An analysis of diet selection by large generalist herbivores [Analyse de la sélection du régime alimentaire par les grands généralistes herbivores].

Résumé

Ecological theory has formulated the objective of feeding strategies as to maximize the capture rate of some nutrient. But large generalist herbivores typically face a limit on how much material they can digest. Thus, their feeding strategy is best considered as aiming to chieve the best nutritional balance within a fixed total bulk of food. The problem of finding the diet which provides the best mix nutrients can be formulated as a linear program. A property of this optimization model is that the contribution of a food to the diet will sometimes vary with the food's content of a particular nutrient and sometimes will not. This might explain why no consistent relations of this kind have been found experimentally. The long-delay learning mechanism is described. If this exists in large generalist herbivores, it would provide a means by which they might pursue optimization of nutrient mix in the diet. It would also account for the very large inter- and intraindividual variability in prefernce reported. Optimization achieved by this means would be imperfect. First, the necessary sampling of foods would modify the diet in the direction of greater variety that predicted by linear model. Second, optimization would not act perfectly on all nutrients. If diet selection is dominated by the need to meet nutritional criteria, the response of the diet to availablity of particular foods should not be continous but take the form of a cutoff at very low availability. This implies that as a food becomes rare in a plant community, grazing pressure on it will increase, at least until the cutoff is reached.