Foraging in a patchy environment: prey-encounter rate and residence time distributions

Small bluegill sunfish, Lepomis macrochirus, foraging among patches in the laboratory did not search systematically within a patch; their intercapture intervals did not differ from a model of random prey encounter within a patch. Patch-residence time, number of prey eaten, and giving-up time (time between last prey capture and leaving the patch) were measured for bluegills foraging in two different three-patch ‘environments’ (a constant environment, in which each patch began with the same number of prey and a variable environment, in which two patches began with low prey density and one patch with high prey density). When compared with three decision rules a forager may use to determine when to leave a patch, the data most closely agreed with predictions from a ‘constant residence time’ rule. Bluegills responded to changes in the distribution of prey among patches, but not by using different decision rules. There was qualitative, but not quantitative, agreement with a model of random residence times. The total number of prey eaten by a bluegill during a foraging bout was similar to the number predicted from a model of random search and random residence times. In both theoretical and empirical work on foraging, investigators have considered three types of information that foragers may use to decide when to leave a patch of prey: (1) the time elapsed since entering the patch, (2) the accumulated reward since entering the patch, and (3) the present rate of energy gain in the patch. From such information, simple decision rules have been proposed that would allow a forager to approximate optimal patch use (Charnov 1976; Krebs 1978; Howell & Hart1 1980). In this study, we considered three of the decision rules discussed in Krebs (1973) and Krebs et al. (1974): (I) a constant residence-time rule, where a forager stays in each patch for a constant optimal amount of time; (2) a constant number rule, where a forager leaves a patch after capturing a certain number of prey; and (3) a constant giving-up time rule, where a forager remains in a patch until the time since the last prey capture reaches some threshold level. This threshold time, or giving-up time, should be inversely related to the average capture rate for the environment as a whole. Iwasa et al. (1981) showed that, in a stochastic system, environments differing in distribution of prey would differ in which of the three strategies would provide results closest to optimal. Their results suggest that if prey is distributed heterogeneously among patches, for example if their distribu* Present address: Department of Zoology, Box 7617, North Carolina State University, Raleigh, North Carolina 27695-7617, U.S.A. tion is a negative binomial, then the best constant giving-up time rule would yield higher capture rates than the best constant residence-time or constant number rule; however, if prey were more homogeneously distributed, the best constant residencetime or number rule would yield the highest rates. Under these conditions, a constant giving-up time could yield the lowest capture rates. We designed experiments to evaluate these three hypotheses of proximal decision-making in foragers and tested the results against a fourth hypothesis, a model of random departure times from patches that precludes the use of any of these decision rules. This fourth possibility means that the animal leaves a patch independent of its experience in the patch. The decision to leave is completely random. In addition, we tested whether foragers adjust to their environments by using different decision rules in environments differing in prey distribution. In a laboratory study, we used small bluegills, Lepomis macrochirus, foraging for midge fly, Chironomus riparius, larvae distributed among patches of artificial vegetation. Under natural conditions, small bluegills forage within vegetated littoral habitats, and midge fly larvae are a major component of their diets (Werner 1967; Keast 1978; Mittelbach 1981). We first conditioned bluegills to an environment in which prey were equally distributed among patches. Then, by varying the distribution, we were able to distinguish among the four hypothesized decision rules. We could also deter0003-3472/89/030444+11$03.00/0 01989 The Association for the Study of Animal Behaviour