Making sense of the patchwork US regulatory system for genetically engineered crops and animals

rube goldberg
Rube Goldberg drawing.
The faster-growing genetically engineered AquaAdvantage Salmon took 20 years of regulatory scrutiny to gain approval, while the non-browning gene-silenced Arctic Apple took a relatively brisk 5 years. A new CRISPR-edited non-browning mushroom appears to have been cleared over a cup of coffee (okay, six months). All of these products were created using genetic engineering—so why were they regulated differently?

[Editor’s note: This is the first of a multi-part series on US biotech regulations and potential reform. In future installments, we’ll look at biotech animal breeding, New Breeding Techniques (NBTs), biotech pharmaceutical and industrial products, and international comparisons and regulatory reform.]

The first thing to understand about biotech regulation in the US is that it was not established through legislation, but rather through executive branch rulemaking under existing laws. In a series of cludge maneuvers, some more arbitrary than others as we’ll see, various categories of products and potential risks were assigned to be regulated under laws already on the books.

In doing the research for this series, I started with a decent sense of how biotech is regulated, but in trying to get a better, more granular view of the nuts and bolts of real-world processes, I’ve been surprised by how well the patchwork quilt covers what you would want it to cover and how much more sense the patchwork system makes than a unified, rationalized system. One criticism of the system is that it’s a Rube Goldberg contraption with too many moving parts. I’d say, to my pleasant surprise, that’s a feature, not a bug. We’ll circle back around to this point as we wrap up.


The most common critique of reformers of the current status quo is that regulation of products should be done by relative, credible risks rather than arbitrarily by process. I think that’s correct, but within that arbitrary divide it’s very much true that the biotech approval process is risk-based, with products that present little relative risks moving through the approval process more briskly and those that raise real questions being subjected to more scrutiny.

Reasonable people can disagree about whether the process is too onerous, insufficiently onerous or just right in the degree of onerousness. As the above comparison of the AquaAdvantage Salmon, the Arctic Apple, and the non-browning mushrooms shows, it is undeniably true that some products are subject to more scrutiny than others.

Coordinated Framework for the Regulation of Biotechnology

Established as a formal policy in 1986, the Coordinated Framework for Regulation of Biotechnology describes the Federal system for evaluating products developed using modern biotechnology. The three main federal agencies responsible for regulating the safe use of genetically engineered organisms are the Department of Agriculture (USDA), the Environmental Protection Agency (EPA), and the Food and Drug Administration (FDA).

The USDA’s duties are carried out by the Animal and Plant Health Inspection Service (APHIS), under the Plant Protection Act. The EPA’s come from the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) and the Toxic Substances Control Act (TSCA). All three agencies are responsible for compliance with the National Environmental Policy Act of 1969 (NEPA).


For plant foods, the FDA reviews genetically modified crops if the new trait’s properties are already understood. If not, the applicant is required to submit data proving that the food additive does not “adulterate” the food — in other words, that the additive is not injurious to human health.

Genetically modified animals are considered “animal drugs”. Thus, a New Animal Drug Application (NADA) must be submitted to the Center for Veterinary Medicine (CVM), which conducts an Environmental Assessment (EA) of potential environmental impacts and determines whether the animal presents any credible risk to human health.

Environmental Protection Agency

Under FIFRA, genetically modified pest control traits are regulated by the EPA as a pesticide. Traits related to pest control, known as Plant-Incorporated Protectants (PIPs) i.e. insect resistance (IR) and disease resistance require serious demonstrations of relatively low risk of serious environmental or health risks under FIFRA.

px Environmental Protection Agency logo svgThe best known PIP is the use of the Bt trait first commercialized in corn and cotton in the mid-1990s to control corn rootworm, cotton borers and initially boll weevils in cotton. The trait is named for the soil bacterium Bacillus thuringiensis, which produces Cry proteins toxic to those pests. Bt has been used for decades, mostly in organic farming as a pesticide spray, to control those pests. Several companies, including Monsanto, transferred a Bt gene into corn and cotton and other crops.


A Bt potato was also developed and approved for use to control the Colorado potato beetle. It was brought to market but quickly withdrawn due to the pressure from what are ostensibly environmental groups. Food and Water Watch and the Center for Food Safety petitioned McDonald’s, THE market maker in potato production, to commit to avoid Bt potatoes.

