Uncovering the Genetic Basis of Grain Yield-Related Traits in Common Vetch (Vicia sativa L.) Through Genome-Wide Association Mapping

Abstract

Common vetch (Vicia sativa L.) is a globally green manure and forage crop, cultivated extensively worldwide. Its seeds serve as an important concentrated feed. Due to the late release of the reference genome, few studies were conducted to analyze the genetic mechanisms of grain yield, which hindered the progress of common vetch breeding. Marker-assisted selection (MAS) is the best and most effective way to accelerate the genetic improvement of grain yield-related traits in common vetch. In this study, we performed a genome-wide association study (GWAS) using the high-density single nucleotide polymorphism (SNP) data obtained through re-sequencing to better understand the genetic basis of grain yield-related traits. In total, six grain yield-related traits were evaluated in 172 accessions mainly sourced from China and Russia, across four environments, including branches per plant (NB), pod length (PL), number of pods per plant (NP), number of grains per pod (NG), hundred-grain weight (HGW), and grain yield (GY). Population structure analysis of the 172 accessions revealed four distinct subpopulations, exhibiting strong geographical correlation. In total, 38 loci have been identified as significantly associated with six grain yield-related traits, accounting for 13.3–31.7% of the phenotypic variances. Among them, qGY1.1 and qNG1.1, qNG2.2 and qPL2.1, qNG3.2 and qGY3.2, qNG4.1 and qPL4.1, qGY4.1 and qHGW4.1, qNG6.1 and qPL6.1, and qNB6.2 and qGY6.2 exhibit overlapping regions, suggesting that these regions are pleiotropic and should be prioritized for further research and breeding. In total, 12 candidate genes encoding auxin response factor, F-box repeat protein, gibberellin receptor, serine/threonine-protein kinase-like protein, and cellulose synthase-like protein were identified. Furthermore, we successfully developed and verified a kompetitive allele-specific PCR (KASP) marker (Kasp-NB6.2) for the number of branches. These findings provide molecular insights into grain yield-related traits in common vetch and offer valuable loci and molecular tools for MAS breeding.

1. Introduction

Common vetch (Vicia sativa L.), an annual legume native to Europe, Western Asia, and North Africa, is cultivated globally for its higher adaptability [1,2,3,4,5,6]. Historically domesticated since the Neolithic era, it now thrives across temperate and subtropical regions, with major cultivation spanning Europe (Spain, France), Asia (China, Turkey), the Americas (U.S., Brazil), and Australia, covering approximately 5–7 million hectares annually [1,2]. Seed yields average 1.2–2.6 tons per hectare, while biomass production reaches 22.5–50.0 tons per hectare. Primarily valued as a high-protein (18–25%) forage crop, it is often intercropped with cereals to enhance livestock feed quality [2,3,4]. Its nitrogen-fixing symbiosis with rhizobium bacteria is a key green manure in sustainable rotations to reduce fertilizer use. Additionally, it aids ecological restoration by stabilizing eroded lands and rehabilitating degraded ecosystems [3,4,5,6]. In addition, common vetch is also an important source of bioenergy, which can be utilized to produce biogas or bioethanol.

Common vetch is an important forage crop globally [3]. Its seeds are rich in protein, making them a valuable source of concentrated feed. However, the grain yield (GY) of common vetch is relatively low [2,4], and the price of forage seeds is high. Therefore, improving the grain yield of common vetch is important for improving its economic value. Grain yield (GY) is a complex and comprehensive trait influenced by the effects of multiple related traits. Among these, an increase in the number of branches (NB) enables the plant to produce more inflorescences and pods, providing a potential foundation for higher grain yield. Additionally, a greater NB may enhance the plant photosynthetic area and nutrient allocation efficiency, indirectly promoting grain formation and development. Pod length (PL) directly affects the number of grains per pod and grain size. Longer pods typically accommodate more grains or support more fully developed grains, thereby positively influencing grain yield. The number of pods per plant (NP) is one of the direct determinants of grain yield. More pods means more grains, significantly increasing grain yield. The increase in NP is closely related to NB. The number of grains per pod (NG) directly impacts the composition of grain yield. A higher NG leads to an increase in grain yield and is associated with pod length and grain development. Hundred grain weight (HGW) reflects grain size and plumpness, serving as an important component of grain yield. A higher HGW indicates more fully developed grains, which can significantly enhance grain yield. GY is the comprehensive result of all the aforementioned traits. The synergistic effects of NB, NP, NG, and HGW determine the final grain yield.

