Genomic prediction of lactation curves of Girolando cattle based on nonlinear mixed models
Knowledge of lactation curves in dairy cattle is essential for understanding the animal production in milk production systems. Genomic prediction of lactation curves represents the genetic pattern of milk production of the animals in the herd. In this context, we made genomic predictions of lactation curves through genome-wide selection (GWS) to characterize the genetic pattern of lactation traits in Girolando cattle based on parameters estimated by nonlinear mixed effects (NLME) models. Data of 1,822 milk control records from 226 Girolando animals genotyped for 37,673 single nucleotide polymorphisms were analyzed. Nine NLME models were compared to identify the equation with the best fit. The lactation traits estimated by the best model were submitted to GWS analysis, using the Bayesian LASSO method. Then, based on the genomic estimated breeding values (GEBVs) obtained, genomic predictions of lactation curves were constructed, and the genetic parameters were calculated. Wood's equation showed the best fit among the evaluated models. Heritabilities ranged from 0.09 to 0.29 for the seven lactation variables (initial production, rates of increase and decline, lactation peak, time to peak yield, persistence and total production). The correlations among GEBVs ranged from -0.85 to 0.98. The concordances between the best animals selected according to the selected traits were greater when the correlations between GEBVs for these traits were also high. Consequently, the methodology allowed us to identify the best nonlinear model and to construct the genetic lactation curves of a Girolando cattle population, as well as to assess the differences between animals and the association between lactation variables.