Australian Journal of Entomology (2006) 45, 292–297 © 2006 CSIRO Journal compilation © 2006 Australian Entomological Society doi:10.1111/j.1440-6055.2006.00553.x Blackwell Publishing AsiaMelbourne, AustraliaAENAustralian Journal of Entomology1326-6756© 2006 The Authors; Journal compilation © 2006 Australian Entomological Society? 2006454292297Original ArticleRisk analysis and native-range research A Sheppard et al. *andy.sheppard@csiro.au Case study Native-range research assists risk analysis for non-targets in weed biological control: the cautionary tale of the broom seed beetle Andy Sheppard, 1 * Melanie Haines 2 and Thierry Thomann 1 1 CSIRO European Laboratory, Campus de Baillarguet, 34980 Montferrier-sur-Lez, France. 2 Bio-Protection and Ecology Division, PO Box 84, Lincoln University, Canterbury, New Zealand. Abstract This case study considers the broom seed beetle, Bruchidius villosus, a narrowly oligophagous species within the Fabaceae, subtribe Genistinae for which in-depth native-range studies have been vital to help understand the likely field host specificity following release. Bruchidius villosus has been used in three countries as a classical biological control agent against Scotch broom, Cytisus scoparius. Original host-specificity testing of a UK population, where this species had only been observed developing on C. scoparius, suggested this population was specific to the target. The beetle was released in New Zealand. Following release, however, the agent exhibited a broader host range than in the tests, but not a broader host range than that of the species as a whole. Subsequent studies in the native range using surveys and field testing have helped to show why B. villosus populations exhibit higher specificity in the native range than would be expected from the species’ host range. This case is used to illustrate the contribution native-range studies can make to science-based risk analysis of biological control agents against weeds. By doing so, they also highlight the associated risks of ignoring native-range studies and adopting of a ‘grab-and-run’ approach to obtaining classical bio- logical control agents. Key words Bruchidius villosus, Cytisus scoparius, field tests, host range, host specificity. INTRODUCTION Evaluating the risk of non-target use by potential weed bio- logical control has been and is a relatively straightforward procedure for the majority of species considered. This is because the majority of potential agents selected for testing are either quickly rejected, or clearly demonstrate enough specificity in even the most conservative quarantine test types (no-choice or starvation tests). Problems arise when: (i) the field host range of the agents released is larger than that dem- onstrated by the quarantine testing (test failure?); or (ii) the results of the quarantine tests demonstrate a fundamental host range unexpectedly large compared with the known field host use in the native range. In both these instances a return to studies in the native range is the only recourse to understand- ing test results and keeping science as the basis of the risk assessment. This case study considers the broom seed beetle, Bruchidius villosus, a biological control agent against Scotch broom (Cytisus scoparius) where native-range studies have been of paramount importance for a complete risk assessment. In this species, as in many others used as biological control agents, host range is primarily defined by adults selecting oviposition sites. Adult oviposition behaviour, however, is a highly complex process driven by a whole range of external and internal stimuli and phenological as well as physiological host compatibility (Miller & Strickler 1984; Sheppard et al. 2005). Also ‘it is very likely that the evolution of host special- isation can result from the evolution of the nervous system that determines host selection behaviour’ (Jermy & Szentesi 2003). This case presents an example where behavioural com- plications arose while determining host specificity and how native-range studies helped untangle them. BACKGROUND The old world genus Bruchidius has over 250 described spe- cies with many more undescribed (Udayagiri & Raj Wadhi 1989). Most species in the genus attack seeds of Fabaceae (Johnson 1981). Most species are monospecific, or specific to one or a few closely related genera (Jermy & Szentesi 2003). Bruchidius villosus (=B. ater, =B. fasciatus) is a European seed beetle used in New Zealand and Australia as a classical biological control agent against C. scoparius (Syrett et al. 1999). Following its accidental introduction into eastern North America, US States on the Pacific coast have also started introducing and distributing the species in the west. In the original host-specificity tests conducted for New Zealand in the 1980s, a UK population of B. villosus (then