Even on a high, dry plateau in central Oregon, it's hard not to find a riot of grass. Farmers grow Kentucky bluegrass for the seed, and lining irrigation canals are wild varieties of rabbit's foot and water bent, along with stubby lawn grasses that, grown thick, would make an excellent putting green.
And recently, a new variety has joined the scene: feral, genetically engineered bentgrass.
Early this past decade, the Oregon Department of Agriculture conducted a field trial just north of the town of Madras, planting 400 acres of bentgrass engineered by the Scotts Co. to resist the popular weedkiller Roundup. Scotts hoped the bentgrass, if approved, would be a financial bonanza, allowing golf courses to douse their fairways with herbicide, mimicking what has become standard practice at farms growing Roundup Ready corn and soybean countrywide.
Problems quickly became apparent. Unlike domesticated crops, which farmers must reseed each year and carefully tend, the bentgrass could thrive and persist in the wild. Only one major planting occurred, in 2002; the fields were fumigated after a year. But over the next half decade, the grass spread far across the plateau, its pollen scattered in the breeze, said Carol Mallory-Smith, head of Oregon State University's weed science program, who studied the migration.
While the grass is environmentally benign, it has proved impossible to eradicate.
"If we went out there right now," Mallory-Smith said, "[it's] still out there."
DNA splicing is nothing new. Nearly all of the cotton, corn and soy grown in the United States carry bacterial genes for pest or weedkiller resistance. Few companies, however, have sought to sell bioengineered plants and animals that can survive and even thrive without human tending. Firms feared a backlash should their creations escape, wary they could spread irreversibly like Scotts' bentgrass.
That resistance is creeping away.
Most prominently, AquaBounty Technologies is nearing approval of its fast-growing salmon that, if cleared by the Food and Drug Administration, would be the first genetically engineered animal on supermarket shelves (Greenwire, Sept. 21). Unlike the engineered cows and pigs that will follow it, AquaBounty's salmon remains a fundamentally wild creature, albeit with a massive appetite. And it is just the beginning of genetic engineering's spread to the wilderness.
Scientists have developed several other fast-growing fish, including trout and carp, and traits like improved nutrition are coming. One biotech firm backed by International Paper Co., ArborGen LLC, has trials of cold-tolerant eucalyptus trees scattered across the Southeast. And biofuel startups are developing biotech prairie grasses that grow fast and full in degraded soil, sparing food prices while surviving for years with little human intervention.
Another decade in the future, the changes get more complex. Companies are developing genetically engineered algae to consume carbon dioxide and sweat oil products, but barring a massive binge in greenhouse construction, the algae, designed to grow aggressively, will be cultivated in open ponds (ClimateWire, July 22). And while they may not be developed in wild plants, firms are close to growing commercial pharmaceutical components in farm crops.
This bioengineering surge is forcing regulators to encounter ecological challenges far knottier than in the past: how to avoid a repeat of the Oregon grass escape; how to judge, in the lab, the environmental risk posed by wild escapees and their genes; and, most fundamentally, how and whether to stop these plants and animals from pursuing nature's ultimate imperative -- reproduction -- in the wild.
"These are wilder species that are difficult to study," said Allison Snow, an ecologist at Ohio State University. The first generation of domesticated crops were hard enough, she said, but "it's more challenging when you have long-lived trees and grasses, and when you have fish and algae. ... As we get into more and more species with more and more traits, there are going to be many more challenges."
Just like in nature, not all genetically engineered life is created equal. Off the farm, several modified plants and animals, their traits obviously benign, are widely sold. Glowing zebrafish armed with jellyfish genes swim in U.S. pet stores. Hawaii's papaya trees are engineered to resist a devastating virus. Similarly, one effort to restore the American chestnut tree depends on splicing in genes to resist blight.
But then there are traits, like cold resistance or speedy growth, that could allow escaped organisms to beat out their wild counterparts in the struggle for food, sunlight and sex. These tweaks could counter the conventional wisdom among scientists that genetic engineering cripples life's ability to survive in the wild. And companies like ArborGen and Aquabounty have realized that, facing these fears, they must find a way to make their creations as sterile as mules.
Dubbed "bioconfinement" by scientists, sterility technology exists in various states of maturity, depending on the species. It is most advanced, thanks to a biological quirk, in fish. But for most plants and animals, bioconfinement is at best in beta, and few proposed systems have faced peer-reviewed trial, said Albert Kausch, director of the University of Rhode Island's Plant Biotechnology Laboratory.
"There are ways we could build in mechanisms to prevent and inhibit gene flow into the environment," Kausch said. "How severe does it need to be? I don't know. You could build in redundancies and fail-safes if necessary."