Current breeding efforts for common vetch primarily focus on enhancing its yield and nutritional traits [3,4,5,6,7]. Marker-assisted selection (MAS) breeding has significantly accelerated the improvement of stress resistance traits (e.g., drought and cold tolerance in soybean), agronomic traits (e.g., biomass and seed yield-related traits in maize), and disease resistance (e.g., rust and powdery mildew resistance in wheat) in field crops [8,9,10,11,12]. Additionally, molecular markers support functional genomics research [13,14]. Collectively, molecular markers serve as efficient tools for genetic improvement, resource utilization, and sustainable production of crops [15]. In common vetch, grain yield-related traits are typical quantitative traits, influenced by polygenic interactions and environmental factors [16]. Genome-wide association studies (GWAS) based on linkage disequilibrium (LD) have been widely applied to identify significant loci for complex agronomic and yield traits in major field crops [17]. The basis for conducting association analysis is the availability of comprehensive and evenly distributed genetic variations. The integration of next-generation sequencing (NGS) with GWAS has proven effective in uncovering the genetic basis of complex traits. In 2022, the release of a chromosome-level reference genome for common vetch [18] made it possible to perform re-sequencing using NGS. Furthermore, kompetitive allele-specific PCR (KASP) markers, a high-throughput genotyping method with high throughput, low cost, and rapid detection, are widely adopted in Poaceae and Cucurbitaceae crops [19,20]. However, to date, there have been no reports on the application of KASP markers in the molecular breeding of common vetch.

In this study, GWAS for six grain yield-related traits in a diverse panel with 172 elite common vetch accessions were used to (1) identify loci significantly associated with grain yield-related traits, (2) provide available molecular tools for common vetch MAS breeding, and (3) search for candidate genes of grain yield traits for further study (Table S1).

2. Materials and Methods

2.1. Plant Materials

In total, 172 common vetch accessions from 18 countries, with the majority originating from China (108 accessions, 62.8%, National Crop Germplasm Resources Green Manure Medium-term Storage Facility), followed by Russia (29, 16.9%), Ukraine (6, 3.5%), Spain (5, 2.9%), Belarus (4, 2.3%), Australia (3, 1.7%), the Czech Republic (3, 1.7%), Poland (2, 1.2%), Germany (2, 1.2%), and Sweden (2, 1.2%), and only one accession each from Bulgaria, France, Greece, Italy, Lithuania, Mexico, the Philippines, and Romania were collected (Figure 1, Table S1). All the 172 accessions were planted in 2020, 2021, 2022, and 2023 in Harbin, Heilongjiang Province (45°51′ N, 126°45′ E). All the 172 accessions are sown in mid-April each year and harvested in late July. A randomized block design was used, with plot planting in three replicates. Each plot had a row length of 5 m, a row spacing of 0.65 m, five rows per plot, and 100 seeds per row.

The investigation indicators NB, PL, NP, NG, HGW, and GY were evaluated according to Cao et al., 2007 [21]. The trait investigation standards referenced the Description Specifications and Data Standards for Green Manure Germplasm Resources [21]. NB includes all branches, including the main stem, whereas NP is the total number of mature pods on a single plant. NG is the total number of seeds contained within a single mature pod, whereas PL is the length of the pod. GY is the total seed yield per unit area from mature pods. HGW means the weight of 100 mature seeds. During the full flowering stage, 10 individual plants were randomly selected from each plot to measure NB and NP. Ten pods were randomly selected from each plant to measure PL and NG. Three 1 m sample sections (excluding border rows) were randomly selected from each plot, threshed, and weighed to calculate GY. HGW is the average of three measurements of 100 randomly selected air-dried mature seeds from each plot.

2.2. Re-Sequencing and Variant Identification

The genomic DNA from seedling leaves was isolated and genotyped by re-sequencing using the Illumina HiSeq 2500 platform (Illumina, Inc., San Diego, CA, USA) by Biomarker Biotechnology Co., Ltd. (Beijing, China). The paired-end read data (PE150) with a sequencing depth of approximately 15× of the common vetch genome were generated. The clean reads were aligned to the common vetch reference genome (GCF_026540005.1) using the BWA (v0.7.8) [22] and converted to binary alignment format via SAMtools (v0.1.19) (https://sourceforge.net/projects/samtools/files/samtools/0.1.19/, accessed on 25 April 2025). The average comparison rate between the sample and reference genome is 99.03%, with an average coverage depth of 15.18× and a genome coverage of 98.18%. Genome coverage analysis revealed that 72.54% of the reference genome (GCF_026540005.1) was covered at ≥1× depth, while 62.93% achieved ≥4× depth. SNP calling was conducted by GATK Haplotype Caller (v4.0.10.1) [23]. The SNPs were filtered by minor allele frequency (MAF) < 0.05 and missing rate > 10% (Table S2) for further GWAS analysis.

