BBC News' Toby Murcott reports on the research findings

If confirmed by larger-scale experiments, it would mean that GM crops developed at great expense would quickly become useless. It would also be a boost to those opposed to GM crops, who argue that the technology cannot solve problems of heavy pesticide use.

The research at the University of Arizona looked at the breeding cycle of the pink bollworm moth, a common pest of cotton. Their results suggest that the bollworm could, in theory, rapidly become resistant to the insecticide produced within a GM cotton plant called "Bt cotton".

Losing battle

This cotton was developed by Monsanto. Their spokesman points out the preliminary nature of the research but adds: "No matter what type of insect protection mechanism you use, short of squashing them under your foot, insects will become resistant to it."

No field-scale, long-term tests of pink bollworm resistance to Bt have been done

He admitted that: "If the Bt cotton was only viable for three or four years that would not be a financially viable option for us."

One million hectares of Bt cotton were planted worldwide in 1998, about three percent of the total cotton grown. The Monsanto spokesman says: "Our Bt cotton is in its fourth year in the US and has saved the application of over three million litres of chemical insecticides. If that is not an environmental benefit to everybody, I don't know what is."

The testing of GM crops has become an increasingly heated issue in the UK, with many trials being destroyed by campaigners. Greenpeace say: "Field trials cannot tell us any of the facts about GM crops: their long-term impact on ecosystems or whether they will adversely affect human health."

Professor Mick Crawley, a biologist at Imperial College, told BBC News Online that the Arizona bollworm study is "an interesting snippet, but these laboratory studies are absolutely useless in risk assessment. They just point out things to look out for in larger-scale trials."

Cotton boll damage

But he says a second generation of larger, longer term trials should go ahead, although they would be "tremendously expensive to do properly".

"Abandoning the trials would be patently absurd as there is no evidence of any danger. If anything, there's evidence of absence of danger. What people are doing with these little snippets are fanning the flames."

Professor Crawley adds: "I'm sure it will turn out that GM crops did not have problems, but it will take time for this realisation to be the one the public accepts.

"And all the while the policy makers have a problem because their voters, by and large, believe we should do away with the technology. But they couldn't possibly contemplate that because they know that Britain's competitors abroad would not stop."

He believes that: "The policy makers just have to do what they are doing and put up with unpopularity - there's no alternative."

Longer to develop

Bt cotton expresses its own insecticide, the Bt toxin, the gene for which is derived from a bacterium. The Arizona bollworm study showed laboratory-bred, Bt-resistant pink bollworm larvae feeding on Bt cotton take five days longer to develop into adult moths than do non-resistant larvae feeding on non-Bt cotton.

This compromises the strategy of "refuges", where non-Bt cotton plants are interspersed throughout the GM crop. Bollworms emerging from these and mating with any Bt-resistant bollworms will result in non-resistant offspring, as the gene is recessive. This delays the spread of resistance.

However, if the resistant and non-resistant bollworms emerge at different times they cannot mate and resistance may spread more quickly.

Monsanto claim resistance should not arise for 50 years but a 1997 report by entomologist Fred Gould of North Carolina State University suggested that resistance in the cotton-eating moth, Heliothis virescens, would emerge in 10 years. Other pests could become resistant to BT cotton in three to four years, Gould said.