The unpredicted host use of Chamaecytisus profiler by the introduced weed biocontrol agent, Bruchidius villosus: a retrospective analysis and explanation

Abstract

This study investigates unpredicted host use by a weed biological control agent in New Zealand. Bruchidius villosus, a seed-feeding beetle, was released into New Zealand in 1986 as a biocontrol agent to aid in the control of Cytisus scoparius (Genisteae), a European shrub that has become widespread in New Zealand. Although host specificity tests had predicted that Chamaecytisus prolifer would not be attacked, in 1999 B. villosus was found utilising this plant as well as C. scoparius in the field in New Zealand. This finding prompted a thorough re-examination of historical rearing records, which showed that B. villosus does in fact exhibit a wider host range than was initially thought and forecast by New Zealand and UK testing. Uncertainty over the identity of B. villosus (which is referred to in some literature as B. ater and other synonyms) appears to have contributed to the misunderstanding of its known host range. To ascertain why host specificity tests (choice tests on cut shoots) failed to predict that Ch. prolifer would be a suitable oviposition host for B. villosus, a series offield and laboratory experiments were undertaken.

The field host range of B. villosus in New Zealand was established: C. scoparius and Ch. prolifer were the only hosts oviposited on and no other native or exotic pod-bearing relatives of C. scoparius were found to be hosts in the field. However, subsequently (2003) beetles have also been collected from Lupinus arboreus in the field.

In no-choice sleeved tests on whole plants in the field C. scoparius, Ch. prolifer and two additional species, Cytisus multiflorus and Genista monspessulana, were oviposited on, and adults emerged from all these species except C. multiflorus. These results supported European literature records that B. villosus has a broader host range.

Mitochondrial DNA for B. villosus reared from a number of host plants from the UK, France, Spain, Germany, Hungary, Canada, and the USA was analysed to determine whether the current concept of B. villosus as a single species is valid. The alternative hypothesis was that B. villosus included a number of sibling species separated either geographically or by host plant. Results indicated that all beetles belong to the same species.

Performance of B. villosus on C. scoparius and Ch. prolifer in the field was compared. Larval survival was significantly higher and development from egg to adult significantly faster on C. scoparius indicating it was the more suitable host for B. villosus. Despite apparently being a less suitable host, B. villosus destroys approximately 40% of available Ch. prolifer seed, compared to about 90% of C. scoparius seed.

Aspects of the methodology of the original testing were examined to identify any feature of the experimental design that may have caused the failure of the original choice oviposition tests to detect the acceptability of Ch. prolifer. Firstly the effect of larval rearing host, adult food type and no overwintering period were studied. In all cases beetles oviposited on Ch. prolifer, though showing a strong preference for C. scoparius. Secondly quantity of oviposition resource, cage size and orientation of plant material were also investigated. Beetles laid eggs on Ch. prolifer when they were presented with equal amounts of test plant and control material in replicates identical in design to the original tests, but when the quantity of control pod resource presented in choice tests was half that of the test plant, oviposition increased on the test plants Ch. prolifer (significantly) and C. multiflorus. Cages of two different sizes and cut shoots presented horizontally or vertically had no significant effect on oviposition.

In order to determine whether a host range expansion had occurred in the New Zealand population of B. villosus since its introduction from the UK, a comparison of oviposition preference was undertaken between New Zealand beetles and a newly imported population from the UK. Sample size was increased from two replicates (ten female beetles) in the original 1985 tests to ten replicates (50 female beetles). Beetles newly imported from the UK laid eggs on Ch. prolifer as well as on C. scoparius which ruled out the development of a host race with a broader host range in New Zealand. In only 40% of replicates were eggs laid on Ch. prolifer indicating that 20 (or less) females found Ch. prolifer acceptable for oviposition. Results from this experiment and from previous investigations suggested that insufficient replication could be responsible for the failure to detect oviposition on Ch. prolifer due to a high level of individual variation in oviposition preference among female beetles.

Oviposition preference by individual female B. villosus beetles, from New Zealand and UK populations, between C. scoparius and Ch. prolifer was measured in a laboratory choice test. The variation in numbers of eggs laid was high, especially on Ch. prolifer (CV = 148% New Zealand population, CV = 166% UK population) compared with numbers on C. scoparius (CV = 97% New Zealand population, CV = 50% UK population). Although both populations showed an overall preference for C. scoparius, New Zealand populations showed higher acceptance of Ch. prolifer (New Zealand population: mean number of eggs 8.6 ± 1.6 SEM on C. scoparius, 6.5 ± 1.8 SEM on Ch. prolifer; UK population: mean number of eggs 12.7 ± 2.0 SEM on C. scoparius, 3.9 ± 2.0 SEM on Ch. prolifer). Nine of 29 New Zealand beetles laid more eggs on Ch. prolifer compared to none of the nine UK beetles that laid eggs. A sibling experiment using three mothers showed that there were no strong maternal influences on overwintering survival, longevity, fecundity or oviposition preference of offspring.

Although as some beetles from the newly imported UK population accepted Ch. prolifer it was possible to rule out the likelihood of a host shift having taken place, there are indications from the differences in individual preferences shown between New Zealand and UK populations that differentiation in populations may be occurring. There seemed to be little maternal influence on preference, but further study is required to determine whether observed differences are inherited and whether host race formation is likely. Beetles reared from Ch. prolifer were bigger, more fecund and had higher overwintering survival. Longevity was similar for beetles reared from either host, but survival from egg to adult was higher and development faster on C. scoparius, the host preferred overall for oviposition. Clearly original choice host specificity tests failed to detect that B. villosus accepted Ch. prolifer because insufficient replication was used: testing 20 female beetles instead of 10 would have reduced the chance of selecting a sample that would lay no eggs on Ch. prolifer from 11 % to 0.02%.

Lessons for host specificity testing in weed biological control are that high levels of individual variation in host preference require suitably high levels of replication to detect non-target effects and avoid type II errors (false negative results). This will inevitably generate larger numbers of type I errors (false positives) requiring detailed follow-up, preferably with individual insects. There are also indications that preference by an imported weed biological control agent for a non-target host plant may have changed over the 15 generations since its release, suggesting that test results for plant species that are new associations should be interpreted with particular care. The release of any agent that in no-choice tests accepts and develops fully on a valued host plant cannot be recommended for release, even if the level of attack is low.