It’s no secret that wind power has experienced a boom in recent years, as demand for renewable energy sources grows. But while the technology is adept at helping curb greenhouse-gas emissions, the thousands of new turbines popping up around the globe do have some drawbacks.

Wind farms have a long-documented history of killing hundreds of thousands of birds and bats each year. As it turns out, the bat toll may be higher than previously estimated.

In a study published Monday, researchers in the United Kingdom found that environmental impact assessments – the main tool used to predict the ecological effects of a new energy development – commonly failed to predict the number of bats that would have fatal collisions with wind turbines’ spinning blades. Even in the few cases where researchers said early assessments accurately predicted the danger to bats, efforts to mitigate those risks often did not succeed.

“The findings highlight the difficulty of establishing with certainty the effect of major developments before they occur,” co-author Fiona Mathews said in announcing the results, which were published in the journal Cell Press.

Mathews, a mammalian biologist at the University of Exeter, and several colleagues surveyed 46 wind farms across the U.K. over the course of a month to estimate bat fatalities, relying heavily on search dogs to locate fallen bats. They then compared their findings from each site to the environmental assessments they were able to access. In most cases, the pre-construction assessments had not accurately predicted the risk of bat fatalities. And even where companies had put in place mitigation measures to try to steer bats clear of the turbines, the researchers found that bats were still killed.

The reasons why aren’t entirely clear. The researchers say it is uncertain whether the acoustic surveys widely used to estimate bat activity are not precise enough or whether bats’ “highly variable” activity means they change their patterns too often to predict with accuracy.

“Bat activity recording during pre-construction surveys may not accurately reflect activity post-construction,” the authors write. “This may be due to bats changing their behavior at turbines, as bats may be attracted to wind farm sites for a variety of reasons, including the emission of ultrasound from turbines and increased prey availability.”

In an in-depth article on the problem this year, Scientific American detailed how the wind industry had put in place voluntary guidelines to halt turbines at low wind speeds, when bats are most active. Conservationists praised the move, even as some scientists said more needed to be done.

Paul Cryan, a bat biologist with the U.S. Geological Survey, told the magazine he worries about the ongoing impact of turbines on bat populations, which are an essential link in certain ecosystems. “Bats are long-lived and very slow reproducers,” he said. “Their populations rely on very high adult survival rates. That means their populations recover from big losses very slowly.”

USGS scientists at the Fort Collins Science Center in Colorado have built a research program aimed at investigating the causes and repercussions of bat fatalities at wind farms, with the goal of minimizing deaths over time even as wind energy proliferates.