Spline functions and nonlinear models for modeling the growth of sheep in Northeast Brazil

Abstract

Modeling animal growth is essential to manage quality meat production systems at a low cost. This study aimed to compare the linear-linear, linear-plateau, and linear-linear-linear spline functions with the nonlinear models Brody, Richards, and von Bertalanffy to model the growth of crossbred sheep in the Northeast of Brazil. Weighing data from the genetic groups of meat sheep Dorper x Morada Nova, Dorper x Rabo Largo, and Dorper x Santa Inês were used. Knots in the spline fitting were considered unknown and estimated under the segmented nonlinear regression approach. Overall, splines were more accurate than traditional nonlinear models in describing the growth of sheep of all genetic groups. The linear-plateau spline was superior in predicting weight as a function of the age of sheep. The linear-plateau function also resulted in estimates of parameters (growth rates, time for animals to reach adult weight, and adult weight) with easier biological interpretations, which can be used in practice for decision-making for herd management.