Interpretation of genotype × environment interaction for grain yield of barley using the GGE biplot method

The identification of the most favorable cultivar(s) with high yield and stable performance is usually done based on the analysis of the genotype × environment (GE) interaction. The yield stability of ۱۶ barley lines with two check varieties was studied in a randomized complete block design with four replications across three years at five locations in a multi-environment trial layout. The dataset was analyzed with a GGE (genotype main effect (G) + GE interaction) biplot method. Results indicated that the first two principal components (PCs) explained ۸۱, ۷۸ and ۷۱% of the GGE sum of squares for ۲۰۱۷, ۲۰۱۸ and ۲۰۱۹ growing seasons, respectively. According to the average environment coordinate abscissa, G۲, G۱۳ and G۱۸ were the best genotypes in terms of grain yield in years ۲۰۱۷ and ۲۰۱۸ while genotypes G۲, G۷ and G۱۴ were the highest yielding genotypes in ۲۰۱۹. When both yield performance and stability were considered simultaneously, the G۲ and G۱۳ genotypes in ۲۰۱۷ and G۲, G۸ and G۱۳ in ۲۰۱۸, were closer to the ideal genotype. In ۲۰۱۹, G۲, G۷ and G۱۴ were the best in terms of grain yield and stability. In the "which-won-where pattern", the five locations in ۲۰۱۷ fell into four sectors with different winning genotypes as G۲, G۵, G۱۴ and G۱۳. In ۲۰۱۸, the five locations fell into three sectors in which G۲, G۴ and G۱۷ were the highest yielding genotypes while in ۲۰۱۹, locations were positioned in four sectors and G۲, G۷, G۱۰ and G۱۳ were chosen as the winning genotypes. However, for practical use of the “which-won-where” pattern, the mean performance of genotypes over three years in the five test locations was taken into account. Although the results revealed six mega-environments, by neglecting small differences, we can assume only one mega-environment in which G۲ (the check variety Khorram) was the best performing genotype.

Keywords:

Barley , GGE biplot , Grain Yield , Multi-environment trials , Singular value decomposition