Aplicação de modelos de normas de reação para estudo da interação genótipo x ambiente sobre pesos corporais em bovinos da raça Nelore

Abstract

The objective was to verify the existence of the Genotype x Environment Interaction (IGA) through the use of Reaction Norm models and their implication in the evaluation of breeders on weight performance indicators of Nelore cattle. We used 55444 and 41195 records of body weights at weaning (P120) and yearling (P450), respectively, from herds participating in the genetic improvement program of the National Association of Breeders and Researchers (ANCP). For each body weight (characteristic), one-step reaction norms models were used that assume the residual variance as homogeneous (MHNR1p_homo) and as heterogeneous (MHNR1p_het). Genetic parameters were obtained using a function of environmental gradients. Thus, Bayesian inference was used. using a Gibbs sampler adopting Markov chains with 600,000 cycles; burn in of 50,000 cycles and thinning of 20 samples, totaling samples of 30,000 cycles. For the diagnosis of Markov chains, the criterion of the Geweke test was used, adopting a significance level of 0.05. To choose the most appropriate model for each characteristic, the Deviance Information Criterion (DIC) was used. In models of reaction norms in which the residual heterogeneous assumption was assumed, they proved to be more adequate in both characteristics. Estimates of posterior means of heritability and correlation varied according to the environmental gradient. For P210, higher averages were obtained in the median gradients (approximately 0.32), with similar values in the extreme gradients (0.25 for gradient 1 and 0.20 for gradient 5). For P450, further heritability estimates were similar across gradients, ranging from 0.26 to 0.29. All subsequent means of genetic correlations for P210 between the gradients were greater than 0.80, indicating that, despite the differences between heritability in the gradients, the animals would be classified similarly between them and, consequently, without IGA influence. On the other hand, for P450, a posterior average of genetic correlation was observed, between gradients 1 and 5, equal to 0.77, evidencing the presence of IGA between more extreme environments. For the other combinations of gradients, all means were greater than 0.80. In this sense, for P450, the genetic evaluation model must consider the presence of IGA.