When you bite into a piece of celery, there’s a fair chance that it will be coated with a thin film of a toxic pesticide called acephate.

The bug killer — also used on tomatoes, cranberries, Brussels sprouts and other fruits and vegetables — belongs to a class of compounds linked to autism, hyperactivity and reduced scores on intelligence tests in children.

But rather than banning the pesticide, as the European Union did more than 20 years ago, the U.S. Environmental Protection Agency recently proposed easing restrictions on acephate.

The federal agency’s assessment lays out a plan that would allow 10 times more acephate on food than is acceptable under the current limits. The proposal was based in large part on the results of a new battery of tests that are performed on disembodied cells rather than whole lab animals. After exposing groups of cells to the pesticide, the agency found “little to no evidence” that acephate and a chemical created when it breaks down in the body harm the developing brain, according to an August 2023 EPA document.

The EPA is moving ahead with the proposal despite multiple studies linking acephate to developmental problems in children and lab rats, and despite warnings from several scientific groups against using the new tests on cells to relax regulations, interviews and records reviewed by ProPublica show.

To create the new tests designed to measure the impact of chemicals on the growing brain, the EPA worked with the Organization for Economic Cooperation and Development, which comprises some of the world’s wealthiest democratic countries and conducts research on economic, social and scientific issues. The OECD has warned against using the tests to conclude a chemical does not interfere with the brain’s development.

A scientific advisory panel the EPA consulted found that, because of major limitations, the tests “may not be representative of many processes and mechanisms that could” harm the developing nervous system. California pesticide regulators have argued that the new tests are not yet reliable enough to discount results of the older animal tests. And the Children’s Health Protection Advisory Committee, a second group of advisers handpicked by the EPA, also warned against using results of the nonanimal tests to dismiss concerns.

“It’s exactly what we recommended against,” Veena Singla, a member of the children’s committee who also teaches at Columbia University, said of the EPA’s acephate proposal. “Children’s development is exquisitely sensitive to toxicants. … It’s disappointing they’re not following the science.”

The EPA’s proposal, which could be finalized later this year, marks one of the first times the agency has recommended changing its legal safety threshold largely based on nonanimal tests designed to measure a chemical’s impact on the developing brain. And in March, the EPA released a draft assessment of another pesticide in the same class, malathion, that also proposes loosening restrictions based on similar tests.

The proposed relaxing of restrictions on both chemicals comes even as the Biden administration has been strengthening limits on several other environmental contaminants, including some closely related pesticides.

In response to questions from ProPublica, the EPA acknowledged that it “will need to continually build scientific confidence” in these new methods but said that the introduction of the nonanimal tests to predict the danger chemicals pose to the developing brain “has not been done in haste. Rather, a methodical, step-wise approach has been implemented over the course of more than a decade.”

The agency said its recent review of acephate included a thorough examination of a variety of scientific studies and that, even with its proposed changes, children and infants would still be protected.

The EPA expects to start accepting public comments on the acephate proposal in the coming months before it makes a final decision. The agency anticipates soliciting comments on malathion this summer.

Some environmental scientists strongly oppose loosening the restrictions on both acephate and malathion, arguing that the new tests are not reliable enough to capture all the hazards a chemical poses to the developing brain.

“It will put children at an increased risk of neurodevelopmental disorders like autism and ADHD that we already know are linked to this class of chemicals,” said Rashmi Joglekar, a toxicologist at the Program on Reproductive Health and the Environment at the University of California, San Francisco.

Health and environmental scientists are concerned about more than the direct impact of having potentially greater amounts of acephate and malathion on celery and other produce. They also worry that using the new tests as a basis for allowing more pesticides on crops will set a dangerous precedent for other brain-harming chemicals.

“I think the companies see this as a new way over a 10- or 20-year period to gradually lobby” the EPA “to allow higher levels of pesticides in food,” said Charles Benbrook, an agricultural economist who has monitored pesticide regulation for decades. “If they can convince regulators to not pay attention to animal studies, they have a very good chance of raising the allowable exposure levels.”

