The objectives of this study were to verify the presence of the genotype by environment interaction and to compare different models to fit Devon cattle post weaning gain standardized to 345 days, via reaction norms obtained by random regression. Data from 14,973 calves collected by the beef cattle improvement program from 1980 to 2005 were used. The INTERGEN program was used for data analysis. First, a standard animal model (AM) was fitted to serve as comparison basis and to provide estimates of the environmental gradient, based on contemporary group (CG) deviations and, next, the reaction norms hierarchical models with homogeneous (RNHM) and heterogeneous (RNHMH) residual variances were run. In these last two models, fixed effects of age of dam and age of calf at yearling and random contemporary group and genetic effects (level and slope of animal's reaction norms) were included. For the RNHM estimates of residual variance, reaction norm level and slope additive genetic variances and their genetic correlation were, respectively, 272.02 ± 8.51; 340.38 ± 17.11; 0.119 ± 0.006 and 0.83 ± 0.01, while for the RNHHM were 401.96 ± 11.91; 177.86 ± 16.60; 0.076 ± 0.006 and 0.72 ± 0.02, respectively. The heritability asa well as the genetic variance increased on the environmental gradient, characterizing that there is genotype by environment interaction. Based on the Deviance Information Criterion, Bayes Factors and Pseudo Bayes Factors, the NRHM provided superior fit to the data, being the most appropriate model to consider the genotype by environmental interaction of this population.
Bayesian inference; beef cattle; genetic evaluation; heterogeneity of variance; random regression
