The stratification of the crop region is a strategy used to control the effect of the genotype x environment interaction. This approach is based on results of an experiment conducted in several years and locations, in which the statistical analysis has difficulties originated from the incomplete data structure and variance heterogeneity of the experimental error. This paper reports results of a research on the magnitude and the nature of the genotype x environment interaction and on the convenience for the stratification of the common bean production region in Rio Grande do Sul State, Brazil, for the control of this interaction. Data from the Common Bean Yield State Trial carried out in 24 locations for eight years were evaluated. Joint analyses of each year, of two sub-groups of four and of a set of eight years, as well as cluster analyses, were performed using a methodology that circumvents the statistical difficulties. The results of the annual analyses revealed highly significant genotype x location interaction. This interaction was also significant in the set of the eight-years, but not significant in the two groups of four-years. The genotype x location x year interaction was highly significant in these three sets of years. The cluster analyses based on the eight years constituted sub-regions incoherent with the sub-regions formed by the annual analyses, which were not consistent among themselves. The results point out for the inadequacy of a regional division of the area of common bean production in Rio Grande do Sul.
Phaseolus vulgaris L.; variance components; environment stratification; cluster analysis