Industry helped fashion EPA’s testing strategy

Since its founding in 1970, the EPA has relied on studies of mice, rats, guinea pigs and other species to set exposure limits for chemicals. The lab animals serve as a proxy for humans. Scientists expose them to different doses of substances and watch to see what levels cause cancer, reproductive problems, irritation to the skin and eyes, or other conditions. Some tests look specifically at chemicals’ effects on the offspring of rats exposed during pregnancy, and some of those tests focus on the development of their brains and nervous systems.

But over the past decade, chemical manufacturers and animal rights advocates have argued for phasing out the tests on the grounds they are impractical and inhumane. The animal experiments are also expensive, and the pesticide industry, which by law shoulders the cost of testing its products, is among the biggest proponents of the change.

The EPA has allowed the chemical industry and animal rights groups to help fashion its testing strategy. Agency officials have co-authored articles and held workshops on the use of the cell-based tests to regulate chemicals alongside representatives of People for the Ethical Treatment of Animals as well as Corteva Agriscience, BASF and Syngenta Crop Protection, companies that make pesticides regulated by the EPA.

The EPA said its scientists have been working to develop the nonanimal tests for decades with other government and scientific organizations, both nationally and internationally.

“It is absurd to describe those scientific efforts as an apparent conflict of interest,” the agency said in a statement.

The EPA’s Office of Pesticide Programs has previously come under fire for its willingness to allow pesticides onto the market without required toxicity testing. In 2018, as The Intercept reported, staff members held a party to celebrate a milestone: The number of legally required tests the office had waived for pesticide companies had reached 1,000. A science adviser to the office at the time said the move spared companies more than $6 million in expenses.

While phasing out animal experiments would save money and animal lives, experiments involving collections of cells do not always accurately predict how entire organisms will respond to exposure to a toxic chemical. The new cell-based tests and computer techniques that are sometimes used with them can be reliable predictors of straightforward effects like eye or skin irritation. But they are not yet up to the task of modeling the complex, real-world learning disorders that have been linked to acephate and malathion, according to Jennifer Sass, a senior scientist at the Natural Resources Defense Council, an environmental advocacy organization.

The new tests can show whether a chemical can kill a brain cell. And they can show if a chemical affects how a brain cell connects with other brain cells, said Sass.

“But these tests can’t show that a kid is going to be able to sit through class and not go to the principal’s office,” she said.

While the cell-based tests may point to certain harms, they are likely to miss others, said Sass, who likens their use to fishing with a loose net. “You only know what you caught — the big stuff,” she said. “You don’t know about all the little stuff that got through.”

A 2023 study revealed the failure of the cell-based tests to detect certain problems. In it, scientists exposed brain cells to 28 chemicals known to interfere with the development of the nervous system. Although the tests were specifically designed to assess whether chemicals harm growing brains, they failed to clearly identify harm in one-third of the substances known to cause these very problems. Instead of registering as harmful, the test results on these established developmental neurotoxins were either borderline or negative.

Because of these potential blind spots and other uncertainties associated with the tests, the Organization for Economic Cooperation and Development has advised against interpreting results of the nonanimal tests as evidence that a chemical doesn’t damage the brain. Several scientific groups have recommended that the EPA do the same.

A federal advisory panel of scientists assembled to advise the EPA on pesticide-related issues published a 2020 report that identified numerous limitations and gaps in the nonanimal studies, finding that they “underestimated the complexity of nervous system development.”

In 2021, the Children’s Health Protection Advisory Committee, a group the EPA created to provide advice on how to best protect children from environmental threats, warned the agency that, “due to important limitations,” the test results “cannot be used to rule-out a specific hazard.”

