Estudo de modelos não lineares para predição da curva de crescimento de bovinos da raça Nelore criados na região Sul do Brasil

Abstract

Aiming to study the growth of Nellore cattle raised in Southern Brazil, nine models were evaluated: Von Bertalanffy, Richards, Gompertz, Bianchini Sobrinho, Logarithmic Square, Brody 23, 24 and 45. The main objective was to evaluate which one promote better data adjustment for weight and age of these animals. Thus, were analyzed 156,561 observations of 27,019 animals borned between 1976 and 2008, weighed between 1976 and 2010, created in 67 municipalities throughout the states of Rio Grande do Sul, Paraná and Santa Catarina. The criteria used to select for the best model were the number of iteration used to reach convergence (NI); the mean square error (QMe); the error or mean absolute deviation (DMA) and the coefficient of determination (R2). Brody 45 model were chose who promoted the best fits, for males and females.