Summary 1.Several genetically modified herbicide-tolerant (GMHT) crops have cleared most of the regulatory hurdles required for commercial growing in the United Kingdom. However, concerns have been expressed that their management will have negative impacts on farmland biodiversity as a result of improved control given by the new herbicide regimes of the arable plants that support farmland birds and other species of conservation value. 2. The Farm-Scale Evaluations (FSE) project is testing the null hypothesis that there is no difference between the management of GMHT varieties of beet, oilseed rape and maize and that of comparable conventional varieties in their effect on the abundance and diversity of arable plants and invertebrates. The FSE also aims to estimate the magnitude and consider the implications of any differences that are found. 3. The experimental design of the FSE is a randomized block, with two treatments allocated at random to half-fields. The target sample is around 60-75 fields for each crop, selected to represent variation of geography and intensity of management across Britain. The experimental crops are managed by commercial farmers as if under commercial conditions. 4. Biodiversity indicators have been selected from plants and terrestrial invertebrates to identify differences between crop management regimes that may result in important ecological changes over larger scales of space and time. Field sampling is at fixed points, mainly along transects from the field boundary, starting before the crop is sown and continuing into following crops. 5. Synthesis and applications. The FSE is best considered as an investigation into the effects of contrasting crop management regimes on farmland biodiversity, rather than a study of the effects of genetic modification. It could become a model for future studies of ecological effects of the way we use and manage agricultural land.

Effects of genetically modified herbicide-tolerant (GMHT) and conventional crop management on invertebrate trophic groups (herbivores, detritivores, pollinators, predators and parasitoids) were compared in beet, maize and spring oilseed rape sites throughout the UK. These trophic groups were influenced by season, crop species and GMHT management. Many groups increased twofold to fivefold in abundance between early and late summer, and differed up to 10-fold between crop species. GMHT management superimposed relatively small (less than twofold), but consistent, shifts in plant and insect abundance, the extent and direction of these effects being dependent on the relative efficacies of comparable conventional herbicide regimes. In general, the biomass of weeds was reduced under GMHT management in beet and spring oilseed rape and increased in maize compared with conventional treatments. This change in resource availability had knock-on effects on higher trophic levels except in spring oilseed rape where herbivore resource was greatest. Herbivores, pollinators and natural enemies changed in abundance in the same directions as their resources, and detritivores increased in abundance under GMHT management across all crops. The result of the later herbicide application in GMHT treatments was a shift in resource from the herbivore food web to the detritivore food web. The Farm Scale Evaluations have demonstrated over 3 years and throughout the UK that herbivores, detritivores and many of their predators and parasitoids in arable systems are sensitive to the changes in weed communities that result from the introduction of new herbicide regimes.

The Farm Scale Evaluations of genetically modified herbicide-tolerant crops (GMHT) were conducted in the UK from 2000 to 2002 on beet (sugar and fodder), spring oilseed rape and forage maize. The management of the crops studied is described and compared with current conventional commercial practice. The distribution of field sites adequately represented the areas currently growing these crops, and the sample contained sites operated at a range of management intensities, including low intensity. Herbicide inputs were audited, and the active ingredients used and the rates and the timings of applications compared well with current practice for both GMHT and conventional crops. Inputs on sugar beet were lower than, and inputs on spring oilseed rape and forage maize were consistent with, national averages. Regression analysis of herbicide-application strategies and weed emergence showed that inputs applied by farmers increased with weed densities in beet and forage maize. GMHT crops generally received only one herbicide active ingredient per crop, later and fewer herbicide sprays and less active ingredient (for beet and maize) than the conventional treatments. The audit of inputs found no evidence of bias.

Farmland biodiversity and food webs were compared in conventional and genetically modified herbicidetolerant (GMHT) crops of beet (Beta vulgaris L.), maize (Zea mays L.) and both spring and winter oilseed rape (Brassica napus L.). GMHT and conventional varieties were sown in a split-field experimental design, at 60-70 sites for each crop, spread over three starting years beginning in 2000. This paper provides a background to the study and the rationale for its design and interpretation. It shows how data on environment, field management and the biota are used to assess the current state of the ecosystem, to define the typical arable field and to devise criteria for selecting, sampling and auditing experimental sites in the Farm Scale Evaluations. The main functional and taxonomic groups in the habitat are ranked according to their likely sensitivity to GMHT cropping, and the most responsive target organisms are defined. The value of the seedbank as a baseline and as an indicator of historical trends is proposed. Evidence from experiments during the twentieth century is analysed to show that large changes in field management have affected sensitive groups in the biota by ca. 50% during a year or short run of years-a figure against which to assess any positive or negative effects of GMHT cropping. The analysis leads to a summary of factors that were, and were not, examined in the first 3 years of the study and points to where modelling can be used to extrapolate the effects to the landscape and the agricultural region.

From 2000-2003 a range of Farm Scale Evaluation (FSE) trials were established in the UK to assess the effect of the release and management of herbicide tolerant (HT) crops on the abundance and diversity of farmland wildlife compared with their conventionally managed non-GM-equivalents. The objective of this research project was to investigate gene flow within the winter (WOSR) and spring oilseed rape (SOSR) FSE trials and to develop a statistical model for the prediction of cross-pollination frequency that can be used to evaluate current separation distance guidelines. Seed samples were collected from the non-GM half of the trial sites and were tested for evidence of cross-pollination from the GM HT halves using a quantitative PCR assay specific to the HT (bar) gene. Rates of cross-pollination were found to decrease rapidly with increasing distance from the GM source. The quantitative data were subjected to statistical analysis and a two-step model was found to provide the best fit for the data. Significant differences were found between the results for WOSR, SOSR and varietal association (VA) crops. The model predicted that the %GM content (including upper 95% confidence limits) of a sample taken at a distance of 50 m away from the GM source would be 0.04% (0.84%) for WOSR, 0.02% (0.39%) for SOSR, 0.77% (21.72%) for WOSR VA and 0.37% (5.18%) for SOSR VA. The data and models presented here are discussed in the context of necessary separation distances to meet various possible thresholds for adventitious presence of GM in OSR.