The Fish That Frightened Everyone: GM Salmon

In the late 1980s, in a laboratory at Memorial University of Newfoundland, a biologist named Garth Fletcher was studying how fish that live in near-freezing water avoid turning to ice. The answer is a set of antifreeze proteins, switched on by the cold, that stop ice crystals forming in the blood. Fletcher and his colleagues were interested in the genetic switch — the promoter — that kept those proteins running through a Canadian winter. Then they had an idea that would take a quarter of a century to reach a dinner plate: attach that always-on switch to a salmon’s growth-hormone gene, and you would have a fish that never stopped growing for the season. The animal they made reached market weight in roughly half the usual time. It also became the most feared fish in the world, and the argument over whether anyone should be allowed to eat it lasted longer than the careers of most of the scientists who built it.

The animal they actually built

Strip away the word “Frankenfish” and the engineering is almost modest. The base animal is an ordinary Atlantic salmon, Salmo salar, the same species farmed in net-pens off Norway, Scotland and Chile. Into its genome the researchers inserted a single growth-hormone gene taken from a Pacific relative, the Chinook salmon (Oncorhynchus tshawytscha), and paired it with the promoter from that antifreeze gene, borrowed from an ugly, eel-like bottom-dweller called the ocean pout (Zoarces americanus).

The trick is in the switch. A wild Atlantic salmon makes growth hormone too, but seasonally — production falls away in cold, dark months, which is why farmed salmon take the better part of three years to reach harvest size. The ocean-pout promoter ignores the calendar. It keeps the transplanted growth-hormone gene switched on the year round, so the fish grows steadily through the winter instead of pausing. The result, marketed as AquAdvantage salmon by a company eventually called AquaBounty Technologies, reaches a market weight of four to five kilograms in about sixteen to eighteen months rather than the usual thirty or so.

This matters because it is a genuine efficiency, and because it is not the thing people picture. The fish is not enormous. A grown AquAdvantage salmon is the same final size as a conventional one; it simply gets there faster and on roughly a quarter less feed. It is a salmon carrying a salmon gene and a fish promoter, engineered to skip the seasonal off-switch. The chimera stitched together from a dozen creatures lives only in the headlines. Holding that plainly in mind is the beginning of understanding both why the science community found the panic frustrating and why the panic was, in several of its parts, entirely reasonable.

The case against it, at full strength

It is easy to caricature the opposition to genetically modified food as squeamishness dressed up as ethics. That caricature is lazy, and it loses the argument by refusing to have it. The serious case against approving AquAdvantage salmon was made by ecologists, consumer advocates and a good many scientists, and at its strongest it ran along four lines that deserve to be stated as their best advocates would state them.

The first and heaviest is ecological. Atlantic salmon in the wild are in trouble — populations across the North Atlantic have collapsed over the past half-century, and several runs are formally endangered. Farmed salmon escape from ocean net-pens constantly, in their hundreds of thousands, and interbreed with wild fish, eroding the local adaptations that let a river’s own salmon survive. Now imagine releasing into that fragile system a fish engineered to grow fast and, plausibly, to feed and mate aggressively. In the late 1990s two Purdue University researchers, William Muir and Richard Howard, modelled exactly this danger and gave it a name that stuck: the Trojan gene hypothesis. If a transgene made a fish more attractive as a mate but its offspring less likely to survive to breed, the gene could spread through a wild population while quietly driving that population towards collapse — a genetic Trojan horse, welcomed in and lethal once inside. The maths was a model, describing a mechanism no one had yet watched play out in a wild river, and it was frightening precisely because it was rigorous.

The second line was about consent, and it produced the word that defined the whole affair. This would be the first genetically engineered animal ever approved anywhere for people to eat — a true precedent, the door through which every future engineered animal would walk. Consumer groups argued that whatever the safety verdict, people had a plain right to know what they were buying, and to choose. The demand for a label was a demand to be told. When opponents branded the fish a “Frankenfish”, the tabloid word did real rhetorical work, but underneath it sat a defensible principle about transparency and the right to refuse a novelty you never asked for.

The third line was about power. AquAdvantage was a patented animal, owned by a company, protected by intellectual property law. Critics pointed to the broader pattern in modern agriculture — a handful of firms controlling the genetics of the food supply, farmers bound to licences, the living world converted into proprietary product. Whatever one thinks of this fish, the model it advanced was one of privatised biology, and that is a legitimate thing to fear on its own terms, quite apart from any question of safety.

The fourth line was philosophical and the hardest to answer: the precautionary principle. We were being asked to release a genuinely novel organism into the food system and, potentially, the environment, under real scientific uncertainty, on the strength of models and contained trials. The burden, opponents argued, should fall on the people introducing the novelty to prove it harmless, and “we can think of no reason it would hurt you” is not the same as proof. When the stakes include an endangered wild species and a first-of-its-kind precedent, caution is not cowardice. It is arithmetic about irreversibility.

Take those four together and you have something no honest person should wave away. The opposition was not a mob. It was, at its best, a careful argument about ecology, consent, corporate power and the ethics of acting under uncertainty.

Where the strongest version starts to give

Now walk the same ground with the evidence in hand, because this is where the good-faith case begins to separate from the folklore that grew up beside it.

