The ridge east of the bay smells like dry grass and hot stone by August.
Every afternoon the valley heats up and pulls air through the gaps in the hills.
Eagles have worked that lift for as long as anyone has watched this place.
The turbines came later, planted on ground the birds already held.
Counted from the road, the coexistence looks settled.
Counted by the age of the birds, it does not.
Why a wind farm turns out to be good eagle country
The pass sits in a low saddle in the northern Diablo Range, where inland heat drags air through the hills almost every afternoon.
That reliable wind is why the machines are there, and part of why the surrounding range holds one of the densest concentrations of breeding eagles anywhere.
An eagle hunting on the wing needs three things. Lift without flapping, open ground to see across, and something on it worth dropping on.
The prey here is the California ground squirrel, which likes short grass, disturbed soil and the margins of roads, and a turbine site supplies all of it.
Access roads, graded pads and cleared verges are squirrel habitat that did not exist before the wind farm.
So the machines did not displace the birds. They sat down inside a working hunting ground and, in one respect, improved it.
What the surveys found on the ground
Inside a two mile ring around the turbines, eagles were about as thick on the ground as anywhere else in the range.
One territorial pair per roughly 7 square miles surveyed inside the ring, against one pair per roughly 8 square miles outside it.
Breeding tracked the same way. Pairs inside averaged 0.37 young fledged, pairs outside averaged 0.30.
Those are the two measures a wind farm gets judged on, and on both the turbine ground reads as ordinary.
It is the third measure that does not.
29 percent of pairs inside the ring included at least one bird still in subadult plumage, against 3 percent of pairs outside it.
What the fieldwork actually covered
The work ran across a 2,000 square mile block of the northern Diablo Range, in most years between 2014 and 2021.
Randomly selected plots were searched repeatedly for occupancy and reproduction, and about 230 territorial pairs were identified across the whole area.
Up to 21 overlapped the turbine ring in any given year, which is the entire sample the wind farm can offer.
The age comparison is thinner still. It rests on a mean of 16 pairs aged inside the ring against 122 outside.
Across the survey years the share of pairs fledging at least one young ran from 0.15 to 0.30, with 0.27 and 0.26 in the final two.
An earlier round on the same ground, across three seasons, found the same pattern and something sharper with it.
What sixteen pairs can and cannot carry
In that earlier round, between 23 and 36 percent of territories within the ring held a subadult member, and not one mixed age pair attempted to nest.
A subadult on a territory means the adult that held it is gone, and the authors read the gap as higher adult mortality or displacement at territories overlapping the wind farm.
But 16 pairs is 16 pairs. A handful of birds either way moves that percentage a long distance, and the report is careful about it.
The turbines are also not the only pressure. A severe fire in 2020 burned through more than 60 historical territories in the study region, which the authors flag as a threat nobody has measured yet.
Separating a fire from a rotor from a poisoned squirrel is not what an occupancy survey was built to do, and the old baseline was brutal on its own, with an early study putting the toll near a thousand raptors a year across 5,000 small turbines.
The full survey data and methods are public, which is more than most sites offer.
What has changed on the ridge since
The hardware is not what it was. Some 569 machines from the 1980s, most rated at 100 kilowatts or less, came down and 23 modern ones went up in their place.
Fewer, taller rotors sweep a band above the height where a hunting eagle spends most of its time, which is the whole theory of the swap.
Operators have also tried cameras that spot a bird half a mile out and stop one machine at a time.
The numbers on those are contested. One study reported an 82 percent drop and a follow up put it at 85, with an interval so wide it runs from 12 percent to complete, and a separate published reanalysis found little evidence of a reduction at all.
That is the honest state of the mitigation case, for the camera towers as much as the new steel.
What the survey does say is narrow and hard to argue with. The golden eagle pairs are still there, still breeding at an ordinary rate, and turning over faster than the birds next door.
