REGRESSÃO NÃO LINEAR EM CURVA DE CRESCIMENTO DE COTURNIX COTURNIX

Abstract

The aim of this work was to study non-linear regression models in animal growth curve studies. Nonlinear regression models have been shown to be suitable for describing growth curves, as they present parameters that can be interpreted biologically. These are mainly characterized by weight at maturity, which represents weight in adulthood, and growth velocity, or maturity rate, which represents a measure of precocity. Different non-linear models are presented in the literature to describe animal growth curves, and each of these models has its advantages and disadvantages. A database of 10 generations of Coturnix coturnix was used, totaling 572 females, from a lineage of meat quails. In the initial phase, up to the 42nd day of life, a feed with 26% crude protein was fed ad libitum. Birth weights (P1), weight at 7, 14, 21, 28 and 35 days were collected. For the non-linear regression analysis, the von Bertalanffy, Brody, Gompertz, Logistic and Richards models were used. To evaluate the best model that best fitted the data, the evaluators of Percentage of convergence (C%), Coefficient of determination (R²), Adjusted coefficient of determination (R²aj), Mean squared error of prediction (MEP), Criterion Akaike Information Criterion (AIC) and Schwarz Bayesian Criterion (BIC). Conclude that the best model depends on the species to be used, the most applied being the Gompertz and von Bertalanffy models.