NEW YORK (Reuters) – As the U.S. growing season entered its peak this summer, farmers began posting startling pictures on social media: fields of beans, peach orchards and vegetable gardens withering away.

The photographs served as early warnings of a crisis that has damaged millions of acres of farmland. New versions of the herbicide dicamba developed by Monsanto and BASF, according to farmers, have drifted across fields to crops unable to withstand it, a charge authorities are investigating.

As the crisis intensifies, new details provided to Reuters by independent researchers and regulators, and previously unreported testimony by a company employee, demonstrate the unusual way Monsanto introduced its product. The approach, in which Monsanto prevented key independent testing of its product, went unchallenged by the Environmental Protection Agency and nearly every state regulator.

Typically, when a company develops a new agricultural product, it commissions its own tests and shares the results and data with regulators. It also provides product samples to universities for additional scrutiny. Regulators and university researchers then work together to determine the safety of the product.

In this case, Monsanto denied requests by university researchers to study its XtendiMax with VaporGrip for volatility – a measure of its tendency to vaporize and drift across fields.

The researchers interviewed by Reuters – Jason Norsworthy at the University of Arkansas, Kevin Bradley at the University of Missouri and Aaron Hager at the University of Illinois – said Monsanto provided samples of XtendiMax before it was approved by the EPA. However, the samples came with contracts that explicitly forbade volatility testing.

“This is the first time I’m aware of any herbicide ever brought to market for which there were strict guidelines on what you could and could not do,” Norsworthy said.

The researchers declined to provide Reuters a copy of the Monsanto contracts, saying they were not authorized to do so.

Monsanto’s Vice President of Global Strategy, Scott Partridge, said the company prevented the testing because it was unnecessary. He said the company believed the product was less volatile than a previous dicamba formula that researchers found could be used safely.

“To get meaningful data takes a long, long time,” he said. “This product needed to get into the hands of growers.”

‘Jeopardize the Federal Label’

Monsanto employee Boyd Carey, an agronomist, laid out the company’s rationale for blocking the independent research at a hearing of the Arkansas Plant Board’s Pesticide Committee in the summer of 2016.

A meeting summary by the Arkansas Legislature’s Joint Budget Committee described Carey’s testimony as follows: “Boyd Carey is on record on Aug. 8 stating that the University of Arkansas nor any other university was given the opportunity to test VaporGrip in fear that the results may jeopardize the federal label.”

Efforts to reach Carey were not successful. Monsanto declined to comment on his testimony.

To be sure, complaints about damaged crops are still under investigation and there is no evidence that independent testing of XtendiMax’s volatility would have altered the course of the crisis. But it would have given regulators a more complete picture of the formula’s properties as they decided if and how to let farmers use it, agriculture experts said.

In the end, the EPA approved the product without the added testing in September. It said it made its decision after reviewing company-supplied data, including some measuring volatility.

“EPA’s analysis of the data has shown reduced volatility potential with newer formulations,” the EPA said in a July 27 statement.

However, EPA spokeswoman Amy Graham told Reuters the agency is “very concerned about the recent reports of crop damage” and is reviewing restrictions on dicamba labels.

Monsanto Chief Technology Officer Robert Fraley said, “We firmly believe that our product if applied according to the instructions on the label will not move off target and damage anyone.”

States Approve Without More Testing

Companies can limit independent testing because the substances are proprietary. When samples are provided to researchers, lawyers hammer out contracts detailing how testing will be conducted and results will be handled, but rarely do agreements limit what the products can be tested for, according to researchers interviewed by Reuters.

For instance, BASF, which introduced its rival herbicide, Engenia, around the same time, said it allowed several university researchers to evaluate its “off-target impact and application parameters.”

Norsworthy, of the University of Arkansas, confirmed he had been permitted by BASF to study Engenia for volatility and that the results showed less volatility than previous dicamba formulations. BASF says its product is safe when properly applied.

The EPA did not answer questions about whether it noticed a lack of input from university researchers about XtendiMax’s volatility or whether it requested such testing.

It also did not address whether the lack of independent research played into its decision to give the product an abridged two-year registration, less than the 20 years experts say is more common. The agency did the same for BASF’s Engenia.

“The EPA placed time limits on the registration to allow the agency to either let it expire or to easily make the necessary changes in the registration if there are problems,” Graham, the EPA spokeswoman, said.

After the EPA signed off, Monsanto sought approval from individual states, which determine whether agricultural products are suitable for their climates and geographies.

To help them do that, Monsanto shared its XtendiMax testing results with state regulators. But it only supplied that data in finished form, Monsanto’s Carey told the Arkansas Plant Board meeting, meaning it withheld underlying data that could be analyzed independently by the regulators.

Only Arkansas wanted more. Terry Walker, the director of the Arkansas Plant Board, said the state asked Monsanto for extra testing, but the company refused.

“As the system progressed and it got closer to EPA approval, the board kept asking for local data,” Walker said. “That did not happen.”

Monsanto’s Fraley said the company could not honor Arkansas’ request within the EPA’s timeline. “Given the timing of the approval… there simply wasn’t the opportunity to do the additional testing,” he said.

Arkansas blocked Monsanto’s product because of the lack of extra volatility testing by universities, but approved BASF’s because it had not limited such testing and the results were acceptable. Thirty-three other states – every other state where the products were marketed – approved both products.

After Arkansas blocked XtendiMax in December, crop damage began to appear in the state anyway. Investigators trying to determine the cause of the damage are considering a range of possibilities including problems with or improper use of Engenia or illegal use of XtendiMax or earlier formulations. In July, the state banned all products containing dicamba.

Some states including Illinois, Missouri and Tennessee said they do not seek the more data if products pass EPA scrutiny.

“The EPA is the federal agency responsible for approving and registering pesticides for sale and use,” said Missouri Department of Agriculture spokeswoman Sarah Alsager. “The Department does not perform field testing or solicit local input.”

Some states are now forming task forces to determine what should be done about the damage.