The objective of this work was to evaluate the genotype × environment interaction (GEI), for the adjusted weight at 365 days of age, in Polled Nellore cattle from Northeastern Brazil, by using reaction norm models through random regression in a Bayesian approach. The analyzed models included the fixed effect of cow's age (linear and quadratic), and the genetic additive and contemporary group random effects. The one-step reaction norm models with homogeneous residual variance (RNHM1sHm) yielded better data adjustments than the models with heterogeneous residual variance, and the standard animal model (AM). The estimates of genetic variance (64.41±13.24 kg² to 842.31±58.29 kg²) and heritability (0.11±0.01 to 0.63±0.02) increased as the environmental gradient improved. Spearman correlation values ranged from 0.69 to 0.99, among the classification of breeders with the highest breeding genetic values in AM and RNHM1sHm, in distinct environments, indicating changes in classification, particularly in the genetic values obtained by AM in high-quality environments. The identification of such degree of GEI requires that specific evaluation of individuals and herds are performed for environments of low, medium and high production.
environmental gradient; Bayesian inference; random regression; genetic value; zebu cattle
