When resources don't rescue: flowering phenology and species interactions affect compensation to herbivory in Ipomopsis aggregata

The ability of plants to tolerate, or compensate for, herbivore damage is highly variable and has been the subject of much research. Although many plants can compensate for herbivore damage, and some even overcompensate, we cannot yet generalize about the conditions that promote a positive response to damage. Here, we asked how abiotic resources (i.e. plant nutrient status) coupled with biotic interactions – i.e. subsequent interactions with pollinators, seed predators and nectar robbing bumble bees – affect the compensatory ability of Ipomopsis aggregata, a monocarpic herb that has been the subject of much previous debate. We hypothesized that compensation to herbivore damage in I. aggregata (Polemoniaceae) would depend first on plants having an ample supply of resources and, second, on the outcome of subsequent interactions with mutualist pollinators and enemy pre‐dispersal seed predators and nectar robbing bumble bees. We used a fully‐factorial experiment in which plants were watered, fertilized or left as unmanipulated controls, crossed with clipping to simulate herbivore damage to the apical meristem. Resource addition enhanced both male and female components of fitness, but resource enhancement did not provide the means for plants to fully compensate for simulated herbivory. Clipped plants produced significantly more inflorescences, but at the expense of a delay in flowering and fewer total flowers. Clipping significantly reduced losses to dipteran pre‐dispersal seed predators by delaying flowering time, but early flowering plants produced higher numbers of seeds despite incurring higher rates of predation. Clipped plants incurred a higher risk to nectar robbers in one of two years. Overall, clipped plants suffered severe reductions (a nearly 50% reduction in total seed set) in female success, but clipping combined with nutrient addition enhanced male function through increases in per‐flower pollen production. However, because clipped plants produced significantly fewer flowers than unclipped plants, whole‐plant pollen production was significantly reduced by clipping. Pollinator visitation and nectar robbing were variable between clipping treatments and between years and (nectar robbing) among sites. Our results demonstrate that the variability in plant response to herbivory can, at least in part, be driven by plant interactions with mutualists and enemies. Thus, accounting for such interactions and their variability is important to fully understanding plant compensation for herbivore damage and will likely go far to explain variation in plant response that appears to be independent of resources.