The aim of this work was to identify among the Brody, Von Bertalanffy, Gompertz, France, Logistic, Modified logistic and Dual-pool logistic models, the one that presents the highest quality of fit for cumulative gas production curve in sunflower and corn silages. The quality of fit was evaluated by coefficient of determination, residual mean square, graphic analysis of the observed and estimated curves, graphic analysis of dispersion of the studentized residual, average percentage error, relative efficiency and number of iteractions to converge. Brody, France and Dual-pool logistic models showed the largest values of coefficient of determination in both substrates, and the difference among them was considered worthless. They also showed the smallest values of residual mean square in sunflower silages, being the difference among them considered worthless. Brody and France models presented smaller residual mean square in corn silages. All the models showed positive dispersion of the residual in both substrates after 144 hours of incubation. The Brody model showed smaller average percentage error and number of iterations in both substrates. The Dual-pool logistic and France models showed higher relative efficiency, respectively, in sunflower and corn silages. Thus, the Dual-pool logistic model showed higher quality of fit to the gas production curve in sunflower and corn silages.
kinetics; in vitro; mathematical models; nonlinear models