2.3. Population Analysis and Association Mapping

Population structure analyses were conducted by ADMIXTURE v1.3.0) [24]. Principal components analysis (PCA) and neighbor-joining trees were analyzed by the Tassel v5.0 (https://www.maizegenetics.net/copy-of-tassel, accessed on 12 April 2025). Broad-sense heritability (H_b²) estimates were calculated as follows: H_b² = σ²_g/(σ²_g + σ²_ge/n + σ²_e/nr), whereas σ²_g represents genetic variance, σ²_ge represents the genotype-environment interaction variance, σ²_e is the residual variance, n represents the environments, and r represents the replicates. A GWAS was performed for grain yield-related traits by the mixed linear model (MLM) model incorporating kinship matrices and principal components as covariates in TASSEL v5.0. Significant marker-trait associations (MTAs) were identified using a stringent threshold (−log₁₀(P) ≥ 5.0). The calculation of genetic polymorphism (π value) for subpopulations was performed using VCFtools v0.1.15 [25], with the default parameters of the software. The genome was divided into non-overlapping sliding windows of 10 Kb.

2.4. Development and Validation of the KASP Marker

To provide molecular tools for common vetch MAS breeding, KASP primers targeting trait-associated SNPs were designed. Fluorescent-labeled primers (FAM: 5′-GAAGGTGACCAAGTTCATGCT-3′; HEX: 5′-GAAGGTCGGAGTCAACGGATT-3′) and corresponding common primers were used. Reaction mixtures were as follows: 46 μL sterile distilled water, 12 μL tailed primer (100 μM), and 30 μL common primer (100 μM). Thermal cycling conditions comprised an initial denaturation at 95 °C for 15 min, followed by 10 touchdown cycles (94 °C for 20 s; annealing from 65 °C decreasing by 0.8 °C/cycle for 25 s), and 10 amplification cycles (94 °C for 20 s; 55 °C for 60 s).

2.5. Quantitative Reverse Transcription PCR (qRT-PCR) for Candidate Genes

In total, 12 candidate genes associated with grain yield-related traits in common vetch were selected for expression analysis. Contrasting phenotypic groups (low-bulk vs. high-bulk) were established using 10 accessions exhibiting extreme trait values (Table S3). At 20 days after flowering, we sampled both the branches and seeds of common vetch to validate candidate genes for PL and NG. At the late grain maturation stage, we collected near-mature seeds to validate candidate genes associated with GY and HGW. First-strand cDNA synthesis was performed using a cDNA synthesis kit (HiScript II), with gene-specific primers designed via GenScript (https://www.genscript.com.cn/tools/real-time-pcr-taqman-primer-design-tool?act=register_success, accessed on 3 April 2025) (Tables S5 and S6). Amplification reactions (20 μL total volume) were as follows: 2 μL cDNA template, 10 μL SYBR Green master mix, and 0.4 μL of each primer (10 μM). Technical triplicates were maintained for all assays, with relative gene expression quantified through the 2^−ΔΔCt method using actin as the endogenous control.

3. Results

3.1. Re-Sequencing for the 172 Common Vetch Accessions

A total of 172 common vetch varieties were mainly collected from China, Russia, and Southern Europe (Figure 1). In total, 3727.56 Gb of clean data were obtained by paired-end read data (PE150) with Q₃₀ about 92.0%. The average comparison rate between the sample and reference genome is 99.03%, with an average coverage depth of 15.18× and a genome coverage of 98.18%. The SNPs were filtered by minor allele frequency (MAF) < 0.05 and missing rate > 10% for further analysis. In total, 4,796,342 high-quality SNP markers were distributed across all six chromosomes, with an average of 2904.6 SNPs/Mb. Chromosomes 1 and 2 are the longest, at 324.8 Mb and 324.6 Mb, respectively, contained the highest SNPs (1,040,667 and 1,011,653 SNPs), and ranked among the top three in marker density. Chromosome 1 showed the second-highest density (3204.4 SNPs/Mb), while chromosome 4, despite its shorter length (290.1 Mb), exhibited the highest density (3184.4 SNPs/Mb), indicating extensive genetic variation in these regions. In contrast, chromosomes 3 (290.7 Mb) and 5 (272.5 Mb) displayed relatively lower SNP densities (2337.1 and 2613.9 SNPs/Mb, respectively), suggesting reduced polymorphism or potential assembly gaps. Notably, chromosome 6, the shortest at 148.7 Mb, contained only 28,588 SNPs but maintained a moderate density (2882.7 SNPs/Mb) comparable to larger chromosomes. The overall marker density (2904.6 SNPs/Mb) provides sufficient resolution for GWAS and MAS breeding, especially in high-density regions. These findings establish a robust foundation for dissecting agronomic trait architectures (

Table 1. The SNPs used for the GWAS by re-sequencing in common vetch.

