Abstract
Honeybees (Apis mellifera) are important pollinators. Their foraging behaviors are essential to colony sustainability. Sublethal exposure to pesticides such as neonicotinoids can significantly disrupt these behaviors, in particular pollen foraging. We investigated the effects of sublethal doses of the neonicotinoid imidacloprid on honeybee foraging, at both individual and colony levels, by integrating field experiments with artificial intelligence (AI)-based monitoring technology and mechanistic simulations using the BEEHAVE model. Our results replicated previous findings, which showed that imidacloprid selectively reduces pollen foraging at the colony level, with minimal impact on nectar foraging. Individually marked exposed honeybees exhibited prolonged pollen foraging trips, reduced pollen foraging frequency, and instances of drifting pollen foraging trips, likely due to impaired cognitive functions and altered metabolism. These behavioral changes at the individual level corroborated the previous model predictions derived from BEEHAVE, which highlights the value of combining experimental and simulation approaches to disentangle underlying mechanisms through which sublethal effects on individual foragers scale up to impact colony dynamics. Our findings have implications for future pesticide risk assessment, as we provide a robust feeding study design for evaluating pesticide effects on honeybee colonies and foraging in real landscapes, which could improve the realism of higher-tier ecological risk assessment.