Lessons from a limpet: modelling decisions of central place foragers

Abstract

Central place foraging behaviour is widespread among both vertebrates and invertebrates. Due to their limited behavioural repertoires, homing limpets (Mollusca Gastropoda) represent an ideal system for the study of optimisation problems in central place foragers. A simple model of limpet behaviour is proposed, based on stochastic dynamic programming, and predictions are compared with published observations. The model was kept as general as possible, so as to be flexible enough to describe the behaviour of almost any central place forager. The state of the forager was characterised by two state variables: the quality of food encountered and the distance from the home site. At each time step, the animal must decide if it is: (a) to accept a patch it has encountered, (b) to continue onwards or (c) to return homewards. Average patch energy content was assumed to increase with distance from home. To simulate different spatial arrangements of resources, the possibility of an autocorrelation in rewards, due to spatial proximity of visited sites, was also allowed. With no autocorrelation, all patches are considered independent of one another, whereas with increasing levels of autocorrelation, one valuable patch is more likely to be followed by another valuable patch. The model allowed reproduction of the typical tri-phasic behaviour observed in many limpet species. This is characterised by a rapid outward phase, a slow central phase, and a rapid return phase. This behaviour changed somewhat with the degree in autocorrelation of rewards due to spatial proximity. When patch contents were highly autocorrelated, the three phases were sharply separated, while with lower levels of autocorrelation, more gradual transitions between the phases occurred. Model results generally agree with published reports of limpet behaviour and they also seem to offer a useful benchmark for understanding more general aspects of central place foraging behaviour in other animals.