Rather than having to explain to a dubious public why lowering insecticide use in potatoes was worth doing and that there were no health concerns with the potatoes, the company bowed to pressure. Potatoes are a crop with a fairly high level of insecticide use. You can’t break up pest cycles easily with crop rotations in potato production. Why supposed environmental groups would oppose a technology that would lower insecticide use in potatoes continues to baffle actual environmentalists to this day.

Microbial pest control agents (MPCA), another form of biocides, are also regulated under FIFRA as pesticides, and by the FDA under the Federal Food, Drug, and Cosmetic Act (FFDCA).

NEPA provides an umbrella framework for assessing environmental risks, while TSCA applies to biotech microbes single-celled organisms bred for production of industrial enzymes and other chemicals, algaes engineered for biofuel production or microbes engineered for use in ocean oil spills. Another example might be RNAi used to restore vulnerability to herbicides in resistant weeds.


The first biotech product to be approved for food production was a yeast to produce chymosin, the enzyme in rennet that causes milk to coagulate into cheese. Microbes engineered for the production of synthetic vitamins like vitamin C for food fortification and nutritional supplements are another common form of biotech microbials regulated under TSCA.

Food and Drug Administration

The FDA primarily regulates biotech products under the Federal Food, Drug, and Cosmetic Act (FFDCA). In a 1992 guidance document, the FDA explained that genetic modification referred to ALL modifications to the genetics of a food plant varieties (meat and animals are regulated under the USDA).

“Genetic modification” means the alteration of the genotype of a plant using any technique, new or traditional. “Modification” is used in a broad context to mean the alteration in the composition of food that results from adding, deleting, or changing hereditary traits, irrespective of the method. Modifications may be minor, such as a single mutation that affects one gene, or major alterations of genetic material that affect many genes. Most, if not all, cultivated food crops have been genetically modified.

imagesWhat they are interested in is determining if the modification, genetically engineered or brought about by traditional breeding, combines properties that are well understood or if it introduces some new uncertainty. This means that if the new trait or traits produce proteins that are not known allergens and are produced in other food products that are well understood, then the new product is cleared under FFDCA with minimal fuss. If it comes with some unknowns, then they will ask for a greater range of evaluations and testing to answer questions about potential risks.


Crops are assessed along a number of variables:

The assessment scheme focuses on characteristics of the new plant variety, based on characteristics of the host and donor species, the nature of the genetic change, the identity and function of newly introduced substances, and unexpected or unintended effects that accompany the genetic change. The assessment focuses on the following considerations:

  1. Toxins known to be characteristic of the host and donor species;
  2. The potential that food allergens will be transferred from one food source to another;
  3. The concentration and bioavailability of important nutrients for which a food crop is ordinarily consumed;
  4. The safety and nutritional value of newly introduced proteins; and
  5. The identity, composition and nutritional value of modified carbohydrates, or fats and oils.

First, they are looking at things that are known about the properties of the host plant and the donor plant through their history of use and compositional analysis. Are both plants considered safe for consumption? Does the new plant have substantially different values for nutrients that the food is counted on providing?

Has the breeding process caused increased production of known toxins, naturally present in the host plant, but produced at sub-toxic levels? (i.e. Lectins and cyanogenic glycosides in legumes. Glucosinolates which may impair thyroid function in cruciferous vegetables. Cucurbiticin, an acute toxin, in squash and cucumber. Neurotoxic lathyrogens in chickpeas). Plants, like other organisms, can have metabolic pathways that are no longer active due to evolutionary changes. The breeding process can, in rare cases, reactivate silent pathways. In some small subset of those rare cases, the reactivated metabolic pathway could produce an allergen or toxin.


When the effects of a new crop can’t be confidently predicted by compositional analysis, the FDA will require toxicological feeding studies. Insect resistance conferred by the Bt trait was an example of a type of trait that required animal feeding studies.

Areas of concern with microbial pesticides are addressed under the Federal Food, Drug, and Cosmetic Act (FFDCA) are acute, subchronic and chronic dietary risks, occupational exposures, drinking water exposures, effects to the immune and endocrine systems, any dose response related information, exposures associated with daycares, residences and schools, exposure of sensitive populations, such as infants or children, aggregate effects for multiple exposures, and cumulative effects. For example, while the EPA would look at how a biological pest control will affect the environment, the FDA would look at potential harms from residues on crops to human health.

