Prediction of Sire Merit for Calving Difficulty

Abstract

Three mixed models were evaluated to identify the best linear unbiased predictors of sire merit for calving difficulty. The best model included relationships due to sires and maternal grandsires of bulls with progeny data and accounted for the unequal variance associated with parity of dams. Adding relationships and simultaneously adjusting for unequal variances by parity reduced the mean square error by 49%. About 90% of the reduction was due to the adjustment for unequal variances. This resulted in a gain in precision of individual sire estimates and in a higher effective heritability for calving difficulty. Addition of relationships alone made no significant change in the mean square error but increased the variation among predicted breeding values 43%.