Résumé
Most—if not all—pollinators make foraging decisions based on learning and memory. In interaction with environmental conditions and competitive pressure, pollinators’ cognition shapes their movement patterns, which in turn determine pollen transfers. However, models of animal-mediated pollination often make simplifying assumptions about pollinator movements, notably by not incorporating learning and memory. Better considering cognition as a driver of pollinators’ movements may thus provide a powerful mechanistic understanding of pollen dispersal. In this exploratory study, we connect pollinator behavior and plant reproduction by using an agent-based model of bee movements implementing reinforcement learning. Simulations of two bees foraging together in environments containing twenty plants show how learning can improve foraging efficiency as well as plant pollination quality through larger mating distances and smaller self-pollination rates while creating spatially heterogeneous pollen flows. This suggests that pollinators’ informed foraging decisions contribute to genetic differentiation between plant subpopulations. We believe this theoretical exploration will pave the way for a more systematic analysis of animal-mediated plant mating patterns, as model predictions can be tested experimentally in real bee-plant systems.