Scientists have received little general guidance from regulatory agencies on what will be required for their wild creations. Until a few years ago, the U.S. Department of Agriculture tended to view all proposed plants in the same light, scientists say, and the agency still has not indicated whether it will require bioengineered trees or grasses to be sterile, and, if so, how strictly they define sterility.
"This is a challenge that has not been addressed," Kausch said. If the government wants zero tolerance, imposing rules that no pollen can escape, "then draw that line in the sand and let the technology people develop what's necessary," he said.
In many ways, FDA and USDA are still improving their ability to evaluate environmental risks from next-generation biotech. FDA has quietly launched a pilot program to overhaul its animal biotech program, hiring new scientists for the task. And this year USDA reorganized its biotech division, creating a dedicated team to assess environmental risks, requesting $5.8 million for the upgrade.
FDA has pushed AquaBounty to add several layers of containment to its proposed salmon, including sterility controls. Most scientists agree these efforts are thorough. But the agency, in pushing these provisions, has so far allowed the company to ignore analyzing what would happen to the waters around its facilities in Prince Edward Island and Panama should fertile salmon escape.
"FDA and the company, AquaBounty, are presumptive," said Yonathan Zohar, head of the University of Maryland, Baltimore County's marine biotechnology department. "They make assumptions. They say, because it's Prince Edward Island, the eggs will not be able to survive. Because it's Panama, the fish will not be able to survive. ... I would like this to be done experimentally."
Unlike bioengineered trees or grasses, which would have to be grown in open fields to achieve any sort of commercial viability, companies tinkering with fish have several confinement advantages. The fish can be grown at inland facilities, as AquaBounty has proposed, which will spare oceans the pollution and escape problems caused by typical pen-based fish farms. And the salmon embryos, after conception, can be shocked into sterility with high efficiency.
But like plant biology, the techniques used to induce fish sterility are never 100 percent. Life is tricky, and often it finds a way to survive, said Tillmann Benfey, a fish biologist at the University of New Brunswick and one of the world's leading experts on aquaculture controls.
"In biology, nothing is perfect, especially when trying to get around sexual maturation," Benfey said. "It's really what evolution is all about."
'Fish are ... very plastic'
The government knows that sterility controls are a looming issue. Six years ago, the National Academy of Sciences released a study on bioconfinement science. Most of the technologies being developed were in their infancy, many mere ideas, it said. And while the research has rolled on since then, bioconfinement remains an untested field, with all degrees of uncertainty and hypothetical methods.
The best-known techniques for imposing sterility come not from manipulating genes, but entire chromosomes, the protein-DNA complexes that hold genes. Most animals carry two sets, one from each parent. If, by biological chance, a mammal ends up with three full chromosome sets, it is unlikely to be born. Mammals are rigid like that.
Fish, however, are a different story.
"Fish are unusual vertebrates in that they're very plastic," said John Buchanan, AquaBounty's scientific director. For example, female fish flooded in testosterone will develop as males, though they only carry female genes, a process AquaBounty uses to produce all-female stocks. And fish that carry three chromosome sets -- called triploids -- are able to function much like more stress-prone varieties of their normal peers, except in one way: They can't reproduce.
No one is quite sure exactly how the fish survive with three chromosome sets, or how it arrests sexual development.
"The mechanism of sterility has never been studied very well," said Serge Doroshov, a fish biology professor at the University of California, Davis. But it works, with particular success in salmon; carp, though, can reproduce despite an added chromosomal burden.
Over the years, aquaculture specialists have discovered ways to use shocks of high pressure to arrest the development of fish eggs just after fertilization, before they have shed their mother's extra chromosome. It is commonly used on trout stocks for sport fishing and has become a well-developed, if not foolproof, technique, said Benfey, whose work on triploids guided AquaBounty.
For its FDA application, AquaBounty has promised more than 95 percent of its bioengineered eggs will be triploid. It is a high rate, probably the best that can be done with current technology in a commercial setting, Benfey said. But it also means each shipment will also carry fertile females.
"That's the story in biology," Benfey said. "Nothing is 100 percent."
Benfey's phrase is a familiar refrain. The inability for one system to ever promise full containment means that, if regulators have any hope of controlling wild bioengineered organisms, multiple systems will need to be in place, said Ohio State's Snow, who was an author on the National Academy report.
"If you're going to even come close, you're going to have a lot of redundant mechanisms," Snow said.
Bioengineered salmon are actually in a promising place because of the need for redundancy, according to AquaBounty's Buchanan. The firm will sell only females to inland fish farms with strict physical separation from natural waters. The company does not have to worry about the wind, about pollination.
"We're in a better place than crops," Buchanan said.
