Evaluation of non-linear models for growth curve in Brazilian tropical goats.

In the present study, 1060 body weight-age records were measured on 300 Canindé goats. The records were collected from Rio Grande do Norte state, North-eastern part of Brazil, and used for evaluation of non-linear models (Brody, Von Bertalanffy, Logistic, and Gompertz) describing growth curve in goats. The non-linear models were compared applying mean squared error of prediction (MSEP), mean squared residue (MSR), mean absolute deviation (MAD), adjusted coefficient of determination (R2aj), and convergence percentage (%C) for determining the most appropriate model describing the growth curve in Canindé goats. The absolute growth rate (AGR) was estimated based on the adjusted model as a function of time to describe the growth rate of the animals, and the effect of sex on the curve parameters was assessed. The Brody model was the most indicated to represent the average growth curve of Canindé goats. Furthermore, the growth curves estimated by the models under study show that the Brody model displayed the best fit, indicating a greater association between observed and estimated weights, suggesting that it can be used to describe the growth behavior and development of the animals analyzed of both sexes.