Random regression models with B-splines to estimate genetic parameters for body weight of young bulls in performance tests

Abstract

The objective of this study was to estimate genetic parameters for body weight of beef cattle in performance tests. Different random regression models with quadratic B-splines and heterogeneous residual variance were fitted to estimate covariance functions for body weights of Nellore and crossbred Charolais × Nellore bulls. The criteria -2 residual log-likelihood (-2RLL), Akaike Information Criterion (AIC), and consistent AIC (CAIC) were used to choose the most appropriate model. For Nellore bulls, residual variance was modeled with six classes of age, and direct additive genetic and permanent environment effects were modeled with quadratic B-splines with two and one intervals, respectively. For crossbred bulls, quadratic B-splines with one interval fitted direct additive genetic and permanent environment effects and nine classes of age were needed to fit residual variance. Pooling classes of age with up to 40% in difference of residual variances does not compromise the fit of the model. Heritability for body weight in performance tests are moderate (> 0.25, for crossbred bulls) to high (> 0.5, for Nellore bulls) and genetic correlation between weights over the test are also high (> 0.65). Then, selection of young bulls in performance test is an efficient tool to increase body weight in beef cattle.