Identification of QTNs, QTN-by-Environment Interactions, and Their Candidate Genes for Rice Grain-Related Traits across Years using 3VmrMLM

Abstract Background Grain shape and grain weight are major determinants of rice yield and quality. Although numerous loci controlling these traits have been identified, their stable and environment-dependent genetic components remain incompletely understood. This study aimed to dissect the genetic architecture of rice grain-related traits across years using a multi-locus 3VmrMLM algorithm. Results A globally diverse panel of 413 rice accessions, genotyped with 1,974,377 high-quality single-nucleotide polymorphisms, was evaluated for grain length (GL), grain width (GW), grain length-to-width ratio (GLWR), and thousand-grain weight (TGW) across years. 3VmrMLM, along with MLM and FarmCPU, was used to analyze their phenotypes and best linear unbiased prediction values. A total of 407 nonredundant quantitative trait nucleotides (QTNs) were identified, almost each explaining small proportion of phenotypic variation. Of these, 116 were consistently detected across at least two analytical datasets, while 44 previously reported grain-size genes (e.g., GS3 , GW5 , GS2 , GW7 , and GS9 ) were located near the QTNs detected by IIIVmrMLM. Multi-year joint analysis further identified 10 significant QTN-by-environment interactions (QEIs) for GL, GW, and GLWR, but none for TGW. Integrating common QTN/QEI positions, developmental transcriptome data, functional annotation, and haplotype analysis prioritized 14 new candidate genes for grain size or weight. Among them, OsFKII/FRK2 , OsPCF5 , and OsSWN3/OsNAC31 were supported by multiple lines of evidence, whereas PHYA showed significant haplotype and haplotype-by-year interaction effects on GW. Conclusions Multi-locus GWAS identified both stable QTNs and QEIs associated with rice grain traits across years. These QTNs, QEIs, and prioritized candidate genes provide valuable genetic resources for functional validation and for marker-assisted and environment-aware breeding.