Evolution of explorative and exploitative search strategies in collective foraging

Abstract

Evolutionary theories of foraging hypothesize that individual foraging strategies evolved to maximize search efficiency. Many studies have investigated the trade-offs between exploration and exploitation and how individual foragers manage them. However, for groups of foragers, these trade-offs can change and individual search strategies may evolve in response to their social environments. For instance, foragers may use social information to collectively find and harvest resources, which might increase competition and decrease the benefits of exploring new resources. Previous work has shown that when learning socially, it is optimal for groups to be composed of highly explorative strategies. However, individual and collective search efficiencies may not align if individual search strategies beneficial for the group are disadvantageous for the individuals. In the present study, we use an agent-based model to investigate the effect of collective foraging on the evolution of individual search strategies and how they relate to group search efficiency. We use a genetic algorithm to evolve L\'evy walk exponents ($\mu$) that govern the balance of explorative versus exploitative foraging. We show that groups can evolve with social learning whose collective performance is more optimal than without social learning. The model shows that exploiters have a selective advantage in scrounging off findings by other agents, but too many exploiters diminished group search efficiencies. We also show that social learning in large groups can increase the payoffs of exploration and lead to the selection of more exploratory groups. Finally, we show that area-restricted search can help explorers exploit found resources and lead to more efficient collective search. Our results demonstrate how exploration and exploitation must be balanced at both individual and collective levels, and how individual search strategies can evolve to the benefit of collective search efficiency.