1. Studies show that plant defenses influence the host-use of herbivores and tend to be evolutionarily more labile than herbivore traits (e.g., feeding preferences). However, all previous studies have focused exclusively on non-volatile plant defenses thereby overlooking the roles of plant volatiles.

2. We hypothesized that volatiles are equally important determinants of herbivore host-use and are evolutionarily more labile than herbivore traits. To test these hypotheses, the following experiments were conducted.

3. We identified the volatiles and non-volatiles of 17 Asteraceae species and measured their relative contents. We also used a highly resolved bipartite trophic network of the 17 host species and 20 herbivorous (pre-dispersal seed predator) tephritid fly species to determine the evolutionary interactions between plants and herbivores. 

4. The chemical data showed that interspecific similarity in volatiles—but not non-volatiles and phylogenetic distance—significantly accounted for the herbivore community across the plant species; this implies that plant volatiles—but not non-volatile compounds and species identity—dictate plant-tephritid fly interactions. Moreover, we observed phylogenetic signal for non-volatiles but not for volatiles; therefore closely related herbivores do not necessarily use closely related host species with similar non-volatiles, but do tend to attack plants producing similar volatiles. Thus, plant volatiles are evolutionarily more labile than non-volatiles and herbivore traits associate with host use. 

5. These results show that the interactions between plants and herbivores are evolutionary asymmetric, shed light on the role of plant volatiles in plant-herbivore interactions, and highlight the need to include data for both volatiles and non-volatiles when investigating plant-animal interactions.