Department of Agriculture

The USDA’s involvement in regulating biotech in crop breeding is carried out by Animal and Plant Health Inspection Service (APHIS) under the Plant Protection Act (PPA). The PPA tasks APHIS with preventing the introduction of plant pests into the United States. The goal is to minimize the exposure of crops to disease and noxious weeds. Of the different ways that existing laws were interpreted in novel and extended ways in order to create oversight for biotech in breeding, this is the most kludgy. The trigger for regulatory oversight in this case is the use of plant pathogens in the introduction of novel genes into crops.

px APHIS svgAgrobacterium tumefaciens and cauliflower mosaic virus promoter (CaMV 35S) are two examples of plant pathogens used as vectors for gene transfer. Because they are harmful to certain plants in the wild, transferring their genetic material into crops for breeding purposes is what gives APHIS authority to regulate the new crop. Other vectors of gene transfer, biolistics (gene gun) for example, do not trigger USDA’s authority to regulate new crops.


Once the authority is established though, APHIS is concerned with a wide variety of potential risks. There are lots of ways that a crop can become a plant pest. APHIS will look at “disease and pest susceptibilities; the expression of gene products, new enzymes, or changes to plant metabolism; weediness and impact on sexually compatible plants; agricultural or cultivation practices; effects on non-target organisms; and the potential for gene transfer to other types of organisms.”

Related article:  Anti-GMO Russia joins global CRISPR 'bonanza,' investing $1.7 billion in gene-edited crop research

Potential weediness is a clear example of how a new problem can arise through breeding. While corn is a crop that has been bred for the full suite of pampering by modern agriculture it’s not going to thrive outside of cultivation a crop like canola does pretty well on its own. A new herbicide-tolerant canola that ends up somewhere that it isn’t wanted goes from crop to weed pretty easily:

Canola shatters badly during harvest. University of California Cooperative Extension farm advisors Doug Munier in Glenn County and Kent Brittan in Yolo County knew that when they became part of team testing RR canola as a possible oil crop for biofuel. Munier, Brittan and others found yields too low to make canola a profitable irrigated California crop at current prices. They gave up on canola, but canola did not give up on California. It is still around in many fields, three or four years after it was grown as a commercial crop.

Canola shattering losses can potentially be huge.

However, there is far more to it than that.

“What makes canola a different critter is that a significant percentage of this shattered seed does not germinate the following year, which is very different from other California field crops,” Munier explained. “When the shattered seed is incorporated into dry soil, it creates what is called secondary (seed) dormancy.” This is a common genetic trait for canola.

A canola crop that yields 1,400 pounds per acre would produce the equivalent of 182 million seeds per acre or 4,177 seeds per square foot, according to one Canadian researcher. Irrigated California canola has yielded 3,000 pounds per acre. A little shatter goes a long way. From a 5 to 10 pounds per acre crop seeding rate, Munier said, shatter at harvest can produce up to 10 times the initial seeding rate to fall on the soil to germinate for years to come.

In canola growing areas, the shatter issue has created the axiom, “Once a canola grower, always a canola grower,” because the shattered seeds remain viable and dormant in the soil for years.

Compounding the problem is the fact this secondary dormancy seed may germinate year-round in climates like California. That can be a nettlesome problem, but it is made far worse when it is herbicide resistant canola.

This became evident in a Sacramento Valley Roundup Ready alfalfa field recently.

“The grower sprayed his Roundup Ready alfalfa and got beautiful weed control except for what he thought was mustard. The herbicide did not touch it. His PCA looked closer and discovered it was Roundup Ready canola. Recognizing the problem, he rogued out the canola,” he said.

With canola’s secondary dormancy, had the grower let the canola go to seed, the problem would only have magnified for years.

In a trial planted by Munier in the Chico area, he planted RR canola in the fall of 2006. Harvested the next summer, the field was summer fallowed in 2007 followed by fall-planted wheat. RR canola volunteers from the 2007 harvest are still sprouting in 2010. “This is not just a weed, but one totally resistant to Roundup, a herbicide that is intensely used in a wide variety of high value crops in California,” Munier said.

The potential weediness of RR canola had to be balanced against the benefits of the crop, as well as the existence of management solutions. This is a pretty striking example, but here it’s a problem for a farmer using a RoundUp Ready weed control system, so it’s more or less kept in the family. For county roadside weed control or orchards, as the article lays out, there are reasonable management solutions. But problems can arise.

