The fitting of nonlinear models to growing pigs data

Abstract

Five nonlinear models were fitted to weight-age data from birth to 168 days collected in 127 Large White and 143 Landrace pigs, tested at Concórdia, SC, Brazil, during 1979 and 1980. The purpose of this experiment was to determine the best equation for fitting, computacional ease and biological interpretability of the parameters. The pigs were fed from 76 to 114 days with a ration containing 3,300 kcal/kg GE and 18% crude protein, and from 114 to 168 days a finishing ration with 3,250 kcal/kg GE and 16% crude protein. Both feed and water were available ad libitum. Weight data at birth at 21 days, and weekly from 60 to 168 days were used to fit the models. Based on the adjusted coefficient of determination (R²), residual mean squares and biological interpretability of parameter fitted, the Logistic model with variable inflection point in M, provided the best results for the present study. Two models provided poor estimates of observed weights both in the initial and final phase of the studied period. The highest rate of maturity was provided by the Logistic model. All models showed good adjustment. The best results were obtained with male swine independently of race.