AUTHOR=Zhang Chunlei , Hong Huilong , Yuan Rongqiang , Zhang Shiyao , Gao Tianjiao , Yan Shuping , Lamlom Sobhi F. , Ren Honglei , Liu Zhangxiong , Wang Jiajun 

TITLE=Genome-wide association study and fine-mapping identify a major quantitative trait locus controlling hundred-seed weight in soybean

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2025.1716186

DOI=10.3389/fpls.2025.1716186

ISSN=1664-462X

ABSTRACT=BackgroundHundred-seed weight (HSW) is a critical yield component in soybean that directly influences productivity and seed quality. Despite its agronomic importance, the genetic architecture underlying natural variation in seed weight remains incompletely understood.MethodsWe conducted a comprehensive genome-wide association study (GWAS) using 554 globally diverse soybean accessions, comprising 453 Chinese varieties (81.8%) and 101 international accessions (18.2%) from 15 countries. Accessions were evaluated across three consecutive years (2022-2024) and genotyped with 78,050 high-quality single-nucleotide polymorphisms (SNPs).ResultsMixed linear model (MLM) analysis revealed a major QTL on Chr.20 that consistently explained the largest proportion of phenotypic variation across all environments. This QTL demonstrated exceptional temporal stability, maintaining genome-wide significance with peak -log10(P) values of 13.4, 12.1, and 10.2 across the three evaluation years. Fine mapping narrowed the critical interval to 493.69 kb containing 25 annotated genes. The lead SNP within Glyma.20G223200 explained 8-12% of phenotypic variance, while multi-SNP models incorporating five high-priority candidates cumulatively explained 14-18% of variance. Expression analysis of candidate genes revealed differential patterns between large-seeded and small-seeded varieties during seed development, with up to 32-fold expression differences.ConclusionsThe environmentally stable Chr. 20 QTL provides immediate opportunities for marker-assisted selection (MAS) in soybean breeding programs. Genomic prediction modeling suggests 35% greater genetic gain compared to phenotypic selection alone, supporting broad applicability for global soybean improvement programs.