Game of theories: an agent-based model of how variation in contest cost-benefits determines the evolution of contest resolution rules

Game-theory based models are used to understand rules that animals use to settle contests over indivisible resources. However, the empirical literature of contests indicates controversial support to models, with some species supporting different models and other species showing no support to any model. Since strategies used to resolve contests may have different associated costs, it is possible that different conditions have determined the evolution of distinct assessment strategies used by animals. We used an agent-based model to explore the importance of the following conditions: resource availability, probability of reproduction with resource, and damage costs on evolution of assessment strategies. We used self- and mutual-assessment models as a heurist framework to build agents with different assessment strategies. In our model, agents competed for resources in scenarios with different combinations of resource availability, probability of reproduction with resource, and damage costs. We found that agents following the self-assessment with damage strategy were prevalent in scenarios with no probability of reproduction without the resource, independently of other variables. We also found that agents following the non-aggressive strategy occurred in all scenarios. However, agents using the non-aggressive strategy were prevalent only in scenarios with probability of reproduction with the resource. Finally, we observed that agents using mutual-assessment occurred only in a scenario with high risk of damage, low availability of resources, and with probability of reproduction without the resource. These results indicate that agents following the self-assessment with damage and non-aggressive strategies may be able to stay at most scenarios.