BackgroundFloral phytochemicals are ubiquitous in nature, and can function both as antimicrobials and as insecticides. Although many phytochemicals act as toxins and deterrents to consumers, the same chemicals may counteract disease and be preferred by infected individuals. The roles of nectar and pollen phytochemicals in pollinator ecology and conservation are complex, with evidence for both toxicity and medicinal effects against parasites. However, it remains unclear how consistent the effects of phytochemicals are across different parasite lineages and environmental conditions, and whether pollinators actively self-medicate with these compounds when infected.ApproachHere, we test effects of the nectar alkaloid anabasine, found in Nicotiana, on infection intensity, dietary preference, and survival and performance of bumble bees (Bombus impatiens). We examined variation in the effects of anabasine on infection with different lineages of the intestinal parasite Crithidia under pollen-fed and pollen-starved conditions.ResultsWe found that anabasine did not reduce infection intensity in individual bees infected with any of four Crithidia lineages that were tested in parallel, nor did anabasine reduce infection intensity in microcolonies of queenless workers. In addition, neither anabasine nor its isomer, nicotine, was preferred by infected bees in choice experiments, and infected bees consumed less anabasine than did uninfected bees under no-choice conditions. Furthermore, anabasine exacerbated the negative effects of infection on bee survival and microcolony performance. Anabasine reduced infection in only one experiment, in which bees were deprived of pollen and post-pupal contact with nestmates. In this experiment, anabasine had antiparasitic effects in bees from only two of four colonies, and infected bees exhibited reduced—rather than increased—phytochemical consumption relative to uninfected bees.ConclusionsVariation in the effect of anabasine on infection suggests potential modulation of tritrophic interactions by both host genotype and environmental variables. Overall, our results demonstrate that Bombus impatiens prefer diets without nicotine and anabasine, and suggest that the medicinal effects and toxicity of anabasine may be context dependent. Future research should identify the specific environmental and genotypic factors that determine whether nectar phytochemicals have medicinal or deleterious effects on pollinators.