GMOs: The Fear That Outran the Science

This is a history of a fear, and it needs to be read carefully, because the fear is really two separate worries fused together, and the two are not equally sound. One of them is a set of genuine, well-documented arguments about who owns the world’s seeds, what gets sprayed on the fields, and what happens to a living landscape when a handful of crops dominate it. The other is a claim that eating an approved genetically modified food is dangerous to the person eating it. The first has real evidence and real people making it in good faith. The second has been tested about as thoroughly as a food question can be, and the tests have not supported it. Most of the trouble in this story comes from the two travelling together as though they were the same worry. Pulling them apart is the whole task, and it has to be done without any contempt for the people who never separated them, because the reasons they didn’t are entirely understandable.

The strongest version of the sceptic’s case

Start with the argument at its best, made by its most serious advocates, because that is the version worth answering.

A farmer who buys Roundup Ready soya or Bt cotton is not simply buying seed. He is entering a contract. For most of agricultural history a farmer saved seed from this year’s crop to sow next year’s; that was the deal between a person and a plant, older than money. The patented seed changed it. Under the technology-use agreements that companies such as Monsanto required, saving and replanting the patented seed was forbidden, and the company enforced that prohibition in court. The most-cited case, Monsanto Canada Inc. v. Schmeiser, went to the Supreme Court of Canada in 2004; the company also pursued a long series of suits against US growers it accused of saving seed. Whatever one thinks of the legal merits, the picture that reached the public was vivid and, in outline, accurate: a corporation suing farmers over the contents of their own fields. That is not a paranoid fantasy. It happened, in courtrooms, with judgments you can read.

Then there is the chemistry. The commercial point of a herbicide-tolerant crop is that you can spray the whole field with glyphosate — Roundup — and kill everything except the crop. This is convenient, and it also means more glyphosate gets used, year after year, on more land. In March 2015 the International Agency for Research on Cancer, the arm of the World Health Organization that classifies carcinogens, placed glyphosate in Group 2A, “probably carcinogenic to humans”. That is an official finding from a serious body, and anyone who waves it away is not being honest about the record. We will come back to what Group 2A actually means, because it is routinely misread, but the classification is real and it is not nothing.

Add the ecology. A landscape planted edge to edge with one genetically uniform crop is a monoculture, and monocultures are fragile — a single well-matched pest or blight can move through them without a firebreak. Heavy, uniform herbicide use also strips out the weeds that insects and birds depend on; the collapse of milkweed along American farmland, and with it the food plant of the monarch butterfly, is a documented ecological worry with the herbicide regime standing squarely in the frame. And there were the so-called “terminator” patents — genetic-use-restriction technologies designed to make a crop’s seed sterile, so a farmer could not save it even if he wanted to. The companies never commercialised terminator seed, and said they would not, but the patents existed, and their very existence told people something chilling about the direction of travel.

Finally, the politics of trust. The agencies that approve these crops draw on data supplied by the companies that sell them; the revolving door between industry and regulator is real; and for a long time consumers in much of the world could not even tell whether a product contained GM ingredients, because there was no requirement to label it. A person who simply wanted to know what was in their food, and to choose, was told that the choice was unnecessary. That is a reasonable thing to resent.

Assembled honestly, this is a strong case. Corporate control of the seed supply, an ever-heavier herbicide load, biodiversity loss, and a regulatory system that asked to be trusted while showing its workings only selectively — none of that is invented, and a reader who feels the pull of it is not being foolish. The distrust of agribusiness that runs underneath the whole GMO panic was earned, in the ordinary way that distrust is earned, by conduct.

The kernel: what the technology actually was

Now the history, told straight, because the fear grew up alongside a real sequence of events.

The first genetically modified whole food approved for sale was a tomato. Calgene’s Flavr Savr, engineered to soften more slowly so it could ripen longer on the vine, cleared US regulators in 1994. It was a commercial disappointment and vanished within a few years, but it opened the door. What came through the door next was bigger. In the mid-1990s the two workhorse traits arrived: Bt crops, engineered to carry a gene from the soil bacterium Bacillus thuringiensis so the plant produced its own insecticidal protein and needed less spraying; and herbicide-tolerant “Roundup Ready” crops, engineered to survive glyphosate so a field could be weeded chemically in one pass. Maize, soya, cotton and oilseed rape were the early carriers. Within a decade these traits covered a large fraction of the acreage of several major crops across the Americas.

The safety question about eating them is narrow and answerable: does a person who eats an approved GM food face greater risk than a person who eats the conventional equivalent? Note what the question is not. It is not “is industrial agriculture good”, or “should one company control the seed”, or “is glyphosate sprayed too freely”. It is a bounded, testable question about the food on the plate. And on that bounded question the scientific bodies converged with unusual firmness. In 2012 the board of the American Association for the Advancement of Science stated that the crops on the market had been shown to be no riskier to eat than conventional foods. The World Health Organization has said much the same. The most exhaustive review came in 2016, when the US National Academies of Sciences, Engineering, and Medicine published a report running to hundreds of pages, having examined some nine hundred studies; it found no substantiated evidence that foods from GM crops were less safe to eat than their conventional counterparts, while being candid about the social and economic questions it could not settle with biology.

The fork: where “unsafe to eat” left the record

The precise point where the food-safety fear parted company with the evidence can be dated, because it hangs on two studies that became famous for the wrong reasons.

