Multiple centroid methodology to analyze genotype adaptability

Abstract

This study was carried out to propose a modification of the centroid method. The method is modified in the choice of the ideotype, to be defined according to the reseacher's interest, using the bi-segmented regression model, rather than based on experimental data, which limit the comparison of genotypes. To illustrate the method, one trait was simulated in 25 genotypes in 8 environments. For the simulations and statistical analyses the software package GENES was used. The multiple centroid is more flexible than the original centroid method, since it can be used according to the researcher’s objective and the desired recommendation strategy. It is also readily interpretable for recommendation and unambiguous. Besides, the number of genotypes of interest can be compared, making a detailed study possible, by separating the genotypes into recommendation classes.