France's farm minister Stephane Le Foll withdrew Syngenta's marketing permit for the pesticide Cruiser OSR in June, citing evidence of a threat to the country's bees.

But a study by Britain's Food and Environment Research Agency with the University of Exeter says the results of the original research were flawed.

The study, published in the journal Science, does not deny that pesticides could be harmful to individual bees but argues there is no evidence they cause the collapse of whole colonies.

"We do not yet have definitive evidence of the impact of these insecticides on honeybees and we should not be making any decisions on changes to policy on their use," said James Cresswell, the ecotoxicologist who led the latest study.

The previous research, led by French scientist Mikaël Henry and published in Science in April, showed the death rate of bees increased when they drank nectar laced with the neonicotinoid pesticide, thiamethoxam, the active ingredient in Cruiser OSR.

Neonicotinoids are among the most widely-used agricultural insecticides.

Henry's work calculated this would cause a bee colony to collapse completely but Cresswell said the French study seemed to have used an inappropriately low birth rate, underestimating the rate at which colonies can recover from the loss of bees.

"They modeled a colony that isn't increasing in size and what we know is that in springtime when oilseed rape is blossoming they increase rapidly," Cresswell told Reuters.

The French study has been cited by scientists, environmentalists and policy-makers as evidence of the impact of these pesticides on bees, which are declining around the world.

"We know that neonicotinoids affect honeybees, but there is no evidence that they could cause colony collapse," said Cresswell. "When we repeated the previous calculation with a realistic birth rate, the risk of colony collapse under pesticide exposure disappeared."

DOSAGE DOUBTS

Cresswell said Henry's research also used a dosage of pesticide equivalent to a whole day's intake by the bees, akin to testing the effect of coffee on people by making them drink eight cups in one go, rather than spread out over the day.

Henry said he was "perfectly comfortable" with the new findings, adding in an emailed response to Reuters: "The model we used predicts a major deviation from the expected colony dynamics, rather than a collapse per se."

The April paper in Science said exposure to thiamethoxam "causes high mortality due to homing failure at levels that could put a colony at risk of collapse".

Syngenta lost a court bid in July to overturn the French ban on Cruiser OSR, which meant the pesticide was not used for rapeseed sowing in August and September.

"It's important for us that what we had argued is now supported by a scientific study," Syngenta France spokesman Laurent Peron told Reuters. "We are going to use the findings of this study but it's too early to say in what way."

The French farm ministry declined to comment.

Environmental campaign group Friends of the Earth argued that sales of the pesticide should be halted while any doubt about their impact on bees remained.

"Neonicotinoid pesticides cannot be given a clean bill of health until they have been properly tested for their effect on all bees, not just honeybees," said campaigner Paul de Zylva.

Cresswell said: "I am definitely not saying that pesticides are harmless to honeybees, but I think everyone wants to make decisions based on sound evidence, and our research shows that the effects of thiamethoxam are not as severe as first thought."

(Additional reporting by Gus Trompiz in Paris; Editing by Belinda Goldsmith and Pravin Char)

• A Permanent Economic Emergency