Leveraging G$$\times$$E interaction to optimize multi-trait selection in lima bean

Climate variability and the increasing demand for resilient crops have intensified the need for breeding strategies that accelerate genetic gains. Lima bean, a protein-rich legume adapted to low-input farming systems, is a strategic crop for food security in tropical and semiarid regions. However, genotype-by-environment (G \(\times\) E) interaction remains a major obstacle to identifying responsive and stable genotypes. This study investigated G \(\times\) E interaction in 40 lima bean breeding lines cultivated across three contrasting Brazilian agroecological regions and identified superior ideotypes through integrated multi-trait selection approaches. Field trials were conducted in Piracicaba (SP), Teresina (PI), and Tianguá (CE), Brazil, using a randomized complete block design with three replications. Variance components were estimated by restricted maximum likelihood (REML), and genotypic values were predicted as best linear unbiased predictions (BLUPs) using mixed models. Genotype plus genotype-by-environment interaction (GGE) biplot analysis was then used to assess genotype performance and stability across environments. Simultaneous multi-trait selection employed the genotype-by-trait (GT) biplot and the desired-gain index, integrating agronomic, phenological, and yield-related traits into a unified breeding framework. The Likelihood Ratio Test detected significant G \(\times\) E interaction for all evaluated traits, revealing contrasting genotypic responses among environments. Combined multi-trait approaches detected lines L20, L27, L21, L06, and L13 as promising candidates because of their superior grain yield, suitable trait profiles, and enhanced phenotypic stability.