Interactions between plants and their insect herbivores are often specialized, making herbivore-host coevolution possible. However, herbivore-host associations are also labile on ecological and evolutionary timescales: new insect-host associations arise due to range expansions, invasions, and host shifts. How should coevolution proceed in novel associations? Parasite-host coevolution theory predicts that for most herbivores selection should favor reduced insect “virulence” (damage done by attack). For most plants, selection should favour increased tolerance. Combining these patterns yields the prediction that impact should be high in novel plant-insect associations, then decline over time as the new interactors coevolve. We tested this prediction for three gallmaking herbivores of goldenrods (Solidago altissima and S. gigantea) that have evolved specialist host races or cryptic species. In each case, the novel host form attacks S. gigantea. We collected galled goldenrods and their nearest ungalled neighbors at four sites in New Brunswick, Ontario, and Iowa, and compared aboveground biomass to estimate herbivore impact. We also imposed clipping and defoliation damage (which are not “novel” on either host) as a control.
Results/Conclusions
Clipping and defoliation had equal impact on the two hosts, but herbivore impact generally differed between hosts. We had 10 datasets, covering three gallmakers at four sites in three years. Eight showed the expected pattern (more herbivore impact on S. gigantea, the novel host; six of these individually significant). One showed the reverse pattern, and in one case impact was nearly equal between hosts. Combining all datasets, there was a significant pattern of higher impact on S. gigantea (sign test, P = 0.01), which we interpret as evidence of herbivore-host coevolution on the ancestral host, S. altissima. Tests of the evolution-of-impact hypothesis appear to be rare, despite the increasing importance of novel interactions in an age of invasions and climate-driven range expansions.