Complex interactions among biocontrol agents, pollinators, and an invasive weed: a structural equation modeling approach.

Herbivores, seed predators, and pollinators can exert strong impacts on their host plants. They can also affect the strength of each other's impact by modifying traits in their shared host, producing super- or sub-additive outcomes. This phenomenon is especially relevant to biological control of invasive plants because most invaders are attacked by multiple agents. Unfortunately, complex interactions among agents are rarely studied. We used structural equation modeling (SEM) to quantify the effect of two biocontrol agents and generalist pollinators on the invasive weed Centaurea solstitialis, and to identify and quantify the direct and indirect interaction pathways among them. The weevil Eustenopus villosus is both a bud herbivore and a predispersal seed predator; the fly Chaetorellia succinea is also a predispersal seed predator; Apis mellifera is the primary pollinator. We conducted this work at three sites spanning the longitudinal range of C. solstitialis in California (USA) from the coast to the Sierra Nevada Mountains. SEM revealed that bud herbivory had the largest total effect on the weed's fecundity. The direct effect of bud herbivory on final seed set was 2-4 times larger in magnitude than the direct effect of seed predation by both agents combined. SEM also revealed important indirect interactions; by reducing the number of inflorescences plants produced, bud herbivory indirectly reduced the plant's attractiveness to ovipositing seed predators. This indirect, positive pathway reduced bud herbivory's direct negative effect by 11-25%. In the same way, bud herbivory also reduced pollinator visitation, although the magnitude of this pathway was relatively small. E. villosus oviposition deterred C. succinea oviposition, which is unfortunate because C. succinea is the more voracious of the seed predators. Finally, C. succinea oviposition indirectly deterred pollinator visitation, thereby enhancing its net effect on the plant. This study demonstrates the powerful insights that can be gained from the SEM approach in understanding the multiple direct and indirect interactions among agents and pollinators and their effects on an invasive weed. Such an approach may improve our ability to manage weeds with biocontrol agents by identifying pathways that could be exploited by future agents and minimizing the possibility of interference with established agents.