Chromosome	Number of SNPs	Chromosome Length (Mb)	Marker Density (Marker/Mb)	MAF Mean	MAF Range
1	1,040,667	324.8	3204.4	0.27	0.05–0.50
2	1,011,653	324.6	3117.0	0.28	0.05–0.50
3	679,356	290.7	2337.1	0.20	0.05–0.47
4	923,806	290.1	3184.4	0.27	0.05–0.50
5	712,272	272.5	2613.9	0.28	0.05–0.50
6	428,588	148.7	2882.7	0.30	0.05–0.50
all	4,796,342	1651.3	2904.6	0.29	0.05–0.50

MAF: minor allele frequency.

and Table S2).

3.2. Population Structure Analysis for the 172 Accessions

The population analysis was conducted with K values ranging from 2 to 10, and the lowest cross-validation error (CV error = 0.58) was observed at K = 4, indicating that dividing the population into four distinct subpopulations named Subpop1-Subpop4. The genetic diversity analysis revealed that Subpop1 (a mix of Chinese and Russian origins) had the highest nucleotide diversity (π = 0.236), while Subpop2 and Subpop3 had nucleotide diversities of 0.165 and 0.152, respectively. Subpop4 (primarily of Chinese origin) exhibited the lowest nucleotide diversity (π = 0.112). These results reflect the influence of geographic origin on genetic differentiation. Among them, Subpop1 comprised 95 accessions, primarily from China (37) and Russia (28); Subpop2 consisted of 50 accessions, mainly from China (48). Subpop3 consisted of 15 accessions, largely from China (13); Subpop4 consisted of 12 accessions, largely from China (10), and two accessions from Australia and Poland. The population structure was further validated through NJ-tree construction and PCA analysis. The findings highlight significant genetic structure linked to geographic origin, providing valuable insights for breeding programs (Figure S1).

3.3. Phenotype Analysis and GWAS for Six Yield-Related Traits in Common Vetch

All 172 accessions were evaluated for six grain yield-related traits across four environments, including NB, PL, NP, NG, HGW, and GY. In this study, various traits exhibited distinct patterns of variation. NB displayed a mean of 10.1, with a standard deviation (SD) of 0.549 and a coefficient of variation (CV) of 0.054, indicating relatively stable branching across all accessions. Regarding pod-related traits, the average PL was 49.4 mm, with an SD of 5.87 mm (CV: 0.119), while NP had a mean of 44.5, an SD of 5.18, and a CV of 0.116. Among the 172 accessions, the average number of NG was 6.39, with an SD of 0.483 (CV: 0.0756). For HGW, the mean value across all accessions was 0.8 g, with an SD of 0.571 g (CV: 0.119). Notably, GY, a complex and comprehensive trait, is significantly influenced by both environmental and genetic factors. In this study, the average GY for the 172 accessions was 2.32 t·hm⁻², with an SD of 0.723 t·hm⁻² (CV:0.312), reflecting substantial variability (Table S1, Figure S2). Table S3 shows the ANOVA results, indicating significant effects (p < 0.01) of genotypes, environments, and genotype × environment interactions on each trait (Table S3). The H_b² estimated for PL, NG, NP, NB, HGW, and GY were 0.68, 0.65, 0.59, 0.67, 0.56, and 0.57, respectively. NB and PL showed significant (p < 0.01) and positive correlations with NP (r = 0.29 and 0.54), and PL showed a significant (p < 0.01) and positive correlation with NG (r = 0.31). In addition, PL and NP showed a significant (p < 0.01) and false positive correlation with HGW (r = −0.56 and −0.34), whereas GY showed a significant (p < 0.01) and positive correlation with HGW (r = 0.45).

