Simulated evolution of food foraging strategies of army ants

Abstract: Many ant species make use of pheromone trails to coordinate food raids, which result in emergent behaviour in the form of complex, dynamic foraging patterns. The foraging behaviour of army ants, in particular, can be modelled as a ’central foraging problem’, where ants leave the nest to collect and bring back food. This thesis implements a previously defined model for foraging army ants to investigate what ant behavioural parameters lead to optimal solutions for three different types of food sources. These food sources are: small and common, large and scarce, and a combination of both of these food sizes and distributions. The model is also extended to investigate similar scenarios where a foraging ant colony is competing with another colony for resources. To find optimal behavioural parameters, the parameter space is searched using a simple evolutionary algorithm. This is used to successively ’evolve’ parameters to find optimal solutions. The results show that the optimal parameters, as well as foraging patterns that emerge, changed when a competing colony was present. The degree to which a solution found can be deemed ’optimal’ was highly dependent on the specific scenario. Finally, the evolution of foraging strategies for small and common was more successful than that of large and scarce. This research has applications in both natural science and computer science, where applications in the former involve, for example, swarm intelligence and optimisation algorithms.