When regulators in Canada and abroad decide whether to allow genetically engineered salmon on the dinner plate, they'll be relying heavily on the research of a West Vancouver-based scientist.

Robert Devlin has been conducting risk assessments of GE salmon since 1989 at Fisheries and Oceans Canada's Centre for Aquaculture and Environmental Research in West Vancouver , trying to answer the question of what would happen if the salmon were released into the wild. It's the only facility studying GE salmon anywhere in the world without commercial funds.

All that work has been building to this moment, as Massachusetts-based AquaBounty has sought approval from the Food and Drug Administration in the U.S. to bring their AquAdvantage genetically engineered Atlantic salmon to American menus.

If they receive final approval, an application in Canada is likely next on the list, as the company plans to produce the eggs at a research facility in Prince Edward Island.

But 20 years into Devlin's research project, he said there's still more questions than answers. Simply put, scientists can't say for sure how fish will respond to different environments, he said, and there is a serious risk if they escape into the environment.

"It's complex. It's really puzzling," he said of the work. "I think I've embraced the fact that doing experiments that uncover uncertainty and informs regulators of that is just as important (as finding all the answers)."

The first 10 years on the project was spent creating the types of fish they wanted to test on -- in particular by altering the growth hormone. While wild salmon have a regulator in their growth hormone that means it's only produced in the pituitary gland and can turn off when it's not needed, GE fish have changed that.

"What we can do is . . . take that gene and remove that regulatory portion of the normal growth hormone gene and replace it with a dumb regulator that just keeps cranking out growth hormone all the time and from many different tissues," he said.

That gives the fish a much greater appetite. It can grow to 10 times the size of a normal fish in the first year when in a tank environment, while also outcompeting wild fish for food -- but in the wild the effects are less clear.

Devlin used several large tanks designed to mimic the natural environment under big white tents just across the fence from Marine Drive, and found different results.

The tanks contain one million litres of seawater and wild-type fish designed to act as a control group to test the similarities to the fish beyond the research facility's fence. In these tanks, the GE fish grow only slightly faster than wild fish, though they show much less fear of predators. When predators are introduced, the GE fish are much more likely to become dinner before ever reaching a restaurant.

That might suggest that GE fish wouldn't do well in the wild, however it's not that simple, said Devlin. GE fish may also venture into new habitats where wild fish don't go. What the research shows is that the environment significantly affects survival and behaviour, he said.

"How do we combine those data sets into a complete synthesis such that we would predict what would happen in the full complexity of nature? Right now, that's beyond our capacity," he said. "What it means . . . for the natural environment is a bit hard for us to predict at that time."

Another question Devlin has examined is whether the fish could breed.

In West Vancouver, as elsewhere, they sterilize the fish, but the process isn't perfect. They're able to achieve an average 99.8-per-cent sterilization rate, and while that sounds like a lot, in previous international cases up to 50,000 fish have escaped at once. That could mean 100 fertile GE fish on the lam.

GE salmon can also successfully breed with wild-type fish without human help, muddying the waters even further.

The risk could be mitigated, however, by regulations requiring genetically modified fish be raised in inland tanks where the risk of escape is greatly reduced, he said.

But while AquaBounty has been pushing the regulatory envelope, there's little interest in the technology still from the Canadian fish farm industry, said Ruth Salmon, spokeswoman for the Canadian Aquaculture Industry Alliance.

The alliance policy, passed in October 2007, before AquaBounty sought regulatory approval in the United States, says it doesn't support transgenic fish for human consumption.

It's not that they have concerns, said Salmon, it's just that it's not needed. "The industry feels we have breeding practices today that are giving us the production of a high-quality product that's growing well," she said, adding consumers aren't interested in seeing genetically modified salmon on the menu, either.

But Devlin has no doubt that a commercial application will come eventually. Whether it should be accepted and under what conditions, he said, is not up to him to decide -- he simply wants the government to have the science to back up their decision.

And to do that work just a few short hops from where he grew up fishing on the Capilano River is even better, he said. "I love science, being able to investigate things that have never been looked at before is quite a privilege."