Predicting non-target impacts.

Biocontrol of invasive alien weeds has produced great benefits, but concerns over undesirable impacts on non-target plants and/or indirect interactions between biocontrol agents and other biota impede the implementation of biocontrol in some countries. Although great strides have been made, continuing uncertainties predicting the realized host range of candidate agents is probably resulting in some being erroneously rejected due to overestimation of risk. Further refinement of host-range testing protocols is therefore desirable. Indirect interactions are inherently harder to predict, and the risk of both direct and indirect non-target impacts may change over time due to biocontrol agents evolving or expanding their range under climate change. Future research directions to better understand the risk of non-target impacts over time are discussed.