Association mapping was performed using the mean values of the six yield-related traits across all four environments. In total, 38 significant loci were detected. Specifically, 11 loci associated with GY were found on chromosomes 1, 3, 4, and 6, accounting for 11.6% to 20.6% of the phenotypic variances (PVEs). For HGW, six loci were identified on chromosomes 1, 3, 4, and 5 and explain 11.6% to 20.3% of the PVEs. Two loci associated with NB were mapped on chromosome 6 and contribute 11.7% to 14.9% of the PVEs. Eight loci related to NG were located on chromosomes 1, 2, 3, 4, and 6, with explained PVEs ranging from 11.6% to 17.1%. Additionally, three loci for NP were distributed on chromosomes 4 and 5 and explain 11.7% to 16.5% of the PVEs. For PL, eight loci were identified on chromosomes 1, 2, 3, 4, and 6 and account for 11.6% to 17.5% of the PVEs (

Table 2. The loci identified by GWAS for six yield-related traits in 172 common vetch accessions.

QTL	Trait	Chromosome	Physical Interval (Mb)	p-Value	PVE (%)
qGY1.1	GY	1	48.1–48.5	7.4–7.9	13.1–14.2
qGY1.2	GY	1	164.7–166.7	7.5–8.6	13.6–16.5
qGY2.1	GY	2	283.9–284.6	7–11.2	11.6–20.6
qGY3.1	GY	3	165.7–166.2	7.0–7.4	11.6–13.2
qGY3.2	GY	3	278.2–278.2	7.4–7.5	13.1–13.5
qGY3.3	GY	3	234.0–235.2	7.0–8.1	11.6–14.9
qGY4.1	GY	4	71.6–72.6	7.2–10.3	12.5–18.5
qGY4.2	GY	4	141.1–142.3	7.0–7.2	11.7–12.5
qGY4.3	GY	4	178.0–178.5	7.1–9.5	11.9–17.5
qGY6.1	GY	6	71.8–71.8	7.1–7.1	11.9–11.9
qGY6.2	GY	6	119.0–120.1	7.1–9.1	11.9–17.1
qHGW1.1	HGW	1	244.5–245.8	7.0–7.7	11.6–14
qHGW3.1	HGW	3	189.4–190.9	7.0–8.3	11.6–15.3
qHGW4.1	HGW	4	71.9–72.3	7.1–7.9	11.9–14.2
qHGW4.2	HGW	4	132.6–112.8	7.0–7.4	11.6–13.2
qHGW4.3	HGW	4	180.0–180.0	7.7–11.1	14.2–20.3
qHGW5.1	HGW	5	266.7–268.7	7.3–7.3	12.8–13
qNB6.1	NB	6	57.7–58.2	7.0–7.5	11.7–13.5
qNB6.2	NB	6	142.8–146.8	7.1–8.1	12–14.9
qNG1.1	NG	1	48.1–48.5	7.0–7.3	11.6–12.8
qNG1.2	NG	1	285.0–292.8	7.3–7.6	12.8–13.8
qNG2.1	NG	2	187.5–187.9	7.7–7.8	14.2–14
qNG2.2	NG	2	245.7–247.7	7.9–9.1	14.9–17.1
qNG3.1	NG	3	196.0–196.6	7.0–8.5	11.6–16
qNG3.2	NG	3	276.0–277.2	7.0–7.3	11.6–13
qNG4.1	NG	4	65.3–72.3	7.0–8.4	11.6–15.3
qNG6.1	NG	6	102.4–104.5	7.0–7.7	11.7–14
qNP4.1	NP	4	81.9–82.2	7.0–8.6	11.7–16.5
qNP5.1	NP	5	146.6–148.5	7.1–7.4	11.9–13.2
qNP5.2	NP	5	193.2–193.2	7.2–7.2	12.5–12.5
qPL1.1	PL	1	108.1–108.9	7.2–9.5	12.5–17.5
qPL2.1	PL	2	245.7–247.7	7.1–7.1	11.9–11.9
qPL3.1	PL	3	130.5–132.0	7–7.1	11.6–11.9
qPL3.2	PL	3	261.2–261.2	7.5–9.2	13.6–17.1
qPL4.1	PL	4	65.8–67.1	7.1–8.5	11.9–16
qPL4.2	PL	4	85.1–85.8	7–7.8	11.6–14
qPL4.3	PL	4	101.8–102.6	8.0–9.4	15.2–17.1
qPL6.1	PL	6	104.6–105.6	7.7–8.4	14.2–16

NB: Branches per plant; PL: Pod length; NP: Number of pods per plant; NG: Number of grains per pod; HGW: Hundred-grain weight; GY: Grain yield; QTL: Quantitative trait loci.

