Eighteen oat genotypes were tested for genetic dissimilarity, with and without shoot disease control. The evaluated variables were grain yield, weight of one thousand grains, test weight, plant stature and days from emergence to flowering. Cannonical variables and clustering methods by Tocher's optimization test and the hierarchical method of nearest neighbor were employed, having as a measure of dissimilarity the general distance of Mahalanobis. The results showed a significant genetic dissimilarity among the genotypes, indicating the presence of genetic variability for the evaluated traits. Tocher's and nearest neighbor methods allowed the separation of genotypes in distinct groups, enabling the identification of potential parents for artificial crosses to obtain progenies with higher heterosis. The major traits contributing to genetic dissimilarity in our study were test weight and plant stature.
Environment; multivariate techniques; cluster methods
