We are currently examining the individual behavioral rules through which foraging swarms of the New World army ant Eciton burchellii are organized. We have measured the body positions of more than 30,000 ants in digital video of swarms from Costa Rica. From these data we generated a map approximating the changing pheromone landscape experienced by the ants over time. This landscape was matched with ant path data enabling us to quantitatively measure how E. burchellii army ants modify their turn angle and velocity in response to pheromone. Our measurements also provided evidence that ants of different lengths perform different functions in the swarm. These results were integrated into an individual-based computer model of army ant swarming that was tested against the video data. 

Eciton burchellii swarms exhibit self-organization in that they collectively perform very computationally complex tasks without centralized control and with access to only local information. To swarm successfully E. burchellii must spatially coordinate as many as 200,000 nearly blind individuals of multiple castes in the face of changing biotic and abiotic factors.  Large swarms can raid over an area larger than 2,000 m² in a day, processing and retrieving more than 30,000 prey items (Franks, N. R. 1985. Experimental behavioral ecology and sociobiology, 31:91-107). Swarms are robust to drastic changes in size and to huge variations in substrate type. Individuals have no means to measure overall spatial distribution or foraging success within different parts of the swarm. Self-organization provides evolution a mechanism by which selection on such simple individual behaviors can generate complex system-level patterns. Our data and model results suggest that simple rules governing how velocity and turn angle change in response to variations in pheromone concentration and local ant density are enough to qualitatively reproduce many of the complex swarming behaviors observed in the field.