Linear (linear-plateau, quadratic, quadraticplateau, square root) and non-linear (exponential and Michaelis-Menten) models to describe the ruminal microbial growth as a function of sucrose levels were evaluated. Is was used a database of 76 observations of microbial growth in roll tubes (mix of bacterial) containing final concentrations of 0.0; 0.375; 0.75; 1.50; 3.0; 6.0; 12 and 24g/L of sucrose. The microbial growth was evaluated by measuring optical density (OD - 600nm) at 0; 6; 12; 18 and 24 hours of fermentation. The evaluation of linear regression parameters of predicted and observed values indicated adequacy of all models in the prediction of microbial growth. The partition of the mean square prediction error did not allow identifying differences in relation to prediction quality of the models. However, the marginal growth as a function of sucrose supply was inadequate to models linear-plateau and quadratic (in low sucrose levels). The square root, Mitscherlich and Michaelis-Menten models can be used to describe biological responses to nutrients. However, differences are observed in estimates of substrates efficiencies, which implies differences in the recommendations of optimum levels of nutrients. In this sense, the choice of these models can be modulated by the level of nutrient supply desired.
marginal response; Michaelis-Menten; saturation kinetics