Genomics of iron and zinc concentration, iron bioavailability, and yield in common bean: Multi‐locus GWAS, structural equation modelling, candidate gene prioritization and genomic selection

Abstract

Abstract Common bean ( Phaseolus vulgaris L.) is a nutritionally dense and widely cultivated legume that serves as a major source of protein, minerals, and dietary fiber for millions of people, particularly in sub‐Saharan Africa and Latin America. However, its production and utilization are constrained by biotic and abiotic stresses, prolonged cooking time, and the presence of bioactive compounds that limit iron bioavailability. Traditional breeding to improve these traits is slow and often yields limited progress. This study used genotyping‐by‐sequencing on 427 bean genotypes to explore the genetic basis of key nutritional and yield traits: iron (Fe), zinc (Zn), bioactive compounds, cooking time (COOKT, in minutes), hydration coefficient (HC), seed coat color, and yield. Through multi‐locus quantitative trait locus (QTL) analysis and structural equation modeling, the study identified 82 QTLs across all 11 chromosomes, including regions on Pv02 (where Pv is Phaseolus vulgaris chromosome), Pv05, and Pv11 that influence both Fe and Zn. Shared genomic regions were also found for Zn‐HC, COOKT‐HC, and yield‐polyphenols. Nine QTLs that may be involved in multiple traits indicate pleiotropic effects, though only two were beneficial across all. However, incorporating these markers into multi‐trait genomic prediction models did not significantly improve prediction accuracy. The model based solely on the genomic relationship matrix achieved the best performance. These findings highlight the complex genetic architecture of key traits and the necessity of customizing genomic prediction strategies to specific breeding objectives.