Contribution of the heterogeneity of regression lines to the recommendation of new cultivars

Abstract

The objective of this work was to show the contribution of the analysis of the heterogeneity of regression lines to the recommendation of new cultivars by the Eberhart & Russell method. An experiment with ten rice genotypes, evaluated in eight environments, was used to illustrate the methodology. A randomized complete block design, with three replicates, was used. The sum of squares of genotype x environment (GxA) interaction was decomposed to evaluate the heterogeneity of regression lines and the accumulated deviations from linearity. The heterogeneity of regression lines was analyzed using the t‑test on the linear regression coefficients of the genotypes. The two orthogonal components of GxA interaction were significant. The heterogeneity analysis of regression lines made it possible to detect inconsistencies in genotype adaptability, which reduced the chances of misleading recommendations of cultivars. Genotypes were ranked according to the efficiency of the Eberhart & Russell methodology to explain the nature of genotypic performance over changing environments. The evaluation of the heterogeneity of regression lines contributes to more effective recommendations of new cultivars with the use of the Eberhart & Russell methodology.