Multivariate analysis of variance (MANOVA) was presented to provide a more efficient way to select superior genotypes from diallel crosses. Gardner and Eberhart (1966) model is one of the most commonly used methodologies for diallel analysis, in which parents (varieties) and heterosis effects are estimated, despite been presented by the authors as an univariate methodology. On the other hand, to get better populations and broaden the inference, plant breeders use to evaluate many traits. An extension of this model to k > 1 traits was adopted. Explicit expressions to least squares estimates of genetic effects and their variances were worked out. Sums of squares and products matrices were also given. Multivariate tests for the hypotheses of interest were presented. MANOVA on diallel crosses can be accomplished to estimate simultaneously the heterosis in several traits.
Diallel analysis; multiple traits selection; heterosis; Gardner and Eberhart model
