Nonlinear mixed models applied to broiler chickens performance data

Abstract

Nonlinear mixed models applied to broiler chickens performance data Due to the large market demand for chicken meat, there is great interest in research aimed at further improving the production efficiency of this product. In this context, the study of broiler performance assists in the process of optimizing meat production and facilitates the understanding of the needs of each growth phase until the poultry slaughtering age. Although it is common to use nonlinear models to describe the growth pattern of birds, it is not common to include random effects in these models, much less the combined modeling of the variables observed in the experiment. In this work, we adjusted the Gompertz, four-parameter logistic, von Bertalanffy, and Richards growth models with fixed and random effects in their parameters to describe the growth curve of 1080 Ross broilers. Additionally, we performed the joint modeling of body weight and feed consumption variables using mixed models. Additionally, we compared the adjusted models using the AICc and BIC information criteria. The results indicated that the fourparameter mixed logistic model was the most suitable for broiler performance data for both univariate and bivariate models.