The geographic mosaic theory of coevolution states that variation in species interactions forms the raw material for coevolutionary processes, which take place over large geographic scales. One key assumption underlying the process of coevolution in plant-herbivore interactions is that herbivores exert selection on their host plants and that this selection varies among plant populations. We examined spatial variation in the existence and strength of phenotypic selection on host plant resistance exerted by specialist herbivores in 17 archipelago populations of the perennial herb Vincetoxicum hirundinaria (Asclepiadaceae). In these highly fragmented populations, V. hirundinaria is consumed by the larvae of two specialist herbivores: the folivorous moth Abrostola asclepiadis and the seed predator Euphranta connexa. Selection imposed on host plants by these herbivores was examined by analyzing the associations between levels of herbivory, plant fitness, and contents of a number of leaf chemicals reflecting plant resistance to and quality for the herbivores. We found extensive spatial variation in the levels of herbivory and in plant fitness. More importantly, the impact of both leaf herbivory and seed predation on plant fitness varied among plant populations, indicating spatial variation in phenotypic selection. In addition, leaf chemistry varied widely among plant populations, reflecting spatial variation in plant quality as food for the herbivores. However, leaf compounds influenced folivory similarly in all the studied plant populations, and interestingly, some of the compounds were associated with the intensity of seed predation. Finally, some of the leaf compounds were associated with plant fitness, and the strength and direction of these associations varied among plant populations. The observed spatial variation in the strength of the interactions between V. hirundinaria and its specialist herbivores suggests a geographic selection mosaic. Because the occurrence and strength of spatial variation varied between the two specialist herbivores, our results highlight the importance of considering multiple enemies when trying to understand evolution of interactions between plants and their herbivores.

Information of the patterns of genetic variation in plant resistance and tolerance against herbivores and genetic trade‐offs between these two defence strategies is central for our understanding of the evolution of plant defence. We found genetic variation in resistance to two specialist herbivores and in tolerance to artificial damage but not to a specialist leaf herbivore in a long‐lived perennial herb. Seedlings tended to have genetic variation in tolerance to artificial damage. Genetic variation in tolerance of adult plants to artificial damage was not consistent in time. Our results suggest that the level of genetic variation in tolerance and resistance depends on plant life‐history stage, type of damage and timing of estimating the tolerance relative to the occurrence of the damage, which might reflect the pattern of selection imposed by herbivory. Furthermore, we found no trade‐offs between resistance and tolerance, which suggests that the two defence strategies can evolve independently.

No abstract

Summary1. Due to geographically variable species interactions, plants may become locally adapted to their sympatric herbivores and pollinators. However, adaptation to the abiotic environment may significantly affect plant interactions with herbivores and pollinators. Local adaptation to the abiotic environment may constrain local adaptation to herbivores and pollinators under contrasting selection pressures, resulting in trade-offs in local adaptation. 2. We studied local adaptation of a perennial herb, Vincetoxicum hirundinaria, in a reciprocal transplant experiment among four populations and measured plant fitness, pollination success and resistance to two specialist herbivores. We also estimated local adaptation of these two herbivores and generalist pollinators to their sympatric plant populations. Local adaptation was compared with within-population genetic variation, genetic and geographical divergence, and with divergence in terms of population size, environmental conditions and plant secondary chemistry. We further compared local adaptation to the environment, to local adaptation to herbivores and pollinators to detect possible trade-offs in local adaptation. 3. The existence and degree of local adaptation varied among the plant populations. Plants from two populations were locally adapted to their sympatric leaf herbivores and plants from two populations were locally adapted to their abiotic environment. Herbivores from one population were locally adapted to their sympatric plant population. Local adaptation of V. hirundinaria to the seed predator increased with increasing among-population divergence in precipitation and temperature. Local adaptation to the seed predator and the environment increased with increasing population genetic variation. Local adaptation of V. hirundinaria in fitness and in herbivore resistance also correlated positively, suggesting lack of trade-offs in local adaptation. 4. Synthesis. These results demonstrate that species interactions can lead to a mosaic of locally adapted plant, herbivore and pollinator populations. In addition to natural enemies, genetic variation, the abiotic environment and mutualistic interactions contribute to the evolution of local adaptation in long-lived plants. These results provide new insights into the patterns and causes of variation in local adaptation and are among the first to demonstrate that conflicting selection pressures within a population do not constrain local adaptation in multiple traits.