The results of the world's largest ever trial of GM crops show that two out of the three tested - oilseed rape and sugar beet - had a worse impact on farmland wildlife than conventional crops.

The researchers stress that results measured the impact on wildlife of the herbicide regimes used - either glyphosate or glufosinate ammonium - not of the crops themselves, which are genetically engineered to be herbicide-resistant.

"Critically, all our results are explained by the application and timing of different herbicides, not by virtue of the plants being GM or not," said Les Firbank, coordinator of the UK-based trials, at the launch of the results in London on Thursday.

But opponents of GM crops are delighted by the findings, which now go to the UK government and its official advisers on GM crops for appraisal. The negative environmental findings could give the British government acceptable grounds under world trade rules to ban the commercialisation of the two crops.

The third GM crop tested, maize, was better for wildlife than its conventional counterpart. But opponents say this result has been invalidated by the impending European ban on atrazine, the weedkiller used on the conventional maize against which the GM maize was judged.

Whole package

Crop yields were not considered in the trials. The organisers say this and many other factors influence the effect of farming on wildlife. "It's the whole package of how the countryside is managed," says Firbank.

Chris Pollock, chairman of the scientific steering committee which oversaw the £6-million trials, added: "We hope our results feed into the wider debate about GM in Britain, and into the debate about what kind of landscape we need."

The organisers believe the four-year "farm scale evaluations" on 280 fields throughout Britain have been a resounding success. "It's the first time a novel agricultural technology has been trialled extensively before it's been introduced, rather than after," said Pollock.

For each crop, individual fields were divided into two with the GM crop grown in one half and the conventional in the other. To gauge the effect on wildlife, scientists monitored weeds and weed seeds, and collected beetles and other insects in traps.

• Oilseed rape: Weeds were 1.7 times as plentiful in the conventional plots and produced five times as many seeds. The store of these seeds left in the ground was doubled after conventional treatment, providing a vital food resource for the birds such as skylarks that have declined dramatically since the start of modern farming 60 years ago. A quarter fewer butterflies were recorded around the GM plots.

• Sugar beet: About 1.3 times as many weeds grew in conventional plots, and treble the number of seeds. Again, there was an increase in the seeds left in soil. There were 40 per cent fewer butterflies.

• Maize: The situation was totally reversed. There were three times as many weeds in the GM plots, and 2.3 times as much seed produced. But puzzlingly, this did not increase the amount of seed left in the soil.

Geoff Squire, a trials coordinator from the Scottish Crop Research Institute in Invergowrie, Dundee, said that if the crops had been grown for the past 10 years, the oilseed rape and the sugar beet would have triggered further declines in wildlife, but the maize would have increased biodiversity on farms.

"Statistically flawed"

Opponents of GM crops were delighted with the results. "Our main reaction is that a set of trials designed to show GM in a good light have had such a dramatic negative impact," said Charlie Kronick of Greenpeace.

However, Greenpeace and others have always denounced the trials as a fix, and activists attempted to sabotage them by physically pulling up the crops. Earlier in 2003, Friends of the Earth said that the results would be statistically flawed.

The researchers defended their results on maize, despite the impending withdrawal of the atrazine against which the GM herbicides were compared. But they acknowledged that results might need to be "recalibrated" and that extra field research might need to be done to gather data on whatever regime replaces atrazine in conventional maize.

The results will be unwelcome for the biotechnology industry. And on the eve of the results' release, Monsanto, the world's largest producer of GM crops, announced the closure of its European cereal seed business. But the company says it remains firmly committed to developing GM crops.

Journal reference: Philosophical Transactions of the Royal Society of London B (vol 358, p 1775 - 1913)