The preference–performance hypothesis predicts that female insects maximize their fitness by utilizing host plants which are associated with high larval performance. Still, studies with several insect species have failed to find a positive correlation between oviposition preference and larval performance. In the present study, we experimentally investigated the relationship between oviposition preferences and larval performance in the butterfly Anthocharis cardamines. Preferences were assessed using both cage experiments and field data on the proportion of host plant individuals utilized in natural populations. Larval performance was experimentally investigated using larvae descending from 419 oviposition events by 21 females on plants from 51 populations of two ploidy types of the perennial herb Cardamine pratensis. Neither ploidy type nor population identity influenced egg survival or larval development, but increased plant inflorescence size resulted in a larger final larval size. There was no correlation between female oviposition preference and egg survival or larval development under controlled conditions. Moreover, variation in larval performance among populations under controlled conditions was not correlated with the proportion of host plants utilized in the field. Lastly, first instar larvae added to plants rejected for oviposition by butterfly females during the preference experiment performed equally well as larvae growing on plants chosen for oviposition. The lack of a correlation between larval performance and oviposition preference for A. cardamines under both experimental and natural settings suggests that female host choice does not maximize the fitness of the individual offspring.

Spatial variation in biotic interactions and natural selection are fundamental parts of natural systems, and can be driven by differences in both trait distributions and the local environmental context of the interaction. Most studies of plant-animal interactions have been performed only in natural settings, making it difficult to disentangle the effects of traits and context. To assess the relative importance of trait differences and environmental context for among-population variation in plant resistance to herbivory, we compared oviposition by the butterfly Anthocharis cardamines on two ploidy types of the herb Cardamine pratensis under experimentally controlled conditions with oviposition in natural populations. Under controlled conditions, plants from octoploid populations were significantly more preferred than plants from tetraploid populations. This difference was largely mediated by differences in flower size. Among natural populations, there was no difference in oviposition rates between the two ploidy types. Our results suggest that differences in oviposition rates among populations of the two cytotypes in the field are caused mainly by differences in environmental context, and that the higher attractiveness of octoploids to herbivores observed under common environmental conditions is balanced by the fact that they occur in habitats which harbor lower densities of butterflies. This illustrates that spatial variation in biotic interactions is the net result of differences in trait distributions of the interacting organisms and differences in environmental context, and that variation in both traits and context are important in understanding species interactions.

Plants have two principal defense mechanisms to decrease fitness losses to herbivory: tolerance, the ability to compensate fitness after damage, and resistance, the ability to avoid damage. Variation in intensity of herbivory among populations should result in variation in plant defense levels if tolerance and resistance are associated with costs. Yet little is known about how levels of tolerance are related to resistance and attack intensity in the field, and about the costs of tolerance. In this study, we used information about tolerance and resistance against larval herbivory by the butterfly Anthocharis cardamines under controlled conditions together with information about damage in the field for a large set of populations of the perennial plant Cardamine pratensis. Plant tolerance was estimated in a common garden experiment where plants were subjected to a combination of larval herbivory and clipping. We found no evidence of that the proportion of damage that was caused by larval feeding vs. clipping influenced plant responses. Damage treatments had a negative effect on the three measured fitness components and also resulted in an earlier flowering in the year after the attack. Tolerance was related to attack intensity in the population of origin, i.e. plants from populations with higher attack intensity were more likely to flower in the year following damage. However, we found no evidence of a relationship between tolerance and resistance. These results indicate that herbivory drives the evolution for increased tolerance, and that changes in tolerance are not linked to changes in resistance. We suggest that the simultaneous study of tolerance, attack intensity in the field and resistance constitutes a powerful tool to understand how plant strategies to avoid negative effects of herbivore damage evolve.

Timing of plant development both determines the abiotic conditions that the plant experiences and strongly influences the intensity of interactions with other organisms. Plants and herbivores differ in their response to environmental cues, and spatial and temporal variation in environmental conditions might influence the synchrony between host plants and herbivores, and the intensity of their interactions. We investigated whether differences in first day of flowering among and within 21 populations of the polyploid herb Cardamine pratensis influenced the frequency of oviposition by the butterfly Anthocharis cardamines during four study years. The proportion of plants that became oviposited upon differed among populations, but these differences were not related to mean flowering phenology within the population in any of the four study years. Attack rates in the field were also not correlated with resistance to oviposition estimated under controlled conditions. Within populations, the frequency of butterfly attack was higher in early-flowering individuals in two of the four study years, while there was no significant relationship in the other 2 years. Larger plants were more likely to become oviposited upon in all 4 years. The effects of first flowering day and size on the frequency of butterfly attack did not differ among populations. The results suggest that differences in attack intensities among populations are driven mainly by differences in the environmental context of populations while mean differences in plant traits play a minor role. The fact that within populations timing of flowering influenced the frequency of herbivore attack only in some years and suggests that herbivore-mediated selection on plant phenology differs among years, possibly because plants and herbivores respond differently to environmental cues.