Heteroscedastic Reaction Norm Models Improve the Assessment of Genotype by Environment Interaction for Growth, Reproductive, and Visual Score Traits in Nellore Cattle.

Abstract

Simple SummaryIn animal breeding, the different climatic conditions and production systems in beef cattle can result in the genotype by environment interaction. These interactions also need to be considered in genetic evaluations. In this study, five reaction norm models were tested on eleven traits related to growth, reproduction, and visual score measured in Nellore cattle breeding programs to check changes in the estimated genetic values of sires according to the selection environment. Using the best fitted statistical model, the presence of genotype by environment interaction was observed for some traits such as age at first calving, scrotal circumference, weaning to yearling weight gain, and yearling weight. For these traits, it is possible to select the best sires to increase productivity and reduce environmental sensitivity. Overall, the reaction norms trajectories for these traits seem to be affected by a non-linear component, and selecting robust animals for these traits is an alternative to increase production and reduce environmental sensitivity.The assessment of the presence of genotype by environment interaction (GxE) in beef cattle is very important in tropical countries with diverse climatic conditions and production systems. The present study aimed to assess the presence of GxE by using different reaction norm models for eleven traits related to growth, reproduction, and visual score in Nellore cattle. We studied five reaction norm models (RNM), fitting a linear model considering homoscedastic residual variance (RNM_homo), and four models considering heteroskedasticity, being linear (RNM_hete), quadratic (RNM_quad), linear spline (RNM_l-l), and quadratic spline (RNM_q-q). There was the presence of GxE for age at first calving (AFC), scrotal circumference (SC), weaning to yearling weight gain (WYG), and yearling weight (YW). The best models were RNM_l-l for YW and RNM_q-q for AFC, SC, and WYG. The heritability estimates for RNM_l-l ranged from 0.07 to 0.20, 0.42 to 0.61, 0.24 to 0.42, and 0.47 to 0.63 for AFC, SC, WYG, and YW, respectively. The heteroskedasticity in reaction norm models improves the assessment of the presence of GxE for YW, WYG, AFC, and SC. Additionally, the trajectories of reaction norms for these traits seem to be affected by a non-linear component, and selecting robust animals for these traits is an alternative to increase production and reduce environmental sensitivity.