Sir

The first genetically modified (GM) crops being proposed for commercial planting in the United Kingdom have been altered to make them less sensitive to broad-spectrum herbicides. These crops are intended to allow more efficient weed management and herbicide regimes for the farmer, reduced frequency and quantity of herbicide applications, and increased market share for the suppliers. Work has now begun on farm-scale studies of maize and spring oil-seed rape, with winter oil-seed rape studies starting later this year. Many objections to GM crops have been raised, and there is pressure on farmers not to take part in this research programme (see Nature 398, 651–656; 1999). Indeed, one farm, near Swindon in the west of England, has recently withdrawn from the study.

Because of the widespread public interest in GM crops, we, as representatives of the consortium commissioned to conduct this research, think it is worthwhile to outline the background and purpose of these farm-scale evaluations. The research addresses the concern that the changing management of the GM herbicide-tolerant (GMHT) crops could result in reductions of weed and invertebrate populations on which farmland birds and other wildlife depend. This concern has been voiced most cogently by the conservation groups English Nature and the Royal Society for the Protection of Birds (RSPB).

We are aiming to test the null hypothesis that there are no differences between the biodiversity associated with the growing of GMHT crops and comparable non-GM crops at the farm scale. The study will look for positive and negative effects. These are likely to be indirect, resulting from crop and rotation management, rather than from a direct effect of the use of GM plant breeding technology. Indeed, if herbicide resistance had been introduced by traditional breeding, the design of the study would have been the same. Farmers will grow and manage both GM and non-GM crops as they would do commercially.

The only constraints are that the varieties being compared are as similar as possible in other characteristics, and also that all other agronomic practices are kept the same, unless commercial reasons dictate otherwise. So we anticipate that the herbicide regimes will differ, but that any insecticide or fungicide should be applied to both treatments on each farm on the same day, unless there are clear agronomic reasons that may in themselves be the result of growing the different crops. We will also evaluate effects on biodiversity in following crops. These crops will be the same, and will be managed in the same way, following both GM and non-GM treatments, unless there are clear agronomic reasons for differences in rotations or management.

It is not practical to record population responses for all species in the arable system. We are using indicators of biodiversity that are likely to be sensitive to the treatments, and reflect processes that may lead to significant ecological shifts that cannot be detected directly given the time and spatial scales available for the study. Effort will be concentrated on species groups that do not forage over wide areas or occupy higher trophic levels. These include vegetation in and around the crop, the field seed bank, earthworms, snails and slugs, ground beetles, bugs (Heteroptera), foliage arthropods and invertebrate biomass (with moth and sawfly larvae measured separately). Bees and butterflies will also be assessed. These are all being recorded using standard protocols that are being tested and refined during this year.

Birds are not included in the field study because they range too widely to show real effects when only single fields are being considered, although data on invertebrates and plants will provide measures of resources available to them.

Work in the first season is on a pilot scale to ensure that monitoring is matched to the details of crop management. There will then be three seasons of the summer crops and at least two of the winter crop, at the full scale of around 20 treatment pairs per crop per year. We will select from the pool of available farms using a stratified random procedure; the experimental treatments will be allocated at random within each farm. The GM and control crops will be grown in a split-field or a paired-field plan; work in this first year will confirm which is the more appropriate. There are valid arguments for and against both configurations. In a split field, the two halves of the field will have had similar histories, reducing the variation in biodiversity indicators before treatment. The paired-field design gives less chance of interference between the treatments and is more realistic in terms of the structure of the field boundaries. Both configurations are included in the first-year sites.

The work is being conducted by a UK consortium of the Institute of Terrestrial Ecology, the Institute of Arable Crops Research and the Scottish Crop Research Institute. It is funded by the Department of the Environment, Transport and the Regions, the Ministry of Agriculture, Fisheries and Food, and the Scottish Office.

A steering committee will oversee the progress of the work and ensure the scientific quality and integrity of results. The committee includes independent scientists, including experts from English Nature, RSPB and the Game Conservancy Trust. Many results will not be available until the end of the project in 2002.

The role of scientists is to provide the evidence on which to base a sound risk assessment of the effects of herbicide-tolerant GM crops on biodiversity.

Our evidence will, we trust, provide an important input into a rational debate about the adoption of GM crops.