Contrasting genetic architectures of mungbean yield components support trait-specific genomic breeding strategies

Mungbean ( Vigna radiata (L.) R. Wilczek var. radiata ) is an important low-input grain legume, but genetic improvement of yield remains constrained by the complex and environment-dependent nature of yield-related traits. We investigated the genetic architecture and genomic predictability of key yield components using a diverse mungbean panel evaluated across multiple environments through the International Mungbean Improvement Network (IMIN). Phenotypic analyses revealed substantial variation among genotypes. Seed weight (SW) showed relatively high heritability and stable performance across environments, whereas pods per plant (PPP) and seed yield exhibited stronger genotype × environment interactions. Genome-wide association studies revealed contrasting genetic architectures among traits. SW displayed recurrent QTL intervals on chromosomes 1, 5, and 7, whereas PPP and seed yield exhibited fewer stable associations and greater environmental dependence. Candidate gene analysis identified genes involved in carbohydrate metabolism, assimilate transport, developmental regulation, and cell wall modification, including two beta-amylase genes (Chr7.781 and Chr7.782) within a major SW-associated interval. Genomic prediction accuracy also varied among traits, with SW consistently showing the highest prediction accuracy across environments, while PPP and seed yield exhibited lower and more variable performance. Prediction accuracy reached a plateau at approximately 5,000 SNPs, and multi-trait models provided little improvement over single-trait models. Together, these findings demonstrate that mungbean yield components differ markedly in genetic stability and predictability. Stable traits such as SW are well suited for marker-assisted selection and KASP marker development, whereas more complex traits are better targeted using genomic selection, providing a practical framework for genomic-assisted mungbean breeding for diverse production systems.