A comparison of computer-generated and naturally occurring foraging patterns in route-network-constrained spider monkeys.

The attribution of goal-directed behavior to observations of primate foraging and ranging requires that simpler explanations for observed behavior patterns be eliminated. Computer-generated simulations of non-goal-directed foraging behavior can be used as null models for higher complexity cognitive foraging, and can provide quantifiable data against which to compare the observed behavioral patterns in wild primates. In this paper, we compare the results of two variations of computer simulated null models with observed foraging behavior of wild spider monkeys (Ateles belzebuth). One model simulates monkeys searching using a modified random-walk model in which monkeys alternate 100-m steps with turn angles derived from observed behavior. The second model constrains travel to an observed route system derived from observations of wild spider monkeys. Simulated monkeys in each model searched among increasing densities of feeding trees ranging from 10 to 1,000. We compared travel distance, travel directness, and accuracy of starting direction for each feeding tree discovered for the two models. We then compared these results with those derived from observations of wild spider monkeys. Route-model monkeys traveled shorter distances and more directly to feeding trees than did randomly foraging monkeys, and discovered trees in the direction they started more often. Observed spider monkeys outperformed simulated monkeys from both models in all variables, allowing us to reject the null hypothesis that observed foraging and ranging behavior could be explained by non-goal-directed travel.