Systematic cultivar selection for weed biological control risk assessment.
Classical biological control is important for long-term, sustainable management of invasive species such as weeds. To be acceptable for introduction, new biocontrol agents must not damage crops, native plants, or other valued non-target species. Host-specificity experiments inform risk assessment of new biocontrol agents by prioritising and testing non-target plant species. Susceptibility to damage can also vary between cultivars of the same species, yet current protocols for the selection of test plant species inadequately address cultivar differences. Poor selection increases the risk of failing to detect non-target damage. We developed criteria for prioritising cultivars and a process chart (the 'decision support tool') to guide the selection of cultivars for biological control host-specificity testing. We reviewed and documented current cultivar selection practice published in prominent biological control journals and government documents to further explore current best practice for cultivar selection. To demonstrate the utility of the tool, we then applied the decision support tool to a complex host-specificity testing case study. In our review, most papers either did not mention cultivars of the cultivated plant species being tested, or they provided incomplete descriptions of cultivars without explaining omissions. Only one of 29 papers fully described the method for selecting and prioritising cultivars and reported the results for each cultivar tested. Our application of the process chart to a potato testing case study generated a feasible short-list of cultivars which can be scrutinized and updated. We showed how selections could be made through a collaborative and transparent process involving key stakeholders and decision makers. We have identified an example of under-reporting in classical biological control which, if not addressed, could impede progress in the sustainable management of invasive species. The decision support tool we developed has broad application in weed biological control risk assessment. We demonstrated that the decision tool is easy to use, can account for uncertainty, is adaptable to different species, and is suitable for both small and large cultivar groups irrespective of complexity. Our approach will result in more transparent, defensible, and reproducible cultivar selection practices, leading to greater confidence in biological control risk assessments.