilibrium Analysis

Linkage disequilibrium analysis between SSR loci is shown in Figure 5. Among the 4005 locus combinations of the 19 SSR loci, a considerable proportion of loci were in linkage disequilibrium. A high degree of linkage disequilibrium (D > 0.5) was observed in 1734 loci combinations, accounting for 43.30%. Additionally, 183 combinations showed highly significant differences (p < 0.01), with a proportion of 4.67%.

3.6. Association Analysis of SSR Markers and Phenotypic Traits of Chinese Yam

The general linear model (GLM) and mixed linear model (MLM) were used to conduct regression analysis of the main phenotypic data and genotype data to identify loci associated with phenotypic traits. A total of 82 marker loci showed significant associations with FL, AT, LAX, LMC, LL, LW, LWR, and TL at the p < 0.05 level. The results from GLM and MLM analyses indicated that seven marker loci were significantly correlated with the phenotypic traits of FL, AT, LAX, LMC, LL, LW, LWR, and TL at the p < 0.01 level (

Table 7. SSR marker sites were significantly associated with major phenotypic traits (p < 0.01) and their explanation rates for phenotypic variation.

Trait	Marker Site	General Linear Model (GLM)		Mixed Linear Model (MLM)
		p	R2 (%)	p	R2 (%)
FL	YM07_2 (400 bp)	-	-	0.0093	13.94
AT	YM06_3 (350 bp)	0.0016	18.03	0.0036	18.58
AT	YM13_3 (400 bp)	0.0085	12.96	-	-
LMC	YM13_3 (400 bp)	0.0013	18.72	0.0035	18.75
LMC	YM37_2 (350 bp)	0.0012	19.03	0.0063	16.25
LMC	YM37_7 (100 bp)	0.0068	13.73	0.0026	20.16
LL	Da1D08_4 (300 bp)	0.0028	16.46	0.0055	16.77
LL	YM19_3 (130 bp)	-	-	0.0067	15.94
LWR	YM06_1 (500 bp)	0.0062	13.79	0.0058	16.70
LW	YM19_3 (130 bp)	-	-	0.0032	19.34

). The marker gene significantly associated with AT was YM06_3 (350 bp) with explanation rates of 18.03%. The marker loci associated with LMC were YM13_3 (400 bp), YM37_2 (350 bp), and YM37_7 (100 bp), with explanatory rates of 18.72%, 19.03%, and 13.73%, respectively. The marker loci Da1D08_4 (300 bp) and YM06_1 (500 bp) were significantly correlated with LL and LWR, with explanatory rates of 16.46% and 13.79%, respectively.

MLM analysis revealed nine loci that are significantly associated with FL, AT, LMC, LL, LWR, and LW. The marker loci significantly correlated with FL and AT were YM07-2 (400 bp) and YM13_3 (400 bp), with an explanatory rate of 13.94% and 12.96%, respectively. The MLM analysis identified loci associated with LMC, consistent with the GLM results. The MLM analysis also identified loci significantly associated with LL, LWR, and LW, including YM19_3 (130 bp), YM06_1 (500 bp), and YM19_3 (130 bp) with explanatory rates of 15.94%, 16.70%, and 19.34%, respectively. Both GLM and MLM analyses identified six loci significantly associated with AT, LMC, LL, and LWR.

4. Discussion

Phenotypic trait analysis is a fundamental and intuitive method for studying plant genetic diversity, as it can reveal plant diversity, explore genetic resources, and provide an important basis for researching the mechanisms of complex traits [47,48,49]. For Chinese yam, the investigation and analysis of phenotypic traits in germplasm resources represent the primary work in germplasm research, the foundational work in breeding, and a simple yet effective approach. This method can reveal the genetic diversity of yam, explore genetic resources, and provide an important basis for studying complex traits. Currently, there are few relevant reports on the comprehensive evaluation of Chinese yam phenotypic characteristics. However, conclusions drawn from phenotypic markers have certain limitations; thus, it is more scientific and accurate to evaluate the genetic diversity of germplasm resources by combining phenotypic traits with molecular markers. Furthermore, association analysis based on population structure was conducted to identify significant loci and favorable alleles associated with target traits, providing genetic resources and molecular markers for Chinese yam breeding.

4.1. Comprehensive Evaluation of Phenotypic Characteristics of Chinese Yam Germplasm Resources

Principal component analysis is a multivariate statistical analysis technique, which is conducive to improving breeding levels and parent selection effects in multi-objective breeding [50,51]. PCA can convert multiple indicators into a few principal components, which are not correlated with each other and can reflect most of the information of the original data. This method has been widely used in the comprehensive evaluation of germplasm resources, but few studies have been reported on Chinese yam. In this study, 23 phenotypic traits were simplified into eight independent principal components through principal component analysis, with a cumulative contribution rate of 73.13%, representing most of the information on phenotypic traits. The F value of each resource was calculated by constructing the function expression of principal components, and the germplasm with excellent comprehensive characters, such as DP24, DP52, DP22, and DP5, was selected, which could be used for germplasm innovation and genetic improvement of Chinese yam. A mathematical model for phenotype evaluation of Chinese yam germplasm was established through comprehensive correlation analysis and stepwise regression analysis, and the key indicators for comprehensive evaluation were selected as FL, AT, LAX, LMC, LL, LW, LWR, and TL to provide a reference for genetic breeding research of Chinese yam.

