A Markov chain analysis of predator strategy in a model-mimic system

A predator which is preying on a model-mimic system can choose either the single-trial strategy or a multi-trial strategy as its behavior in learning to prudently harvest such a prey system. In this learning behavior, an important and often-posed problem is to determine which among these two strategies is better suited for the predator and why one is preferable over the other. We present in this article, using Markov chain methods, an extensive analysis of these strategies (and also of eat-everything, strategy). We conclude that the multi-trial strategy is the one that the predator should adopt (but we will also describe the situations when the single-trial strategy seems to be better). Our conclusions are based on the comparisons of quantities such as the mean benefit to the predator, energy derived by a predator from the model-mimic system and (a newly introduced notion of) contagion in eating mimics and models (these quantities are computed for different strategies). The first two quantities are functions of the abundancep and noxiousnessb of models. The contagion is a function of onlyp; and, though independent ofb, it is also in support of multi-trial, strategy. We introduce, in the present context, a biological analog of the d'Alembert principle and also derive functions describing the influences of eating a specified type of prey at a given time on eating any type of prey at a later time. Various results of Estabrook-Jespersen (single-trial strategy) and Bobisud-Potratz (multi-trial strategy) are re-derived as special cases of our more general results. A central limit theorem under the eat-everything strategy is given.