Relationships of host plant phylogeny, plant chemistry and host plant specificity of a prospective biological control agent of yellow starthistle

Plant species used for host specificity testing are usually chosen based on the assumption that the risk of attack by a prospective biological control agent decreases with increasing phylogenetic distance from the target weed. Molecular genetics methods have greatly improved our ability to measure genetic relatedness among test plants; however, biological control agents respond to the chemical, physical and behavioral characteristics of plants, rather than directly to their genetics. In general, phylogeny serves as a useful surrogate for the chemical, physical and behavioral (e.g., phenology, habitat) characteristics; however, convergent evolution could produce similar characters in distantly related plants, and divergent evolution could produce differences among closely related plants. Thus, relying only on phylogeny risks omitting a more distantly related plant that is vulnerable to attack. We analyzed the phylogenetic relationships of test plants in the tribe Cardueae which have been used to evaluate the specificity of Ceratapion basicorne, a prospective biological control agent of yellow starthistle, Centaurea solstitialis. The host plant specificity of the weevil mirrored phylogeny at a gross scale (tribe, subtribe), but not at a fine scale (subgenus). We measured the relative abundance of volatile organic compounds (VOC) of the test plants to try to determine if this helps explain host plant preference.