Food Plant Selection by a Generalist Herbivore: The Moose

A model of food plant selection by a generalist herbivore was developed. The model was designed to predict the species composition of the diet of an herbivore based upon the joint probabilities of whether or not an individual of a plant species satisfied two threshold values: some nutritional minimum and a size limit (both minimum and maximum), and the probability that it was encountered while foraging. The model was tested using moose (Alecs alces) at Isle Royale National Park, Michigan, USA. Initially the threshold values for food selection were determined empirically from the moose's observed behavior, but these empirical values were later shown to be based upon time—energy considerations. Although the model satisfied some of the criteria of optimal foraging contingency models, it appeared that the perfect knowledge assumption was not met. Rather, moose appeared to utilize a strategy of risk aversion.