Análises de dados longitudinais em bovinos Nelore Mocho por meio de modelos não lineares

This study was carried out to estimate the best different nonlinear models to predict growth curve in Polled Nellore cattle. The models used (Brody, Von Bertalanffy, Gompertz and Logistic models) were fitted by Gauss-Newton method to weight-age data of birth to adult age of Nellore cattle from Brasilia - DF. The adjustment quality was showed by mean square error, adjusted determination coefficient, prediction error, absolute mean deviation residuals, asymptotic weight (A) and maturity rate (k) biological parameters, graphics of observed and estimated weights, of instantaneous and relative growth rate and absolute maturity. The convergence criteria were attained for all nonlinear models. Estimative mean for A and k was 543.40 kg and 0.00138 %/day; 402.00 kg and 0.00323 %/day; 378.00 kg and 0.00410 %/day; and 346.00 kg and 0.00657 %/day for Brody, Von Bertalanffy, Gompertz and Logistic nonlinear models, respectively. Data adjustment for weight-age of the animal provides information describing the growth curve and the future prognostic for animals managed under the same environmental situation. In the period analyzed, Brody nonlinear model was indicated to estimate the body weight of Polled Nellore.