Influence of dominance relationships on the estimation of dominance variance with sire-dam subclass effects.

Two data sets from the USDA Livestock and Range Research Laboratory were analyzed to study dominance variance and the influence of dominance relationships. The first consisted of 4,155 birth weight (3,884 weaning weight) records of inbred USDA Line 1 Herefords. The second consisted of 8,065 birth weight (7,380 weaning weight) records from a line-cross experiment with five lines. Two models were used. Both included fixed effects of year-sex of calf and age of dam, and covariates for calving date, inbreeding of animal, and inbreeding of dam. For the second set, additional covariates were line composition and heterozygosity coefficients. Random effects were direct and maternal additive genetic, maternal permanent environment, sire-dam subclass, and residual. Model 1 considered sire-dam subclasses unrelated. Model 2 related sire-dam subclasses with a parental dominance relationship matrix. Variance components were estimated using REML. Differences between estimates with Model 1 and 2 were unimportant except for dominance variance. For the first data set, estimates with Model 2 of relative genetic direct and maternal variances, direct-maternal correlation, permanent environment, and dominance variances for birth weight were .35, .13, -.02, .03, and .25, respectively, and they were .39, .11, .04, .06 and .14 for the second data set. For weaning weight, the first data set estimates were .20, .15, -.37, .19, and .11, respectively, and they were .16, .20, -.07, .18, and .18 for the second data set. Changes, decreases and increases, in estimates of dominance variances may be due to increased information from relationships and family types other than full-sibs. The assumption of unrelated sire-dam subclasses might not be appropriate for estimation of dominance variance in populations with many dominance relationships among sire-dam classes.