, Figure 2).

3.4. Twelve Candidate Genes for Grain Yield Traits Were Identified in This Study

In total, 12 candidate genes encoding auxin response factor, F-box repeat protein, gibberellin receptor, serine/threonine-protein kinase-like protein, and cellulose synthase-like protein were identified and validated by qRT-PCR. qRT-PCR was performed to validate the expression levels of the candidate genes in bulk with extreme phenotypic traits. Significant differences in expression were observed for 12 candidate genes (

Table 3. The candidate genes identified for grain yield-related traits in this study.

Candidate Gene	QTL	Annotation
jg55197	qGY3.2	Auxin response factor 5
jg39866	qGY1.1	Cellulose synthase Acatalytic subunit4
jg51171	qPL3.1	Cellulose synthase-like protein
jg46961	qHGW1.1	F-box/FBD/LRR-repeat protein
jg57145	qNG3.2	F-box/kelch repeat protein
jg32764	qHGW4.2	Gibberellin receptor
jg33049	qGY4.1	GTP-binding protein BRASSINAZOLE INSENSITIVEPALEGREEN2
jg44419	qGY1.2	Chitin-inducible gibberellin-responsive protein
jg21506	qHGW5.1	Probable cellulose synthase Acatalytic subunit 3
jg30806	qPL4.2	Serine/threonine-protein phosphatase
jg2488	qPL6.1	Granule-bound starch synthase
jg10056	qNG2.2	Serine/threonine-protein kinase-like protein

). Table S4 shows the accessions for extreme bulk used for expression analysis. Table S5 shows the primer of candidate genes used for analysis expression analysis. The candidate genes jg55197, jg32764, jg39866, jg10056, jg2488, and jg30806 exhibited 3.20–4.56-fold higher expression, respectively, in accessions with larger corresponding traits. The candidate genes jg46961, jg44419, jg51171, jg21506, jg33049, and jg57145 exhibited 2.59–4.63-fold higher expression, respectively, in accessions with lower corresponding traits (Figure S3). Among these, jg55197 (located at qGY3.2) encodes the auxin response factor. The candidate gene jg32764 for qHGW4.2 is encoding the gibberellin receptor, while jg44419 (from qGY1.2) produces the gibberellin-responsive protein. At qGY1.1, jg39866 encodes the cellulose synthase catalytic subunit, and jg51171 (at qPL3.1) encodes the cellulose synthase-like protein. Similarly, jg21506 (located at qHGW5.1) also encodes the cellulose synthase catalytic subunit. In addition, jg30806 (at qPL4.2) encodes the serine/threonine-protein phosphatase, and jg10056 (from qNG2.2) encodes the serine/threonine-protein kinase-like protein. The candidate gene jg2488 (located at qPL6.1) encodes granule-bound starch synthase 1 (chloroplastic/amyloplastic). Additionally, jg46961 (at qHGW1.1) encodes the F-box/FBD/LRR-repeat protein, and jg57145 (from qNG3.2) encodes the F-box/kelch-repeat protein. Finally, jg33049 (located at qGY4.1) encodes the GTP-binding protein (brassinazole insensitive pale green).

3.5. Development and Validation of Kompetitive Allele-Specific PCR (KASP) Markers

To facilitate the MAS breeding in common vetch, six SNPs exhibiting stable or elevated PVEs were selected for conversion into KASP markers. Table S7 shows the details of the KASP markers, whereas the genotype of Kasp-NB6.2 and NB for the 172 common accessions were shown in Table S8. However, only the SNP located at qNB6.2 was successfully converted into a functional KASP marker and named Kasp-NB6.2. Genotypic consistency analysis revealed that Kasp-NB6.2 demonstrated over 85% concordance with re-sequenced data, confirming their accuracy in reflecting true genotypes (Tables S7 and S8). To assess practical utility, Kasp-NB6.2 was evaluated in the 172 diverse panel. The accessions carrying the favorable allele (AA, 81 lines) exhibited significantly higher NB (10.4) compared to those with the unfavorable allele (GG, 58 lines with mean NB of 9.7) (p < 0.05) (Figure 3). These results indicate that the Kasp-NB6.2 could be used to elucidate genetic mechanisms underlying common vetch grain yield traits, demonstrating its potential utility in common vetch MAS breeding.

