Patch-level nontarget attack by the biocontrol weevil Mogulones crucifer: A temporary spillover effect requiring host plant presence?

Background/Question/Methods

Biological control can effectively reduce density of invasive plants, but presents risk of nontarget impact.  Concerns over potential nontarget attack by biocontrol insects threaten the future of this tool. Arguably, such concerns should be weighed against the risk of allowing invasive plants to spread unabated.

A primary question in assessing risk is whether an agent threatens the persistence of nontarget plant populations, since insect damage to individual plants may not translate to population-level effects. In this context, it can be argued that persistent attack is likelier to be more damaging than short-term feeding (i.e., temporary spillover). Here we describe one part of a larger study aiming to assess the risk to a native nontarget plant occurring in biocontrol release sites in Western Canada.

Groups of 300 biocontrol weevils (Mogulones crucifer) were released on nine rangeland sites in Alberta, Canada.  Sites contained natural populations of a native nontarget plant, Hackelia micrantha growing with or without the confamilial invasive weed: houndstongue  (Cynoglossum officinale). Both species were harvested from these sites two years later and dissected in the lab to evaluate M. crucifer attack.  Additional sites were harvested in the year of release, or 3-4 years after release to further examine temporal effects. 

Results/Conclusions

Overall, houndstongue plants were consistently attacked significantly more often and more severely than nontarget plants. Furthermore, on sites where houndtongue was absent, there was no attack found on nontarget plants two years post release. Older release sites where both species persisted (3-4 years post release), showed consistent strong attack on houndstongue and infrequent and generally low attack on the nontarget. Finally, attack intensity on houndstongue plants increased with plant size, whereas H. micrantha showed no size effect on attack intensity, even though the size variation of both species was similar.

We suggest that these findings indicate that weevils do not persist on sites containing nontarget  plants in the absence of houndstongue. We attribute the occurrence of nontarget attack on sites with houndstongue to chance encounter of weevils (emerging from houndstongue hosts) with nontarget plants (i.e., spillover attack). These results will be discussed with respect to risks to nontarget species in biocontrol systems.