Modeling the effects of colony size and food heterogeneity on the foraging behavior of ants

Modeling the effects of colony size and food heterogeneity on the foraging behavior of ants

Kenneth Letendre, Departments of Biology and Computer Science, University of New Mexico

Melanie Moses, Department of Computer Science, University of New Mexico

Among species of desert seed-harvester ants, Genus Pogonomyrmex, territory size scales sub-linearly with number of foragers per colony. This sub-linear scaling relationship might result from central place foragers' need to trade off territory quality against increasing travel distance and time from the nest. However, field observations of foraging ants failed to reveal an increasing relationship between colony size and time spent searching, nor total time per foraging trip. We hypothesized that larger colonies support a higher density of foragers on their territories by recruiting foragers to high quality food patches; by virtue of having larger absolute territory sizes, larger colonies have access to more information about the locations of heterogeneously distributed foods, and are able to reduce search times by recruiting foragers to high quality patches. In order to investigate this hypothesis, we developed an agent-based model of ant foraging. We parameterized our model based on observations of ants in the field, and data on the foraging behavior of Pogonomyrmex available from the literature. In order to optimize the remaining parameters, we used our foraging model as the fitness function in a genetic algorithm (GA). We ran this GA for a range of colony sizes foraging on foods varying in the heterogeneity of their distribution. We used two different models of trail laying: a simple recruitment model in which ants lay a pheromone trail on the return trip each time they pick up a piece of food; and a density-dependent recruitment model in which ants can decide to lay a pheromone trail based on the number of other seeds in the neighborhood. We found a main effect of colony size on recruitment behavior; a main effect of food heterogeneity on recruitment behavior; and an interaction of colony size x heterogeneity. We found that in our simple recruitment model, recruitment behavior declined quickly with the addition of randomly distributed foods; whereas recruitment behavior in our density-dependent recruitment model was more robust and persisted until all heterogeneity was removed from the food distribution. We compare the behavior of our models to our observations of ants in the field, and compare to recruitment systems in other ants. We discuss possible applications of our findings to ant colony algorithms.