The coyote model of W. C. Pitt and co-workers is a good example of an IBM that describes a species with a quite complex social behavior but nevertheless is quite simple. The purpose of the model is to support management decisions. Individuals have the state variables sex, age, social status (alpha, beta, pup), and the group (pack) they belong to. The model considers packs but not territories, and is thus not spatially explicit.
The most important rules of the coyote model are for individuals leaving the group and density-dependent mortality and reproduction. Coyotes between 1.5 and 2 years old have a probability of leaving their pack that is proportional to the square of pack size. Coyotes that leave their pack enter a pool of transients. Mortality of transients increases with the total number of transients and thus tends to be higher than that of pack members. The number of offspring produced is assumed to be density-dependent, that is, to decrease with pack size.
The coyote model was implemented using the Swarm software library. This had the advantage that Swarm's framework for implementing an IBM's schedule could also be used for communicating this schedule (Table 1). This makes it easier to understand and reimplement the model. The schedule also shows that the coyote model consists ofactions that are very simple. The complexity of IBMs is more in its implementation and analysis, not necessarily in its formulation.
The model was verified by using five currencies: mean pack size, proportion of transients, average offspring survival rate, average litter size, and proportion of females breeding. Model prediction matched observations surprisingly well, even without fine-tuning of parameters that were taken from the literature. An important insight gained from the coyote model was that the transients are buffering population dynamics. On the one hand they limit, due to their density-dependent mortality, population growth. On the other hand they buffer the loss of alpha individuals.
Was this article helpful?