Nine mung bean genotypes were evaluated for grain yield performance at four locations in 2016 main cropping season. The experiment was arranged in 3x3 lattice design with three replicates. Additive main effect and multiplicative interaction (AMMI) combined analysis of variance revealed that genotype, location and genotype x location interaction (GLI) significantly affected (p<0.01) grain yield of mung bean. The significant effect of GLI indicates the difference in ranking of genotypes across locations for yield performance as well as the variability of the test locations. In addition, the result indicates simultaneous consideration of both mean grain yield and stability. Both interaction principal component axis, IPCA1 and IPCA2 were highly significant (p<0.01) captured 71.60% and 22.78% of the GLI variability, respectively. AMMI and GGE comparison biplot enabled identification of stable and specifically adapted genotypes very well. Consequently, G3 (Black bean) was the most stable and the highest yielding genotype. Genotypes, G2, G5 and G9 were specifically adapted to Sekota; G6 and G8 adapted to Pawe; G1 and G7, adapted to Miesso location. Genotype plus genotype by environment interaction (GGE) biplot identified the presence of two mega-locations with their winning genotypes. Melkassa, Miesso and Pawe together formed one sector and G3 as their winning genotype, whereas Sekota alone formed the second sector with G9 as its favourable genotype. However, GLI should be exploited further to confirm those specifically adapted genotypes for possible future release.

Keywords: AMMI, Vigna radiata, GLI, Specific adaptability