PCA can convert multiple indicators into a few uncorrelated principal components, which collectively retain most of the information from the original data. This method has been widely used in the comprehensive evaluation of germplasm resources, but few studies have been reported for Chinese yam. In this study, 23 phenotypic traits were simplified into eight independent principal components via PCA, with a cumulative contribution rate of 73.13%, thus capturing most of the phenotypic trait information. By constructing a function expression for these principal components, we calculated the comprehensive score (D value) for each germplasm resource and identified accessions with excellent comprehensive traits, such as DP24, DP52, DP22, and DP5. These can be used for germplasm innovation and genetic improvement of Chinese yam. Through comprehensive correlation analysis and stepwise regression analysis, a mathematical model for phenotypic evaluation of Chinese yam germplasm was established. The key indicators selected for comprehensive evaluation were FL, AT, LAX, LMC, LL, LW, LWR, and TL, providing a reference for genetic breeding research on Chinese yam.

4.2. Analysis of Genetic Diversity of Yam Germplasm Based on Phenotypic Traits and SSR Molecular Markers

Genetic diversity levels can be reflected by the coefficient of variation and Shannon–Wiener index, where a high diversity index indicates rich diversity. The coefficient of variation reflects the variation characteristics of phenotypic traits, while the Shannon–Wiener index reflects their diversity [52]. The Shannon–Wiener index can also reflect the range of variation and the distribution of genotype frequencies, making it widely used to evaluate phenotypic trait diversity [53]. In this study, the phenotypic diversity of 53 Chinese yam germplasm accessions was analyzed using 17 qualitative traits and 6 quantitative traits. High genetic diversity was observed for several phenotypic traits of Chinese yam, including FL, LS, PC, and TSC (I > 1), which is consistent with our previous analysis of phenotypic genetic diversity in five Dioscorea species [41].

SSR markers, characterized by high abundance, stability, and polymorphism, have been widely used in evaluating genetic diversity, constructing genetic maps, and determining species lineages across many plants. For the 19 pairs of SSR primers selected in this study, the average values of Na, Ne, I, Ho, He, H, and PIC were 4.211, 2.846, 1.004, 0.556, 0.529, 0.523, and 0.585, respectively—similar to our previous results on genetic diversity in five yam species [41]. Zhang et al. (2025) developed SSR markers via high-throughput sequencing and selected 42 polymorphic primer pairs to analyze the genetic diversity in 101 yam accessions [36]. Their reported mean values for Na, I, and PIC were 5.409, 0.551, and 0.586, respectively. Compared with the present study, their Na was higher, I was lower, and PIC was similar. These discrepancies may be attributed to differences in primer selection and plant materials. As a geographically restricted endemic to China, Chinese yam remains understudied, with few reports on SSR marker-based genetic diversity analyses. Zhou et al. analyzed the genetic diversity of 28 Chinese yam cultivars using ISSR markers and reported an I value of 0.3191, which is lower than that in our study [54]. Wang et al. analyzed the genetic diversity of 51 Chinese yam accessions from six species using ISSR, ISAP, SRAP, and SCAR markers, with reported H and I values of 0.4950 and 0.6762, respectively [55]. These differences may stem from variations in the type and quantity of test materials, as well as the types of molecular markers used. Additionally, primers YM02, YM06, YM07, YM09, YM32, YM35, YM37, Da1D08, and SSR-17—with PIC > 0.7 and clear amplification profiles—were identified as optimal for evaluating the genetic diversity of Chinese yam germplasm resources.

4.3. Analysis of Population Structure of Chinese Yam Germplasm Materials

Population structure analysis based on SSR markers revealed a clear genetic division of the 53 Chinese yam germplasm accessions into two groups (K = 2), supported by consistent results from PCA and neighbor-joining tree analyses (Figure 4B–D). This genetic divergence may be attributed to multiple factors, including geographical isolation, historical domestication events, and artificial selection.

The two genetic clusters showed a weak but noticeable correlation with geographical origins. For example, the first group (11 accessions) included materials from northern China (e.g., Hebei, Shandong, and Shanxi) and southwestern regions (e.g., Yunnan and Sichuan), while the second group (42 accessions) was dominated by accessions from central and eastern China (e.g., Henan, Jiangxi, and Hubei). This pattern aligns with the historical cultivation centers of Chinese yam, where northern varieties adapted to temperate climates (e.g., DP39 “Baiyu Yam” from Shandong) and southwestern accessions evolved under subtropical conditions (e.g., DP52 “Honglong Yam” from Yunnan). The presence of admixed individuals (e.g., DP41 “Chenji Tiegun Yam” from Shandong) suggests gene flow between regions due to human migration or trade.