For years, pressure has been building for farmers to move off growing corn for ethanol, given fears widespread cultivation could cause a repeat of 2008's food price rise. Instead, future biofuels should be derived from perennial plants -- poplar trees and prairie grasses -- that can grow in poor soil and survive for many years. Of course, they can survive outside the fields, too.
Companies developing bioengineered perennials are facing a future reckoning, according to URI's Kausch, an influential figure in the field. If they want to reach commercial size, sterility controls have to be embedded in the plants' platform, he said. They have to be as fundamental as carburetors were to early automobiles. "You don't have a car without a carburetor," he said.
So far, there is no carburetor, but the company getting closest to installing one is ArborGen, a biotech tree firm backed by International Paper that, early this month, announced plans to hold an initial public offering. ArborGen has developed cold-tolerant eucalyptus that it wants to sell to plantations across the Southeast, turning the tree into the region's biological equal of Applachian coal (Greenwire, Jan. 29).
ArborGen's eucalyptus also carries a system to limit pollen spread. The control depends on a bacterial gene to produce a toxic protein that slices apart genetic material in a cell, causing death; the gene is expressed only in the tree's reproductive organs. ArborGen has yet to publish any peer-reviewed data on the system's effectiveness, though it has flowering trials operating for several years, and has sought full deregulation of its trees since 2008. The data should be out by early 2011, according to Nancy Hood, ArborGen's communications director.
Kausch has worked with the same sterility controls used by ArborGen. Their technology is well-done but unlikely to work all the time, he said. Some pollen will escape. And then there is the issue of mutation, the fungibility of genes. Relying on a single gene to harness reproduction's gallop is a fraught notion, said Ohio State's Snow.
"If there's just one gene causing the tree to be sterile, what if there's a mutation in that gene and it doesn't work anymore?" she said.
There are a host of alternate sterility controls being considered, and it could be that several will need to be stacked together, Kausch said. Trees could be bred so their flowers do not bloom, or sterility could be passed through plastids, plant cell components with their own DNA. Some grass hybrids are by nature sterile, the plant equivalent of mules. And then there is always seed sterility -- the infamous "terminator" technology proposed, to wide public outrage, a decade ago for row crops.
USDA, though it commissioned the National Academy report, has given little indication on what future crops or traits could require sterility controls. (It did, however, put strict controls on a transgenic switchgrass trial.) It is a great guessing game whether the department will set a threshold for pollen escape, or mandate zero tolerance for any gene flow, Kausch said. The latter would be a tall order, but possible, given rapid advances in genomics and reproduction science, he said.
Surely, Kausch added, developing robust biological controls is better than the other option: relying on humans. This was the idea championed by Scotts for its bentgrass, which is still pending a deregulation decision from USDA, by far the agency's longest active case. The grass would be cut by golf courses before it flowered, the company argued, so there was no need to worry about pollen spread.
"Yeah, except when the golf course is abandoned," Kausch said. "In other words, leaving it up to human management. My experience with that practice [is] I've always been let down."
When it comes to evaluating risk, often the best place to start is human failure. The containment system devised by AquaBounty may be rigorous, but when up to 5 percent of the eggs it sells are fertile, and each batch includes up to 200,000 eggs, the odds push forward, inexorably, toward possible escape, whether through theft, loss or shipping error.
For several scientists, it has been frustrating that FDA's review of AquaBounty's salmon proposal has not required the company to tackle what risk its growth-enhanced salmon, if escaped, could present to wild relatives.
While FDA and most scientists maintain the fish is safe to eat, there is a wariness that the FDA study could set a dangerous precedent of substituting confinement layers for risk analysis.
"There is an assumption in the FDA report that the genes are going to be purged out if they eventually escape," Maryland's Zohar said. "Is this true? This is all assumption. I would like to see this done experimentally."
Each additional aquaculture farm seeking to grow AquaBounty's eggs will require a new FDA application, though the environmental assessments accompanying these applications may only copy AquaBounty's current application with facility-specific tweaks, according to Shannon Cameron, an FDA spokeswoman.
"It is very possible that the new [assessment] could be essentially the same as a previous [assessment], with only the addition of new information that is applicable to the new facility being added," she said.
Legally, the agency is not obliged to release these new analyses until after the farm is approved, a particular sore point for groups critical of the agency's decision to regulate the salmon under its animal drug rules. However, the agency has not yet determined if it will go beyond this legal requirement and release the reports prior to an approval, Cameron said.
FDA would be wise to address, in some deeper way, the public's environmental concerns, rather than relying on confinement, said Bill Muir, a population geneticist at Purdue University who has developed an essential model used in evaluating risks posed by modified fish.
