Background: Hundreds of studies on environmental effects of genetically modified (GM) crops became available over the past 25 years. For maize producing insecticidal proteins from Bacillus thuringiensis (Bt), potential adverse effects on non-target organisms are a major area of concern and addressed in risk assessments. Reviews and meta-analyses have helped various stakeholders to address uncertainties regarding environmental impacts of the technology. Many field studies from Europe and other parts of the world have been published in the last decade, and those data are often not covered by previous meta-analyses. Therefore, we conducted a systematic review to answer the question: “Does the growing of Bt maize change abundance or ecological function of non-target animals compared to the growing of non-GM maize?” Methods: Literature published until August 2019 was searched systematically in 12 bibliographic databases, 17 specialized webpages, and reference sections of 78 review articles. Defined eligibility criteria were applied to screen titles, abstracts, and full texts of the retrieved references. A custom-made database was developed with quantitative data on invertebrate abundance, activity density, or predation/parasitism rates. Eligible data that did not fit the quantitative database were captured in detailed tables and summarized narratively. For the first time, a critical appraisal scheme for field studies on non-targets in GM crops was developed to estimate the risk of bias (internal validity) and the suitability to answer the review question (external validity) of all primary data. Meta-analyses on different taxonomic levels, functional groups, and types of Bt maize were conducted. Untreated Bt maize was either compared with untreated non-Bt maize, or with insecticide-treated non-Bt maize. The influence of contributions by private sector product developers on reported effects was investigated. Review findings: The database on non-target effects of Bt maize field trials contains more than 7200 records from 233 experiments and 120 articles. Meta-analyses on different taxonomic levels revealed only few and often nonrobust significant effect sizes when both Bt maize and non-Bt maize were untreated. Bt maize harboured fewer parasitoids (Braconidae, Tachinidae) of the European corn borer, the main target pest of Lepidoptera-active Bt maize, compared with non-Bt maize. Similarly, sap beetles (Nitidulidae), that are associated with Lepidoptera damage, were recorded less in Bt maize. In some analyses, a negative effect of Bt maize was observed for rove beetles (Staphylinidae) and hoverflies (Syrphidae) and a positive effect for ladybeetles (Coccinellidae), flower bugs (Anthocoridae), and lacewings (Neuroptera). However, those effects were not consistent for different analyses and often related to individual articles. When untreated Bt maize was compared with pyrethroid-treated non-Bt maize, more effect sizes were significant. In particular, populations of predators were reduced after pyrethroid treatment, while few data were available for other insecticides. Funnel plots showed no evidence for publication bias and the analyses of private sector contribution revealed no evidence for influence of vested interests. Conclusions about potential effects of Bt maize on vertebrates or on animals inhabiting off-crop habitats were not possible, because only few such studies fitting the format of direct Bt/non-Bt comparisons on plot or field level were identified. Conclusions: The current work largely confirmed previously published results. The effects of Bt maize on the community of non-target invertebrates inhabiting maize fields were small and mostly neutral, especially when compared with the effects of broad-spectrum pyrethroid insecticide treatments.
