Diet-dependent mortality and cognitive impairment reveal species-specific vulnerabilities to a microbial biopesticide in social bees

The increasing use of microbial biopesticides in sustainable agriculture requires a deeper understanding of their potential impact on non-target pollinators. Although biocontrol agents are generally considered safer than synthetic pesticides, they may still cause subtle but ecologically relevant adverse effects on non-target organisms, especially when exposed to multiple stressors that are often overlooked in current risk assessment frameworks. Among these, nutritional stress, caused by habitat loss, fragmentation and reduced floral diversity, is becoming increasingly widespread. In this study, we investigated the lethal and sublethal effects of the bacterial biopesticide Bacillus velezensis (formerly B. amyloliquefaciens) strain QST713 at field-relevant concentrations on two key pollinators: Apis mellifera and Bombus terrestris. For the first time for a biopesticide, oral toxicity was assessed under environmental stress represented by diets with varying sugar concentrations (optimal and suboptimal) to identify potential synergistic effects on bee health. Sublethal effects were examined by studying learning performance and memory retention through a conditioning experiment under laboratory conditions. The results showed marked species-specific differences. While B. velezensis did not impact bee survival under realistic nutritional conditions, we observed a synergistic lethal effect in B. terrestris when biopesticide exposure was coupled with extreme nutritional stress (sugar deprivation). Similar species-specific differences emerged at the behavioral level: unlike A. mellifera, B. terrestris showed impaired visual learning and early long-term memory recall. Taken together, these results show that sublethal cognitive endpoints and multi-stressor contexts may reveal vulnerabilities not immediately evident through mortality-based assessments alone. Our findings also highlight the importance of including multiple pollinator species in risk assessment, as sensitivity to biopesticides might greatly vary among species.