Predicting the host range of a biological control agent prior to release is one of the most important steps in the development of new agents. Knowing which species are most at risk of this non-target damage improves the predictability of these tests. To predict safety, the potential agent is exposed to a subset of the entire flora that represents valued native, agricultural and ornamental plant species. The list of plants includes those species that are the closest relatives to the target weed. To identify these species, molecular phylogenies can be useful tools that potentially identify the most vulnerable plant species. While conducting biological control research of the invasive weed Brazilian peppertree, Schinus terebinthifolia, we conducted nuclear ITS1 and chloroplast trnL-F analysis of agricultural, commercial and native plants that are related to the weed. The results of this analysis support recent phylogenetic studies that have established two subfamily clades of the Anacardiaceae, the Anacardioideae and the Spondioideae. Moreover, our results indicate that species of concern from the genera Cotinus, Lithrea, Pistacia, Rhus, Toxicodendron, and Schinus, group together in the Anacardioideae, whereas the Spondias species, group in the Spondioideae subfamily clade. Further, the closest relatives to the target weed, those species most at risk of non-target damage by biological control agents include members of the Schinus and Lithrea genera. A review of the host testing results of 17 potential biological control candidates indicated that three of these species show a significant phylogenetic signal in their host range. These three species, the sawfly, Heteroperreyia hubrichi, the foliar gall inducing psyllid Calophya latiforceps, and the thrips Pseudophilothrips ichini have been recommended for release by
