PREDICTION OF BREEDING VALUES AND DOMINANCE EFFECTS FROM MIXED MODELSWITH APPROXIMATIONS OF THE DOMINANCE RELATIONSHIP MATRIX

A simulation study was performed to evaluate the performance of mixed model equations for estimation of additive and dominance effects with two different dominance relationship matrices, one contained all relat ionships, assuming no inbreeding, and the other contained only dominan ce relationships between full-sibs. The latter is equivalent to a mode l containing additive effects and a full-sib or litter effect. In both models, the effect of inbreeding depression on the mean was accounted for by a regression on the inbreeding coefficient. The simulated trai t was controlled by 500 unlinked loci each with two alleles. Gene effe cts were equal at each locus and dominance was complete. Gene frequenc ies at each locus were either 0.2, 0.5 or 0.8. Phenotypic selection on performance was carried out over 10 generations. Both models performe d similarly at each gene frequency. Estimates of additive effects show ed no significant bias during the studied time span for either of the two models. Dominance effects were biased upwards for high gene freque ncies and downwards for low gene frequencies. Predictions of additive effects were sensitive to priors assumed for dominance variance. Prior s that were too large gave negative bias in genetic trend and vice ver sa.