Application of AMMI analysis in the assessment of yield stability in soybean

The objective of this work was to evaluate the influence of the genotype environment (GE) interaction on the grain yield of a soybean (Glycine max L.) lines group. Yield data from I I trials (environments) conducted in the State of Goias, Brazil were used. In each trial 18 genotypes were tested, from which four were commercial cultivars as checks. The statistical method was the AMMI analysis (additive main effect and multiplicative interaction analysis). A significant GE interaction pattern was captured only for the first principal AMMI axis, which explained 36% of the original square sum of the GE interaction, suggesting contamination of the classic GE interaction matrix by noise arising from unpredictable factors, assuring that AMMI analysis provides a better prediction of phenotypic responses than traditional methodologies. About yield stability, most experimental lines outstands (with low GE interaction) over checks. However, the checks, mainly the cultivar Conquista, showed higher yield averages. Among the experimental lines, the genotypes L-16, L-13 and L-14 appeared to be the most promising for cultivars recommendation.