Relative discriminating powers of GGE and Ammi models in the selection of tropical soybean genotypes

Abstract

Selection of crops is preceded by multi-locational testing in plant breeding; however, it becomes difficult for breeders to determine which genotypes should be selected in the presence of genotype by environment (GEI). Six
genotypes of soybean (Glycine max (L.) Merr.) were evaluated at ten locations in Nigeria for grain yield and stability. The analysis of variance revealed significant (P < 0.05) GEI effect. Mean grain yield of the soybean
genotypes ranged from 1148 kg ha-1 for genotype M351 to 1584 kg ha-1 for TGx 1448-2E. Ilorin in the southern guinea savanna of Nigeria was the most variable with high interaction principal component axes (IPCA); while
Bauch in the northern guinea savanna was identified as more stable location in evaluating the soybean genotype. Mega-environments and the best yielding soybean genotypes in each mega-environment were revealed by the GGE biplot analysis. Furthermore, TGx 1448-2E and TGx 1440-1E, were established as the most promising, and stable genotypes across the test locations. Stability model of GGE biplot was superior, effective and informative in mega-environment analysis compared to AMMI analysis.

Key Words: Glycine max, principal component analysis