In comments to the EPA, California’s Department of Pesticide Regulation also cautioned the agency against using the tests to conclude that a chemical doesn’t cause specific harms. The California regulators emphasized that the traditional battery of animal tests was still necessary to understand complex outcomes like the effects on children’s developing brains.

“To abandon it at this time would be to abandon a critical support for health-protective decisions,” they wrote.

EPA accused of double standard

As much as 12 million pounds of acephate were used on soybeans, Brussels sprouts and other crops in 2019, according to the most recent estimates from the U.S. Geological Survey. The federal agency estimates that up to 30% of celery, 35% of lettuce and 20% of cauliflower and peppers were grown with acephate. Malathion is used on crops such as strawberries, blueberries and asparagus. (The U.S. Department of Agriculture prohibits the use of most synthetic pesticides, including acephate and malathion, to grow and process products certified by the agency as organic.)

Acephate and malathion belong to a class of chemicals called organophosphates, which U.S. farmers have used for decades because they efficiently kill aphids, fire ants and other pests. But what makes the pesticides good bug killers — their ability to interfere with signals sent between nerve cells — also makes them dangerous to people. For years, there has been a scientific consensus that children are particularly vulnerable to the harms of pesticides, a recognition that led the EPA to strengthen restrictions on them. But with both acephate and malathion, the agency is now proposing to remove that extra layer of protection.

The EPA effectively banned another organophosphate pesticide, chlorpyrifos, in 2021, based in part on evidence linking it to ADHD, autism and reduced IQ in children. (In response to a lawsuit brought by a company that sells the pesticide and several agricultural groups, a court vacated the ban in December, allowing the resumed use of chlorpyrifos on certain crops, including cherries, strawberries and wheat.) While some health and farmworker groups are petitioning the EPA to ban all organophosphate pesticides, the agency is arguing that it can adequately protect children by limiting the amount farmers can use.

Several studies suggest that, even at currently allowable levels, acephate may already be causing learning disabilities in children exposed to it while in the uterus or in their first years of life. In 2017, a team of University of California, Berkeley researchers, partly funded by the EPA, found that children of Californians who, while pregnant, lived within 1 kilometer of where the pesticide was applied had lower IQ scores and worse verbal comprehension on average than children of people who lived further away. Two years later, a group of UCLA scientists reported that mothers who lived near areas where acephate was used during their pregnancies had children who were at an increased risk of autism with an intellectual disability.

The EPA considered this research when deciding to relax the limits on acephate use but stated that flaws and inconsistencies made these epidemiological studies “not compelling.” The agency also dismissed a rat study submitted to the EPA in 2005 in which the pups of mother rats exposed to higher levels of acephate were, on average, less likely to move than the pups of mothers exposed to lower levels. The EPA told ProPublica that “no conclusions could be drawn” from the experiment, citing the “high variability of the data” it produced. But some scientists outside the agency find that study a particularly worrisome indication of the pesticide’s potential to harm children.

In its proposals to increase the allowable amount of both acephate and malathion on food, the EPA also had to look past other potentially concerning test results. Some of the cell-based tests of acephate showed borderline results for interference with brain functions, while some of the tests of malathion clearly indicated specific problems, including interference with the connections between nerve cells and the growth of certain parts of nerve cells. Several scientists interviewed by ProPublica said that such results demand further investigation.

Some scientists see a double standard in the agency accepting the imperfect nonanimal tests while citing flaws in other research as reasons to dismiss it.

“They’re acknowledging limitations in epidemiology while at the same time not acknowledging the even greater limitations of using a clump of cells in a petri dish to try to model what’s happening in a really complex organism,” said Nathan Donley, a scientist at the Center for Biological Diversity, an environmental advocacy organization.

Asked about the criticism, an EPA spokesperson wrote in an email to ProPublica that the agency “does not believe there was a double standard applied.”

ProPublica is a Pulitzer Prize-winning investigative newsroom. Sign up for The Big Story newsletter to receive stories like this one in your inbox.