Uso de modelos não lineares para descrever o crescimento do perímetro escrotal em touros da raça guzerá criados em pastoreio extensivo

Abstract

This study aimed to compare five non-linear models to describe the growth curve of the scrotal circumference (SC) in Guzerat bulls. Brody, Logistic, Gompertz, Richards and Tanaka models were evaluated by the iterative methods Gauss-Newton and Marquardt. Parameter A is the estimate of SC at maturity, B is an integration constant, k is the index of maturity and, for the Richards and Tanaka models, m determines the inflection point. In the Tanaka model, A is the indefinite length of SC, B e k adjust the shape and inclination of the curve. The models were evaluated by the parameters and standard error, the error mean square (EMS), the coefficients of determination (R2), the mean prediction error (MPE), the mean absolute deviation (MAD), the Durbin-Watson test (DW), graphs of residuals and the degree of computational difficulty. Richards model did not reach the convergence criterion. With the Gauss Newton method, the Tanaka model estimated values of 1.31 for parameter A, 0.04 for B, 26.36 for k, and 19.34 for the m parameter. Brody estimated the highest value for A (43.49 cm), followed by Gompertz model (38.85 cm) and Logistic (36.70 cm). The logistic model estimated the highest value for k (0.07 cm), followed by Gompertz (0.05 cm) and Brody (0.03 cm). The inflection point was estimated at 8.24 months for Gompertz, with 14.4 cm of SC, 14.5 months for Logistics, with 18.4 cm of SC and 19.3 months with 23.7 cm for the Tanaka model. The EMS and MAD were lowest for the Tanaka model, followed by the logistics model. The Tanaka model showed lowest MPE between 7 and 48 months. The Tanaka model, followed by the Logistic were more appropriate