4. Discussion

Most studies focus on the cultivation and application of common vetch [26,27]. To date, limited research has been conducted on the genetic analysis of yield-related traits in common vetch, with no molecular markers currently available for MAS breeding. Significant progress will be made in common vetch genetic research following the release of the first chromosome-level vetch reference genome in 2022 [18], alongside advancements in genotyping technologies and the application of KASP markers. This study utilizes 172 common vetch accessions to investigate the genetic architecture of six grain yield traits by GWAS [28,29].

Population structure analysis revealed that the 172 accessions were primarily divided into four subgroups. Subpop1 is primarily from China and Russia, whereas Subpop2, Subpop3, and Subpop4 are also largely from China. These findings align with previous studies demonstrating that Chinese common vetch varieties are fundamentally divided into two or three distinct groups. Historical analysis suggests that the divergence of vetch populations occurred approximately 100,000 years ago, with the two groups cultivated in China undergoing independent domestication processes. To ensure the reliability of association mapping results, we implemented stringent genetic background control measures in the MLM (PCA+K) to prevent spurious associations [30,31,32,33].

Grain yield-related traits of common vetch are typical quantitative inheritance and controlled by multiple minor genes [7]. In total, 38 loci significantly associated with six yield-related traits were identified in this study. Until now, there have been no studies on the genetic mechanisms related to grain yield traits in common vetch, nor have there been any clear reports on relevant genetic loci [11,15,30]. Moreover, this study utilized resequencing and derived SNP markers, which cannot be directly compared with traditional markers (SSR) identified in previous studies [11]. Therefore, all 38 yield-related genetic loci detected in this study may be novel findings.

Additionally, several loci with multi-effects were identified across different chromosomes. For instance, qGY1.1 and qNG1.1 are co-located on chromosome 1 (48.1–48.5 Mb), while qNG2.2 and qPL2.1 are found in the 245.7–247.7 Mb region of chromosome 2. On chromosome 3, qNG3.2 and qGY3.2 are positioned within the 276.0–278.2 Mb region. Moving to chromosome 4, qNG4.1 and qPL4.1 are located between 65.3 and 72.3 Mb, and qGY4.1 and qHGW4.1 are situated within the 71.6–72.6 Mb interval. On chromosome 6, qNG6.1 and qPL6.1 are found in the 102.4–105.6 Mb region, while qNB6.2 is nearly with qGY6.2. Overlapping QTLs (e.g., qGY1.1/qNG1.1 at 48.1–48.5 Mb) suggest pleiotropic regulation. In Medicago sativa, an orthologous region harbors MtNST1, a NAC transcription factor that coordinately regulates pod development and branch formation through auxin transport modulation [34]. The candidate gene jg55197 (auxin response factor) in qGY3.2 is similar to MtARF7, in which Medicago truncatula controls seed filling by activating cell expansion genes while suppressing branching through TCP transcription factors. This mechanistic link explains the negative correlation between HGW and NB observed in our study [35].

NB, PL, NP, NG, and HGW are all critical components of GY. In addition to these traits, GY is also influenced by a combination of factors such as agronomic traits (e.g., plant height), disease resistance, stress tolerance, and nutrient and water use efficiency. Furthermore, heritability analysis revealed that GY has an H_b² of 0.57, significantly lower than that of NG (0.65) and PL (0.68), indicating that it is regulated by unmeasured factors such as photosynthetic efficiency. Therefore, while composite traits like GY overlap with yield components and agronomic traits, they also include independently inherited traits. For example, qGY3.2 (Chr3: 277.2 Mb) does not overlap with loci of other traits and has the highest contribution rate (PVE = 20.6%), demonstrating the existence of independent genetic regulation. Similar phenomena have been observed in other species. In this study, we selected both comprehensive traits (GY and HGW) and yield component traits (NB, PL, NP, NG) to provide insights into common vetch breeding at both the holistic and individual trait levels.

