Grouping of genotypes according to environmental adaptation

In a set of genotypes, in general, several patterns of environmental adaptation may occur, characterized by different combinations of distinct levels of average yield and sensitivity of response in unfavorable and favorable environments. The method of segmented linear regression (Silva & Barreto, 1985; Silva, 1995a, 1995b) allows the identification and discrimination of these distinct genotype behaviors, through inferences related to the parameters of a linear model. This paper presents an extension of this method for grouping genotypes according to the similarity of the estimates of the parameters of adaptability of the segmented linear model. This method of grouping may be convenient when the characterization and classification of the genotypes according to the distinct pattertis of adaptation is desired. The method is based on a dissimilarity index provided by the F statistics for testing the hypothesis of homogeneity of regression lines. The method is illustrated through its application to data of an experiment of comparison of upland rice cultivars.