Estimation of environmental parameters in foraging theory: data driven sieve method

Abstract

Consider a discrete stochastic environment, X~B(n, p), p ~ beta (α,β) , , to describe the distribution of food. In literature, many ecologists provide different optimal foraging rules under different considerations, for example, Green's dynamic rule (1980), GUT (giving-up time) strategy, and MVT (marginal value theorem). In this talk, we consider the parameter estimation problem from observed data by means of dynamic programming as a pseudo tool whatever foraging rule adopted by the predator. It is a realistic problem, in fact, we can not know exactly what foraging rule adopted by the predator. Here, we not only have elegant estimators but also provide a visible skill to identified predator's foraging rule. Furthermore, from the statistical viewpoint, we also try to examine where data information comes from.