Sensitivity of two metabolic models of dairy cattle digestion and metabolism to changes in nutrient content of diets

To better understand how specific nutrients influence predicted outputs by two metabolic nutritional models (MOLLY and NRC) a sensitivity analysis was completed. This analysis examined the responses of the nutritional models when nutrient content of the diets was changed. Three diets were formulated and simulated in two nutritional models. These same diets were then simulated after 10 g/kg of diet DM then replacing it with the same mass of neutral detergent fibre, acid detergent fibre, starch, lignin, crude protein (CP), soluble CP, rumen undegradable protein, fat, or ash. Sensitivity estimates were calculated as the percent change in response variable from the diet with added nutrient compared to the original diet. Approximately 0.75 of the sensitivity estimates were less than 0.01 indicating that most changes to single nutrient concentrations did not have large impacts on simulated outputs. The incidence of large sensitivity estimates varied between models. Within the confines of this study, both models were more responsive to nutrient changes that altered protein content than to nutrient changes that altered energy content of the diet. Both models had nutrient inputs that had little or no impact on the outputs evaluated in this study. However, neither model was completely unresponsive to changes in any nutrient; suggesting that accurate description of nutrient supply inputs for these models is important and should be considered when evaluating predictions from the models. By coupling the knowledge of how specific nutrient inputs influence model predictions with desired animal outcomes, nutritionists that use these models can better allocate their feed and economic resources and time.