Pesticide testing is flawed—and it’s harming our birds and bees

Computer modeling of both chemicals and the organisms affected by them can help farmers make better choices.

tractor spraying insecticide over field
A tractor sprays insecticide over a fieldAqua Mechanical

Since the DDT days of the 1950s and 60s—when the pesticide caused alarming declines in bird species, including iconic birds of prey—regulators and chemical companies have adopted far more rigorous testing guidelines. But the biodiversity crisis still looms, and we’ve continued to see drops in avian and insect species, among other losses. While habitat loss is likely the main cause, herbicides and insecticides used in agriculture are also thought to contribute to some of the declines.

We’ve tried to regulate which chemicals get into the environment, but they’re clearly still causing harm. And according to a paper published Thursday in Science, the reason could be fundamental flaws in the testing process. “We’re seeing a decline in biodiversity, which we can’t directly link to pesticides,” says Christopher John Topping, an ecotoxicologist at Aarhus University in Denmark who lead the study. “But pesticides are heavily regulated, so as part of that we have to be sure that we're asking the right questions, when we're using public money to regulate these pesticides.”

According to Topping and his coauthors, the tools we use now to assess the safety of pesticides are inadequate. The current environmental risk assessment used by the European Union, United States, and other countries doesn’t provide a complete picture of how pesticides are actually applied on fields, preventing us from understanding the risk. The researchers argue that computer modelling, which has advanced substantially in recent years, offers a more holistic approach. Such an approach would offer better information to farmers and regulators, though it would be hard to implement anywhere there isn’t a ton of data on pesticide use, including the US.

Topping thinks the current tiered method of pesticide testing is flawed. To determine toxicity, the first tier that a chemical often undergoes first is an animal-based lab test that determines acute toxicity; Basically, you find out what amount will kill a rat. Then you calculate how much of the pesticide a species you’re interested in protecting will ingest if it only eats contaminated food—insects or plants that were sprayed with the pesticide. When the amount an organism could hypothetically ingest in nature is greater than the lethal amount in the lab, the pesticide fails the first test and moves up to the next tier of testing (if it passes, it’s considered safe to use). And many pesticides fail the first round, says Topping.

In the next tier, researchers estimate how much the non-pest insect, bird, or mammal might be exposed to in a more realistic setting using field or model-based testing. But the problem is that this step is often based on a “single-crop, single-product” approach that doesn’t paint a complete picture.

Under that approach, field tests for a given pesticide typically involve applying the chemical to a plot of a single type of crop, then estimating the impact on organisms. But Topping argues that ignores the reality that farmers often use multiple different chemicals on their lands, which can introduce interactive effects that the tests don’t account for.

Such tests also fail to account for the cumulative effects of pesticide use across time. If the reproductive health of a population of birds is hurt a little each time a pesticide is applied, their ability to recover gradually slows. Eventually, they’re not able to rebound like they may have after an initial test of the pesticide—and the birds’ numbers permanently decline.

Using just one plot as a representative sample also doesn’t provide an accurate picture of species’ ability to recover in other locations. For example, say a specific species of pollinator is lost after a spray of insecticide. If there isn’t other habitat nearby hosting more of the bugs, they might not be able to return and recover their numbers.

“Certainly the single-crop, single-product approach to pesticide risk assessment does not reflect the reality out in the field,” says Valery Forbes, an ecotoxicologist at the University of Minnesota, who was not involved in the study. “I think the authors capture some of the major flaws in current approaches to pesticide risk assessment.”

In the EU, government data on agricultural crops and inputs could be the key to a new way of pesticide testing, says Topping. Using this data, researchers can build sophisticated landscape models for different crops across various regions and climates, down to one meter in resolution. These can show how pesticides applied will move across the area, and what species are present. “This is really quite a lot like a SimCity game,” says Topping. “You'll have the farmers who are going out in the model and monitoring their crops, applying the herbicides, pesticides, insecticides, fertilizers, changing the crops in the fields. And the animals will be moving around inside this kind of virtual reality ... So these are very detailed models.”

This type of modelling would help farmers make more informed decisions on which pesticides to use and in what quantities, as well as understand how long it takes organisms to recover. Continued data from farmers and scientists could help make the information more accurate too.

Still, Topping says the environmental risks need to be balanced with the reality of needing enough crops. “We want these pesticides because they’re needed to kill the pests which are otherwise threatening food production,” he says. “There has to be a balance, and part of that balance is understanding the real effects on the environment.”

In the States, it would likely be more challenging to implement such a system, says Anne Fairbrother, an ecotoxicologist with Exponent, a consulting company. “Most states don’t even collect data in terms of pesticide use at the farm level.” Even so, the federal government could still improve guidelines for assessing impacts on non-pest organisms, adds Fairbrother, who has also worked for the EPA. Instead of just calling a pesticide safe or not, as the current process does, updated guidelines could convey a more precise level of risk. Another improvement would be addressing how a chemical affects the recovery of endangered or sensitive species. “None of our toxicity testing and risk assessment really asks the question of how is this pesticide actually going to affect the populations of things that we’re concerned about,” says Fairbrother.

Toppings adds that a future system based on modelling and monitoring agricultural landscapes wouldn’t necessarily be more restrictive. Instead, it would help create tailored “prescriptions” for what pesticides to use for a given pest in a given region. It’s even possible that fewer chemicals would be banned outright—their application would more likely be constricted to a given window in which the timing and amount is least likely to affect non-pest species. “[It would be] very much like going to the doctor and getting a prescription for antibiotics,” says Topping. “If you’ve got a really bad disease, you can get a very strong [medication] but you’re not going to get that for everything because just like antibiotics, it’s not good to use very strong pesticides.”