EDMONTON - Preventing a West Nile outbreak doesn't require killing every mosquito, says a new study by three researchers at the University of Alberta's centre for mathematical biology.

Using dead-bird counts and other disease data from New York state, where the North American outbreak of the virus began in 1999, the trio calculated that a 40- to 70-per-cent reduction in mosquito numbers in the spring would have been enough to forestall the continuation of the state's outbreak in 2000.

"What our model tells us is you don't need a mosquito-free city," said Marjorie Wonham, who with Mark Lewis and Tomas de-Camino-Beck, wrote the paper published today in the prestigious British journal, Proceedings of the Royal Society.

"You can figure out using the model how low a mosquito population you need to prevent an outbreak," said Wonham, who studied biology before learning the math used to model disease transmission.

Rather than using dead birds as a West Nile indicator, the model uses the number of mosquitoes per crow, one of the birds particularly susceptible to the virus.

In New York, the researchers estimated there were 7.5 to 15 mosquitoes per bird in the spring. If the ratio had been below 4.6 per bird, they suggest, the outbreak could have been prevented.

The model suggests that if you know the mosquito threshold you need to reach, you could avoid spraying more insecticide and filling more wetlands than necessary, Wonham said.

They also found that getting rid of birds such as crows, jays and magpies that are among the most susceptible to the West Nile virus increases the chance of an outbreak.

What matters is the number of mosquitoes per bird, Wonham and Lewis said.

"If there were fewer birds, each bird would get bitten by more mosquitoes. Then more birds would be infected. Then in return, when a mosquito bit a bird, it would have a higher chance of getting infected," said Lewis, who heads the centre for mathematical biology.

"This was an interesting finding for us because we certainly read about ideas of controlling birds," he said.

"Our conclusion is that is counter-productive."

West Nile virus was first discovered in the West Nile area of Uganda in 1937. In the summer of 1999, seven people in New York state died and the virus subsequently spread rapidly across the U.S. and most of Canada, reaching Alberta last summer.

Last year, nearly 9,000 cases of West Nile infection and 212 deaths were reported in the U.S.

Alberta had more than 272 cases while across the country Health Canada reported 851 probable and 466 confirmed cases. Ten Canadians died of the disease.

One thing the U of A group's model doesn't do is predict how bad the next West Nile season will be. Nor will it be of much use to areas outside of New York until more disease data is available.

It's really just a starting point, Wonham and Lewis said, and will need refinement and further testing. One big assumption, they added, is that the numbers they used in modelling the New York outbreak, culled from several sources, were valid.

"The public health surveillance data they have there is unparalleled," Wonham said. "They've got the number of infected mosquitoes by species, the number of infected birds, the number of dead birds."

In Edmonton, it will likely take another couple of years to get similar data, Wonham said.

One advantage Edmonton does have is that its northern climate makes mosquito populations a seasonal phenomenon and easier to control than such areas as Louisiana, where mosquitoes are a year-round problem.

aogle@thejournal.canwest.com