Phenotypic divergence between the two genetic clusters may reflect adaptation to local selective pressures. Group 1’s combination of shorter tubers and higher flesh weight suggests selection for yield optimization in northern regions (e.g., Shanxi), whereas Group 2’s elongated tubers align with traditional shape preferences in central China. These ecotypic adaptations were consolidated via PCA, which linked tuber morphology and leaf architecture to primary axes of variation (Table 4).

The genetic structure provides a foundation for rational germplasm conservation and breeding. Group 1 contains rare wild accessions (e.g., DP52 from Yunnan), which should be prioritized for in situ conservation to preserve adaptive alleles. Group 2, dominated by cultivated varieties, offers potential for trait-specific breeding (e.g., high-yield lines from Henan). Future studies should integrate environmental variables (e.g., altitude, temperature) to disentangle the roles of natural selection and human intervention in shaping the population structure.

4.4. Correlation Analysis Between SSR Markers and Phenotypic Traits of Chinese Yam

Association analysis between SSR markers and phenotypic traits is pivotal for linking genetic diversity to agronomic performance, providing a foundation for marker-assisted selection (MAS) in breeding programs [56,57]. Hansen et al. identified two AFLP markers associated with genes influencing the vernalization requirement before bolting in Beta vulgaris ssp. maritima [58]. This study represented the first reported application of association analysis at the whole-genome level, which subsequently became a widely investigated approach. However, relatively few studies on association analysis for Chinese yam have been reported to date.

In this study, 10 SSR loci were significantly associated with eight key traits (e.g., flowering, leaf morphology, and tuber characteristics) at p < 0.01 using GLM and MLM (Table 7). It is noteworthy that the GLM and MLM models detected seven and nine associated loci, respectively, with an overlap of six common loci, highlighting their distinct detection patterns and complementary nature. In the GLM analysis, only the Q-matrix derived from subpopulation membership was employed, whereas the MLM analysis—by incorporating both population structure and kinship—typically achieves a significant reduction in false-positive loci [59]. In this study, the AT-associated locus YM13_3 showed significant association exclusively in GLM but was undetectable via MLM, suggesting potential false positives due to population stratification. The loci consistently identified by both models (e.g., YM37_7 associated with leaf vein color, LMC) likely represent the most reliable associations, with the MLM model explaining 20.16% of the phenotypic variance. Future studies should prioritize further validation of loci identified through the MLM framework. These association results highlight the potential of SSR markers in elucidating the genetic basis of complex traits and accelerating genetic improvement in Chinese yam. The locus YM07_2 (400 bp) showed a significant association with flowering traits (FL) (p = 0.0093, R² = 13.94%), suggesting that it may be related to regulatory genes controlling flowering. Delayed or absent flowering in Chinese yam is a critical bottleneck for seed-based breeding due to its dioecious nature and reliance on vegetative propagation. This marker could serve as a diagnostic tool to screen early-flowering germplasm, facilitating hybridization and genetic studies. The marker YM19_3 (350 bp) explained 19.34% of the variation in leaf width (LW), indicating that this trait may be under strong genetic control. Leaf morphology influences photosynthetic efficiency and stress adaptation, and this marker could guide the selection of germplasm with optimal leaf architecture for specific environments. Consistent with the present study, Zhang et al. identified 15 loci associated with stem diameter, leaf shape, leaf length, leaf width, petiole length, petiole color, sex type, and single-head weight using 44 SSR markers [49]. In summary, conducting significant correlation analysis between phenotypic traits and SSR markers provides insights into the genetic basis of agronomically important traits in Chinese yam, thus enhancing breeding efficiency and the success rate of cultivar development.

5. Conclusions

This study aimed to analyze the genetic diversity, population structure, and trait-associated molecular markers of 53 Chinese yam germplasm accessions using phenotypic traits and SSR markers, providing a theoretical basis for germplasm identification, innovation, and breeding. Phenotypic analysis revealed high genetic diversity in Chinese yam, with Shannon’s diversity index (I) > 1 for traits such as flowering (FL), leaf shape (LS), petiole color (PC), and tuber skin color (TSC). Tuber traits (e.g., tuber flesh weight) exhibited a coefficient of variation up to 71.45%, indicating significant potential for yield improvement. These results systematically characterize the diversity level of the germplasm, fulfilling the goal of evaluating genetic variability. SSR-based population structure analysis divided the 53 accessions into two subpopulations, corresponding to geographical origins in northern/southwestern China and central/eastern China, respectively. This pattern reflects the combined effects of climatic adaptation and human selection, providing a basis for tracing germplasm origins and targeted conservation. Association analysis identified 10 SSR loci significantly linked (p < 0.01) to key traits, including YM07_2 (flowering, R² = 13.94%), YM37_2 (leaf margin color, R² = 19.03%), and YM19_3 (leaf width, R² = 19.34%). These markers offer direct tools for marker-assisted breeding, addressing the objective of mining functional genetic loci. In summary, by integrating phenotypic and molecular data, this study establishes a comprehensive evaluation system for Chinese yam germplasm. It not only reveals genetic diversity and population differentiation but also identifies elite accessions (e.g., DP24, DP52) and functional markers for breeding, laying a foundation for germplasm conservation, genetic improvement, and industrial utilization of this species.