The expression analysis was conducted using bulk composed of varieties with extreme trait values. Consequently, in selecting candidate genes, we placed particular emphasis on those coding genes that exhibited no differences in their coding regions but harbored variations within their promoter regions. In total, 12 candidate genes were identified and validated through qRT-PCR. Among these, jg39866 (for qGY1.1), jg51171 (for qPL3.1), and jg21506 (for qHGW5.1) encode cellulose synthase-like proteins, which are essential for cellulose synthesis, determine plant morphology, and enhance mechanical resistance. The genes jg46961 (for qHGW1.1) and jg57145 (for qNG3.2) encode the F-box repeat proteins, which regulate hormone signaling, photomorphogenesis, and flowering time. Their substrate specificity ensures precise control over developmental transitions and stress adaptation [36,37]. Additionally, jg30806 (for qPL4.2) and jg10056 (for qNG2.2) encode the serine/threonine-protein phosphatases, which play crucial roles in cell-cycle progression, signal transduction, and hormone signaling by modulating protein activity [38]. The gene jg2488 (for qPL6.1) encodes the GBSS, a key enzyme bound to starch granules. It catalyzes the elongation of amylose via α-1,4-glucan chains using ADP-glucose, making it critical for starch biosynthesis in storage organs [39,40]. Furthermore, jg55197 (for qGY3.2) encodes the auxin (IAA) response factor, while jg32764 (for qHGW4.2) and jg44419 (for qGY1.2) encode the gibberellin receptors. The gene jg33049 (for qGY4.1) encodes brassinazole insensitive pale green, a protein associated with brassinosteroid signaling. IAA are vital plant hormones that regulate seed germination, growth, flowering, fruit ripening, and leaf abscission. GA promotes stem elongation, breaks seed dormancy, stimulates germination, and enhances fruit development. BR, a relatively newer class of plant hormones, promotes cell elongation and division, enhances stress resistance (e.g., disease and cold tolerance), and may function synergistically with IAAs and GAs [41,42,43,44,45,46]. jg32764 (gibberellin receptor) showed higher expression in high-HGW accessions. This aligns with soybean GID1b-overexpression lines where 100-seed weight increased by 22% via upregulation of α-amylase genes [47]. Conversely, jg33049 (BZR1 homolog) was upregulated in low-yield genotypes, potentially reducing photosynthetic efficiency by repressing chlorophyll biosynthesis genes as reported in pea BR mutants [48].

In 2022, the first genome of common vetch was released. The genome is still being refined, which poses significant challenges for gene cloning and genetic mechanism elucidation in common vetch. Currently, most of the segments identified through GWAS are within the 1–8 Mb range, containing a multitude of high-confidence genes within these genetic intervals. Therefore, it is challenging to definitively identify the target genes solely through GWAS segments, annotation information, and qRT-PCR. The next steps include (1) configuring RIL populations in accordance with GWAS results to identify stable and highly effective target segments; (2) constructing multiple near-isogenic lines and derived populations for the target segments, using methods such as KASP markers for finer localization; (3) narrowing down the segments to a 200–500 Kb range, and then combining annotation information, RNA-seq, and qRT-PCR to identify and clone the target genes; and (4) attempting to further validate the function of the target genes through transgenic and gene editing. Application of high-throughput available molecular markers is a key point for MAS breeding and has proven successful. Recently, KASP offers an effective and low-cost approach. Conversion of SNPs into KASP markers could pave the way for common vetch MAS breeding. In this study, only one SNP located at qNB6.2 was successfully converted into the KASP marker Kasp-NB6.2 and validated in the 172 diverse accessions. In addition, genotypic consistency analysis revealed that Kasp-NB6.2 demonstrated over 85% concordance with re-sequenced data, confirming their accuracy in reflecting true genotypes (Tables S7 and S8). Additionally, accessions with more favorable alleles and superior grain yield-related traits, such as GLF304, GLF310, GLF311, GLF313, GLF315, GLF345, GLF369, GLF529, HZMC1353, HZMC1363, and HZMC1367, are recommended as parental lines for improvement of grain yield-related traits in common vetch.

Among the 172 common vetch accessions we collected, the majority originate from China, Russia, West Asia, Southern Europe, and Eastern Europe, but there is a lack of materials from North America and Latin America. The molecular markers we developed can be effectively used for MAS breeding in materials primarily from China and Russia, but their applicability in materials from other regions with potential population structures remains uncertain. However, since the origin of common vetch is in Southern Europe and Western Asia, the materials we collected still exhibit high genetic polymorphism. Common vetch in regions such as the Americas likely spread from these areas, so the markers we have validated should theoretically be applicable there as well. In the future, we will collect additional common vetch germplasm from the United States, Canada, Mexico, and other regions and conduct further resequencing and validation of molecular markers to provide references for common vetch breeding.

5. Conclusions

In this study, GWAS for grain yield-related traits were conducted in 172 common vetch accessions, and 38 significant loci were identified with explaining 13.3–31.7% of the PVEs. In addition, 12 candidate genes were identified for further research. Furthermore, Kasp-NB6.2, which can be utilized in common vetch breeding, was developed and validated. Our study uncovers the genetic architecture of grain yield-related traits in common vetch and provides genetic loci, available KASP markers, and outstanding accessions for common vetch molecular improvement.