Flowers are hubs of attraction, but it is unclear how positive and negative species interactions shape floral evolution, especially for chemical traits. The evolutionary outcomes of multi-species interactions in different abiotic environments are often profoundly different from simplified pair-wise interactions. While selection on floral traits by pollinators and herbivores has been documented in many systems, few studies examine the impact of multispecies interactions, which generate both positive and negative outcomes, at the population level using field and experimental approaches and fewer studies have investigated floral scent as the nexus of these interactions. Our approach expands upon correlative studies by combining multi-year, field-collected observational data and experimental manipulation of floral scent. We test the hypothesis that floral antagonists exert stronger selective pressure on floral scent chemistry than do pollinators in the Arkansas River Valley evening primrose, Oenothera harringtonii, which is polymorphic for the floral scent compound, (R)(-)linalool across its range. Both nocturnal pollinators (Hyles lineata) and herbivores (Mompha definitella, seed predator) perceive (R)(-)linalool at relevant dosages.
Results/Conclusions
Our data show that spatial and temporal variation in the dual interactions of pollinators and herbivores play an integral role in the maintenance of the linalool polymorphism in the O. harringtonii system. Flowers experimentally augmented with linalool showed decreased fitness associated with higher seed herbivory despite having increased pollen deposition. However, this pattern is not maintained across the distribution of the species, as revealed in a transplant experiment where linalool production did not influence seed predation. These independent lines of evidence indicate that herbivore-mediated selection, rather than pollination, drives local adaptation through fragrance diversification and is context-dependent. Our study challenges the primacy of pollinators as selective agents on floral traits and shows that herbivores can select against signals important to pollinators. Our findings emphasize the need to rethink the paradigm that pollinators are chiefly responsible for floral trait evolution and thus, pollinator-mediated diversification of angiosperms more broadly.