NON-LENAR MODELS APPLIED TO GROWTH DATA OF CROSSBRED BOVINES WITH WAGYU

Abstract

ABSTRACT

The objectives of the present study were to select and adjust a non-linear model (MNL) to the growth of crossbred bovines Wagyu-Brahman under grazing conditions in the humid tropic of Gualaca, Panama. About 173 weights of Wagyu-Brahman (F₁) and Wagyu-European-Brahman (3R) bovines were considerate. Evaluated models were: Brody, Logistic, Von Bertalanffy and Gompertz. The estimated growth parameters were: adult weight (PAD) and growth rate (TM); and the growth indicators were: age (EPI), weight at inflexion point (PPI), except for Brody, and maturity grade at 540 days (GM540d). Selection of the model was based on the mean prediction error (EPP), variance of the predictive error (VEP), coefficient of multiple determination (R²), adjusted coefficient of multiple determination (R²_aj), Akaike Information Criterion (AIC), mean square error (CME), Durbin-Watson test (D-W) and errors distributions. Brody model showed best adjustment, followed by Von Bertalanffy. With Brody, values of AIC, VEP and EPP of the F₁ were 995,04; 1253,8 and -9,45129, respectively; and 1439,62; 1438,9 and -6,14342 of the 3R. With Brody R² and R²_aj for F₁ and 3R were >0,910. CME were 1348,3 and 1482,9 for F₁ and 3R, respectively. D-W test did not showed autocorrelations among residuals at the four models. With Brody values of PAD and TM in F₁ were 354,5±13,0 Kg and 0,00299±0,000349 kg.kg^-1.day^-1 and in 3R were 363,1±11.2 kg and 0,00287±0,000278 kg.kg^-1.day^-1, respectively. With Von Bertalanffy EPI, PPI and GM540d were 86,2 days, 100,2 kg and 90,3% for F₁ and 94,7 days, 103,0 kg and 88,8% for 3R, respectively. Brody model was the best and describes slow TM at differences of the sigmoidal curves reported in template weathers.