The ecological argument is the one that held up longest and had to be answered directly, so the company and the regulator answered it with engineering they could point to. The commercial fish are engineered around two safeguards. Every one is female, and every one is triploid — made to carry three sets of chromosomes instead of the normal two, a condition induced by a pressure shock applied to the newly fertilised egg. Triploid animals are functionally sterile; the extra chromosome set jams the machinery of making eggs. An all-female, sterile population cannot found a wild lineage even if it escapes.

Honesty requires the next sentence, and the FDA supplied it: triploidy induction is not perfect. In any batch a small percentage of fish may slip through as fertile diploids. So the second safeguard is physical. Unlike ordinary farmed salmon, which grow in cages in the open sea, AquAdvantage fish were raised in land-based tanks behind multiple barriers. The eggs were produced at a facility on Prince Edward Island in Canada; the fish were grown out first in a tank complex in the highlands of Panama, far from any salmon habitat, and later in a converted plant in Albany, Indiana, an American state with no coastline on any ocean. Screens, filters and the simple fact of being a long way inland stood between the fish and any wild river. The FDA’s 2015 assessment leaned on this stacked containment: sterility that is very good but imperfect, backed by physical isolation that closes the remaining gap.

The safety-to-eat question turned out to be the least dramatic part. After reviewing the company’s data over many years, the United States Food and Drug Administration concluded in its November 2015 approval that the fish was as safe to eat as conventional Atlantic salmon and nutritionally equivalent to it. The growth hormone involved is a salmon protein that salmon already make; the finished fish sits within the normal range for its species on the measures that matter. Health Canada reached the same conclusion, approving the fish for sale in 2016. On the specific fear that eating this salmon might poison or transform you, the record simply does not cooperate. It is a salmon, and it eats like one.

The regulator’s strange choice

If the science settled quietly, the paperwork did not, and here one of the opposition’s instincts was vindicated in a way worth dwelling on. The FDA had to decide what kind of thing this fish was, legally. It chose to regulate the inserted gene construct as a “new animal drug” — the same statutory category used for a veterinary medicine you might give a sick cow. The reasoning was that the engineered DNA is intended to affect the structure and function of the animal’s body, which is the legal definition of a drug. The consequence was that the review ran under a framework built for pharmaceuticals, much of it confidential, rather than one built for food or for environmental release.

Critics found this genuinely troubling, and they had a point that had nothing to do with whether the fish was safe. A framework designed to keep a drug company’s trial data commercially confidential is a poor fit for a decision the whole public has a stake in. It kept much of the process opaque during the years it mattered most, and it stretched a definition until a fish became a drug. You can accept every one of the FDA’s safety findings and still think the pathway it travelled to reach them was the wrong one. That is exactly the sort of distinction this desk exists to draw.

The delay that followed proved the point. Approved in 2015, the fish still could not be sold in the United States, because Congress intervened and blocked imports until the FDA issued guidance on how it should be labelled — the consumer-consent argument, winning in the legislature what it could not win in the laboratory. American sales did not begin until around 2021. By then the economics had soured, and in 2023 and 2024 AquaBounty wound down its American farming operation, the Indiana plant it had once held up as the future. A quarter of a century from the first submission to the regulator to a retreat from the market: the fish spent longer in limbo than most farmed animals spend alive by orders of magnitude.

What was fear, and what was folklore

Separating the two is the whole task, and it can be done cleanly if you are willing to be fair to both sides at once.

Several of the fears were reasonable, and some were right. Containment was a real risk that demanded a real answer, and the honest answer was sterility backed by physical isolation, with the data laid out to show for it. Labelling transparency was a legitimate principle, and the public won it. The wisdom of shoehorning a food animal through a veterinary-drug statute is fairly criticised to this day. And the worry about a food supply owned by a handful of patent-holders is a live political question that this fish did nothing to settle. A worried shopper who held any of these concerns was thinking clearly.

Other fears curdled into something else. The idea that the fish was a monster — swollen, mutated, alien — described an animal that was never built; the real thing is a normal-sized salmon carrying a salmon gene. The notion that eating it might harm you, or that it was nutritionally strange, ran directly into two national regulators’ findings of equivalence. The “Frankenfish” that stalked the headlines was a folk creature assembled from a real precedent and a very old story about hubris and things made in laboratories. That folk creature is a cousin to the panics that attach to ordinary food all the time — the conviction that the pink of farmed salmon is a sinister dye rather than a managed pigment, or the recurring fear that a bright additive in a familiar boxed dinner is quietly harming children. In each case a genuine, answerable question — about welfare, or transparency, or dose — gets crowded out by a mythic one about contamination and betrayal.

What the AquAdvantage story really records is how badly our institutions are built for arguing about novelty. The public was handed a binary — miracle or monster — when the truth was a list: a clever, contained, safe-to-eat fish, approved through a strained legal pathway, sold under a hard-won label, defeated in the end by economics rather than biology, and shadowed the whole way by an ecological risk that was managed but never quite reducible to zero. Every worried person in that story was reacting to something real. The fish that frightened everyone was mostly a salmon that grew up fast, and the fear it provoked tells us less about the animal than about how little practice we have had, and how little help we were given, in reasoning together about something genuinely new.