The issue here is that these kinds of problems are not confined to biotech crops. I’m reminded of the story of the cotton breeder who was growing organic blue cotton. This was driving all her neighbors crazy, because the pollen from her fields was blowing into their standard cotton fields. As far as they were concerned, her blue cotton wasn’t a differentiated, value-added product, it was a pathogen ruining their crops.


A few parting thoughts

One criticism of the system is that it’s a Rube Goldberg contraption with too many moving parts. I’d say, to my pleasant surprise in looking more closely at the system, that the Goldberginess is a feature, not a bug.  The weakness of the system, beyond the kludginess of how it was glommed on to existing statutes and institutions, is that it regulates first by the breeding process, and only then by credible, relative risk. We’ll talk more about regulating strictly by properties and risk rather than by process when we get to regulatory reform, but it’s worth noting that it would still be an ad hoc process. That is, even if you were to create a single inter-agency office as a point of entry to shepherd a new crop through the process, and regulate all novel crops by their properties and credible risks rather than the process they were created with, the process would still have a loose, improvisational aspect to it.

Each crop and new genetic trait or set of traits will be in different environments, where they present different potential risks. Not every new trait warrants a feeding study because the changes don’t impact the composition of the edible crop in any way that might credibly change the nutrition or toxicity. Nor does every new crop require a battery of environmental impact evaluations the way a new trait for insect resistance might. Nor do the same traits in new crops or similar traits in old crops require the same level of scrutiny now that their predecessors did two decades ago.

In putting together this piece, I asked Alan McHughen, a geneticist at UC Riverside who has been involved in shaping regulation in the US and Canada and has written extensively about it (he literally wrote the book on the subject), to take a look at an early draft of this piece and offer some feedback. He was generous enough to offer some important corrections (and copy edits! – like I said, generous). In a phone conversation in follow up he stressed to me that the process is iterative, it’s based on what has come before.

When I asked why something as simple and seemingly risk-free as the non-browning trait in the Arctic Apple took a full five years of review to be deregulated, he pointed out that it was the first apple, so there were no relevant comparators. Nor was there institutional knowledge on what to assess, what tires to kick when it comes to managing risk in apples. There is lots of experience with corn and soybeans, so regulators know what questions they want answers to. Likewise, a new Bt trait with a new Cry protein isn’t going to need the same kind of scrutiny that it did twenty years ago, because we know a lot about how Cry proteins work in the crops they’ve been used for the last twenty years.


He also stressed that the review under the FDA’s banner was voluntary. This is something anti-GMO activists seize on to question the integrity and rigor of the regulatory process. But it’s not an argument that holds much water. When Grist journalist Nathanael Johnson looked into the question in 2013, he concluded that it’s voluntary in name only because the FDA can pull a food off the market. McHughen had a different take. The FDA would only pull the product off the market if there was evidence that it wasn’t safe. What he pointed out was that, though it’s perfectly legal to bring a crop to market with only approvals from the EPA and the USDA and bypassing the FDA, nobody has done that. And the reason for this is that the company would get spit-roasted, either by anti-GMO groups or by their competitors for putting a product on the market without the FDA signing off on its safety.

In a sense, I can see the voluntary aspect actually giving the FDA license to be more rigorous, because there is that escape valve that the company can always just go to market if they don’t want to keep jumping over hurdles. If the company was required to keep jumping over hurdles, the pressure to show a little bureaucratic mercy (if there is such a thing) might take its toll. Likewise, complaints to legislators or top administration officials over being put through the wringer sounds a little whiny.

Either way, in digging into the nuts and bolts of how the system works, the process for crops is more rational than I expected and has improved over time. Where it really hasn’t found its footing yet is in regulating biotech animals, and that’s what we’ll cover next.

Marc Brazeau is the editor of Food and Farm Discussion Lab. Follow him on Twitter @eatcookwrite.

Outbreak Daily Digest
Biotech Facts & Fallacies
GLP Podcasts
Infographic: Here’s where GM crops are grown around the world today

Infographic: Here’s where GM crops are grown around the world today

Do you know where biotech crops are grown in the world? This updated ISAAA infographics show where biotech crops were ...
News on human & agricultural genetics and biotechnology delivered to your inbox.
glp menu logo outlined

Newsletter Subscription

* indicates required
Email Lists
Send this to a friend