Background
The bacterium Bacillus thuringiensis serovar israelensis (Bti) is commercially produced in various formulations for use as a larvicide worldwide, targeting especially the aquatic larval stage of mosquitoes. However, there is a concern that repeated Bti treatments may have both direct and indirect impacts on non-target organisms (NTOs) and the ecosystems they inhabit. This review evaluates the evidence for such impacts.
Methods
Literature was searched using six bibliographic databases, two search engines, and on specialist web sites. Eligibility screening was performed in two steps on (1) title/abstract, with consistency among reviewers assessed by double-screening 557 articles and (2) full text. Articles included after full text screening were critically appraised independently by two reviewers. Disagreements were reconciled through discussions. Key parameters of included studies are presented in narrative synthesis tables, including risk of bias assessments. Meta-analyses comparing treated with untreated ecosystems and using either the raw mean difference or log response ratio as effect size were performed.
Review findings
Ninety-five articles covering 282 case studies were included in the review. From these, we identified 119 different response variables, which were divided into nine outcome categories. Most studies investigated NTO abundance or life history (reproduction related outcomes), but diversity and community composition are also well represented as outcome categories. The studies are highly variable in methodology, rigor, and spatio-temporal scale, spanning 1 day to 21 years and from < 1m2 to > 10,000 m2. Our metanalyses revealed a consistent negative effect of Bti treatment on abundances of Chironomidae and Crustacea, and also on chironomid emergence, although from a more restricted set of studies and regions. For most remaining response variables, we judged meta-analysis unfeasible, due to low study numbers or insufficient reporting of methods and results.
Conclusions
There is now a significant body of studies documenting effects of mosquito control using Bti on NTOs or other ecosystem properties, especially associated with negative effects on Chironomidae, as apparent from our meta-analyses. Accordingly, we suggest the potential for negative NTO or other ecosystem effects of Bti treatment should not be discounted a priori. Once a decision to proceed with Bti treatment has been taken, priority should be given to a well-designed program of ongoing monitoring and assessment. The paucity of rigorous studies conducted with low bias risk for most response variables undermines our capacity for evaluating how common many of the effects documented might be. Future research would benefit from a rigorous and well-replicated approach to studying Bti impacts in semi-field mesocosms or in the field, combined with a greater rigor in reporting key methodological details. A greater focus is needed on understanding the environmental factors which regulate the wider effects of mosquito control using Bti on NTOs and ecosystems, to enhance our capacity for predicting where and when Bti is most likely to have additional, negative and indirect ecological impacts.
