The ridge country of Converse County, Wyoming, east of Casper, hums with turbines from dawn until the wind drops.
Bolted onto steel poles among them sit camera towers, each one scanning the sky.
They were built for a single job.
Spot a golden eagle before it gets close to a spinning blade.
For years, one number followed that system everywhere it was mentioned, an 82 percent drop in eagle fatalities.
Then two federal statisticians went back through the same data.
How a detection system is supposed to save an eagle
Each tower carries a ring of wide angle cameras plus a swiveling pair that zooms in once something crosses the frame. Software classifies the bird in real time, and the system can detect one as far as a kilometer out, roughly six tenths of a mile. A base station then decides whether any turbine in range needs to stop.
That decision travels into the wind farm’s own controls, and at this project turbines could be curtailed as quickly as 20 seconds. Engineers tune the classifier to miss as few eagles as possible, which means it also halts turbines for birds that turn out to be something else. The mechanical chain is short: a camera watches, a computer decides, a blade slows.
A ridge of turbines and the birds circling above it
The site sits on open rangeland where golden eagles hunt ground squirrels and rabbits by riding thermals along the slope, the same rising air that made the ridge attractive for wind power in the first place. That overlap between good wind and good eagle habitat is not unique to Wyoming, and it is a legal problem as well as a biological one.
Operators in this part of the state work under federal eagle permits. The neighboring 66 turbine project in the same county holds an eight year permit authorizing potential take of bald and golden eagles, with required conservation and compensatory mitigation measures attached. Cameras that stop blades are one of the tools offered against that risk.
What the carcass counts actually showed
Finding a dead eagle under a turbine is unreliable work. Scavengers carry bodies off, grass hides them, and searchers miss some fraction every time, so researchers count the carcasses they find and then model how many went undetected.
A team from a raptor conservation nonprofit, a wildlife consulting firm and the U.S. Geological Survey compared corrected counts before and after the cameras came on, using the 66 turbine site about nine miles away as an untreated control. At the camera site, fatalities declined by 63 percent between the before and after periods, while fatality numbers at the control site increased by 113 percent. Combining those two changes produced the widely repeated 82 percent figure.
A later paper by the same group put the reduction at 85 percent and estimated the fatality rate fell by 4.91 more eagles per year at the treated site than at the control, drawing on additional monitoring through 2021. In raw numbers the change was modest: the rate fell by about 2.85 eagles a year at the treated site and rose by about 2.26 a year at the control. Neither detail traveled as far as the round number did.
Why a reanalysis produced a much smaller number
The challenge came from inside the same research community. In 2023 the Journal of Applied Ecology published a reanalysis by Manuela Huso and Daniel Dalthorp, both affiliated with the U.S. Geological Survey, whose title stated plainly that it found little evidence of a reduction in eagle mortality rate.
They argued the published result stemmed from four major errors which, corrected, give an estimate of 50 percent, or possible increase, in the fatality rate. One error they identified was comparing the average number of fatalities over four years before activation with the number in a single year after, ignoring annual variation. They also noted that 7 of 42 reported eagle fatalities were not included in the analyzed data.
The original authors answered in the same journal and the same year, disputing the claim that major errors rendered their work meaningless and arguing the reanalysis ignored pertinent data and included inappropriate observations. The dispute is unresolved in the literature, and readers should treat the 82 percent figure as contested rather than settled.
What still works and what nobody can promise yet
None of this means the cameras are useless. The reanalysis authors themselves wrote that automated curtailment has obvious potential for reducing eagle fatalities in theory, but that several more years of data at several locations and appropriate statistical analyses will be required to evaluate its effectiveness.
What the dispute really challenges is the confidence anyone should place in a single comparison between two ridgelines over one study window. The same difficulty shows up wherever renewable projects try to prove a mitigation measure worked while the surrounding landscape keeps shifting on its own, from wildlife corridor planning at desert solar sites to the pronghorn corridors mapped around a large Nevada solar project.
For now, the towers keep scanning the ridges of Converse County. The eagles keep riding the same thermals that drew both the birds and the turbines there in the first place. The honest number may be smaller, and far less certain, than the one that made headlines, but the birds overhead are still being watched, camera by camera, every single day.
Read the whole thing?
Get the week's signal, not the noise
Our sharpest reporting on energy, climate and nature — free, once a week.
