Regression models grouping for genotype adaptability and stability analysis

Abstract

The objective of this work was to provide evidences of difference among regression models, obtained by the Eberhart & Russell method, for genotype adaptability and stability analysis, and to propose a methodology for grouping similar models. The test to verify the identity of models was used in the evaluation data of 14 corn genotypes in eight environments. Once the hypothesis of equality of regression models was discarded, model grouping was performed through estimating the mean square of the reduction (MSQRed) between pairs of regression models. After obtaining these values, the smallest one was selected and its significance was verified by the F‑test. If this MSQRed is not significant, the model pair related to it comprises the initial group. The models identity verification method can be successfully used for grouping linear regression equations, obtained by the Eberhart & Russell method, in order to study genotype adaptability and stability. The similar models grouping allow the formation of genotype groups with the same statistical behavior.