"Shit always happens," Muir said. "If shit happens and they end up somehow in the ocean ... maybe it's hypothetical to the FDA, but people would like to know what happens."
There is an inherent contradiction in judging the viability of escaped salmon, of course, one carried over from invasive species studies. "If you want to know exactly how a transgenic salmon would behave in the open ocean, you would have to release it," said Francesc Piferrer, a fish reproduction expert at the Institute of Marine Sciences in Barcelona, Spain.
Clearly, with release not an option, scientists settle for semi-natural experiments, which mimic stream or ocean environments. And while FDA has not required such research, scientists in Virginia, using USDA funds, have conducted several studies on AquaBounty's salmon. The studies, which have not yet been published, looked at the salmon's survival against wild Atlantic salmon in food-abundant conditions well-suited to fast-growing salmon, according to Eric Hallerman, a fisheries professor at Virginia Tech, who led the research.
Just after hatching, the fish competed equally well for food. The bioengineered salmon -- which several scientists described as "little eating machines" -- also did poorly in starvation conditions. And AquaBounty's young male salmon, despite their growth advantage, had trouble winning over wild females, Hallerman said.
"[They] didn't compete well for mates at all," he said.
Thanks to limited funds, however, the group did not study the all-female salmon, sterile and not, that would be approved by the FDA. Hallerman would like to do that research, growing them out for several years, but in the meantime he would approve AquaBounty's initial proposal. Trying to get a rigorous handle on the full suite of environmental risks would "shut down this whole line of research," he said.
"I would say, let this go forward on very circumscribed conditions, and keep learning," Hallerman added.
Limits on lab testing
During FDA's salmon hearings last month, one commenter made the comparison between AquaBounty's proposed containment and the Deepwater Horizon, the doomed rig at the center of the Gulf of Mexico oil spill in the Gulf of Mexico. Both cite complex methods of containing a theoretical escape to avoid a full study of environmental risks. It is a fair comparison, Maryland's Zohar said.
"There is always a possibility of escapees, and the FDA admits it," he said.
Should more testing be done, it must simulate a greater variety of conditions, not just environments that are well-suited to the salmon, said Anne Kapuscinski, a sustainability science professor at Dartmouth College who has, quite literally, written the book on assessing environmental risk from transgenic fish.
Even after these tests, however, there is a deep scientific disagreement between Kapuscinski and Purdue's Muir on how these results can be extrapolated to the wild. Muir has spent the past decade refining a model, called net fitness, that judges foreign gene escape. It can predict whether an exotic gene -- dubbed a Trojan gene -- can spread through wild population, and whether its expression could then undermine the species' overall ability to survive, causing eventual extinction.
Using Hallerman's data, based on their prediction of optimal conditions, Muir found that AquaBounty's salmon and its gene went effectively extinct in a few generations, he said. If anything, the model overestimates risk, he added.
"This is a good result," Muir said. "It's always good to overestimate risk so that we have precaution. ... If these models predict that the transgene will be eliminated, we're very confident under natural conditions."
Kapuscinski and a team of Canadian researchers are not so certain in Muir's results. Such a system treats the Atlantic salmon as too uniform and does not account for genetic variation.
"I wish it was a simple as Bill is saying," she said. "But our work is saying it's not that simple."
Kapucinski's lab is testing Muir's model in aquarium fish with overactive growth hormones. Her research has found that the net-fitness model may not encompass all the genetic variability found in fish. More weight needs to be paid to environmental conditions, she said, and testing the salmon in best-case conditions for their survival may not be enough to justify extrapolation to the wild.
A forthcoming paper by two scientists in Canada echoes Kapuscinski's warnings. Written by Robert Ahrens and Robert Devlin -- whose work testing fast-growing coho salmon is widely influential -- the study, heavy in math, finds that transgenes can evolve with salmon over time, and in response to specific environments, that could allow the fish to survive in poor conditions.
"I would like to see a little more consideration [from agencies] for the potential responses in terms of [natural] selection," Ahrens said. "I understand the pressure from industry, but it would be wise to do a little bit more experimentation."
In the end, there will be a limit on what lab and field trials can predict, said Oregon State's Mallory-Smith, who tracked the escape of Scotts' bioengineered grass. "You're still working with such a limited population that you really don't know what's going to happen," she said.
AquaBounty's salmon may not escape the facilities established for them. ArborGen's eucalyptus are almost certain to spread, but they may not be invasive. But should either trait, or the many others making their way toward market, present a risk to the natural world, that trait will likely get out into the wild, Mallory-Smith said. Scotts' bentgrass is still growing in Oregon. It is not going away.
"If it's too risky to have gene flow," Mallory-Smith said, "it's probably too risky to do."