The first was the Árpád Pusztai affair of 1998. Pusztai was a respected researcher at the Rowett Institute in Scotland who, in a television interview, said his rats had suffered damage after eating potatoes engineered with a snowdrop lectin, and that he would not eat such food himself. The remark detonated. The Rowett suspended him, the Royal Society reviewed the work and found it did not support the conclusions drawn from it, and the episode became, depending on who was telling it, either the moment a brave scientist was silenced by the establishment or the moment a preliminary and flawed experiment was blown far past what it could bear. Both readings kept the story alive. What the underpowered potato study could not do was establish that GM food was dangerous to eat, and that was the claim being hung on it.

The second, and more consequential, was the work of Gilles-Éric Séralini. In 2012 his team published a paper in Food and Chemical Toxicology reporting that rats fed Roundup Ready maize, or given glyphosate in their water, developed grotesque tumours; the paper came with photographs of rats with growths the size of ping-pong balls, and those images went everywhere. The trouble was in the design. The study used Sprague-Dawley rats, a strain that develops tumours spontaneously at high rates as it ages, in groups too small to separate a real effect from ordinary background noise. After sustained criticism the journal retracted the paper in 2013, citing the inconclusiveness of the results given the strain and sample size. The paper was later republished in another journal, and that republication was waved as proof the science had been vindicated. It had not; a paper reappearing in a different venue does not answer the objections to its method. But the photographs had already done their work. You do not un-see a tumorous rat, and no calm sentence about statistical power competes with the picture.

That is the fork. Two studies, neither able to bear the weight placed on it, both retracted-or-repudiated by the scientific process, both far more powerful as images and headlines than as evidence. The claim “GM food gives you cancer” was born less from data than from a photograph and a suspended scientist, and it detached from both and travelled on its own.

Hazard, risk, and a butterfly

The glyphosate classification deserves its own careful paragraph, because it is where an honest person can be most easily misled, and where honesty on the science side matters most.

The IARC works in hazard categories, not risk. A hazard is anything capable of causing harm under some conditions; risk is the probability of harm under the conditions people actually meet. IARC’s Group 2A — “probably carcinogenic” — is the same category that has, at various times, held red meat, very hot beverages and shift work that disrupts the body clock. The classification asks whether a thing can cause cancer at all, at some dose, not whether it does so at the exposures a shopper encounters. This is why other serious agencies reading much of the same literature reached different conclusions about real-world risk: the European Food Safety Authority in 2015, and the US Environmental Protection Agency, judged that glyphosate was unlikely to pose a carcinogenic risk to people at expected exposures. Reasonable scientists disagreed, partly because they were answering slightly different questions and weighting different studies. Holding both facts at once — a genuine hazard flag from IARC, and a real dispute about actual risk — is the honest position. Collapsing it into “Roundup causes cancer, official” is not.

The monarch butterfly shows the same shape of trouble from the other direction. In 1999 the Nature journal carried a short paper by John Losey and colleagues at Cornell reporting that monarch caterpillars died when fed milkweed leaves dusted with pollen from Bt maize in the laboratory. It was a striking result and it became the emblem of GM harm to wildlife. But it was a lab study, and the field follow-ups that came over the next few years found that under real conditions the exposure of monarchs to Bt pollen was generally low enough that this direct effect was small. The deeper threat to the monarch turned out to be the ecological argument from earlier — herbicide-driven loss of the milkweed the butterfly needs — which is a genuine problem, and one belonging to the herbicide-and-monoculture column, not to the “the engineered protein poisons butterflies” column. The lab image survived; the field correction, as usual, did not travel with it.

What the fear was really about

Golden Rice is where the cost of the confusion becomes hardest to look at. Engineered to carry beta-carotene, a precursor of vitamin A, it was developed as a humanitarian project aimed at childhood vitamin A deficiency, a leading cause of preventable blindness and death among children in poorer countries. It was never Monsanto’s profit centre; it was, in intent, a gift. Greenpeace and other groups campaigned against it for years as the thin end of the GM wedge, and its progress through trials and regulation was slow for many reasons, some scientific and some political. Whatever share of the delay belongs to the opposition, the episode showed the trap closing: a genuine, well-earned distrust of agribusiness had generalised into resistance to a crop whose whole purpose was to help the poorest children, because it wore the same word, “GMO”, as the corporate products people rightly mistrusted.

That is the thing the whole history turns on. The distrust of agribusiness was real and largely justified, in the same way that distrust of drug companies and food companies has been justified when they have earned it — the pattern traced in Big Sugar’s decades of manufacturing doubt, or in the medical betrayals that made every later reassurance harder to believe, which is the machinery underneath the MMR-and-autism fear. People had watched industries lie about safety before. So when a new technology arrived, patented and corporate and sprayed across the fields of the world, the instinct to distrust it was not irrational. The error was one of aim. The distrust was pointed at the molecule on the fork, where the evidence would not sustain it, instead of at the contract, the herbicide load and the concentration of seed ownership, where the evidence was strong all along. A very similar swap happened with genetically modified salmon, where anxieties about industrial fish farming and ecological escape — real issues — got expressed as a fear of the fish itself as food.

So the honest close is not a verdict badge, because the story does not resolve into one. The sceptics were right about a great deal: seed patents did put farmers in an uncomfortable new relationship with the companies; glyphosate use did climb and is genuinely contested; monoculture and biodiversity loss are real costs still being paid. The scientists were right too, on the narrow thing they were actually asked: the approved foods, tested exhaustively, have not shown themselves more dangerous to eat. The fear outran the science because it was carrying a legitimate cargo of grievance and could not tell which part of that cargo the evidence would hold up. Understanding the panic means seeing that the people who felt it were mostly right to be uneasy — they had simply attached a sound suspicion to the one claim it could not support, and a photograph of a tumorous rat is very much easier to remember than a paragraph about the difference between hazard and risk.