Comparison of three methods used for the study of adaptation and phenotypic stability in the common bean (Phaseolus vulgaris L.)

The objective of the present work was to compare the acuity of 3 different methods of studying phenotypic stability for determining the variability in av. yield data of the common bean. Data from 12 preliminary yield trials of 76 common bean genotypes grown in 8 Brazilian locations were used. Two classical procedures based on linear regression analysis and the recent proposition of bisegmented linear regression were studied. Based on the proportion of estimates of coefficients of determination that were statistically significant or non significant for the genotypes studied, as well as on the magnitude of such estimates, it was concluded that the segmented linear regression procedure was slightly superior to the methods based on linear regression analysis. The 2 methods of linear regression showed similar results. The advantage of the bisegmented linear regression method was more evident when the behaviour of the genotypes in response to the environmental variation did not follow linearity. (AS)