ABSTRACT
The intense genotype x environment interaction present in many cultures make studies of adaptability to specific environments an important part of vegetal improvement programs. The different response of genotypes to favorable and unfavorable environments can be studied using methodologies based on three main strategies: analysis of variance, linear regression and non-parametric statistics. This work presents a new methodology for the study of genotype environment interaction, using principal components analysis, in order to facilitate the recommendation of specific genotypes to certain environments. This methodology, called centroid method, consists in the comparison of cartesian distance values between the genotypes and four ideal references using principal components. It differs from methods based in analysis of variance due to the aiming of genotypes according to environment variation and the facility of genotype identification, dispensing the analysis of several parameters as in methods based on regression. In order to exemplify the use of this method an experiment of Eucalyptus grandis planted in random blocks design and four environments was studied. Four clones of general adaptability were identified besides others of specific adaptation that can also be recommended aiming to capitalize the interaction effect. The results of this study were compared with the ones obtained through other methodologies by Eberhart e Russel (1966) and by Lin and Binns (1988) and allows to conclude that the centroid method was efficient in the identification of differentiated performance E. grandis clones; associated to the easiness of recommendation and ordering of the genotypes into specific adaptability groups.
Key words:
genotype environment interaction; Eucalyptus grandis; adaptability; principal components