wind-energy wildlife birds planning habitat-loss biodiversity

The Ground the Eagles Gave Up: Wind Farms, Birds of Prey and Predictions Rarely Tested in Public

Tracked golden eagles effectively abandoned inner turbine positions at two Scottish wind farms. The harm measured is lost ground, and predictions made before consent are rarely tested in public.

hostile.eco

At two wind farms in the Monadhliath mountains of Scotland, researchers followed young golden eagles carrying GPS tags and watched how they used the ground around 99 turbines. What they found was not a pile of carcasses. It was an absence: “After operation, eagles effectively abandoned inner turbine locations, and flight line records close to rotor blades were rare.”

The usual case against badly sited wind farms is made with dead birds and with migration routes. On the British evidence gathered for this article, neither carries it. Collisions involving birds of prey at British wind farms are recorded in small numbers: three red kites in 50 months of searches at one site, one golden eagle found at another. No British raptor migration corridor is documented in any source we read. What the evidence does carry is narrower and harder to dismiss. Birds of prey lose ground to wind farms, measurably. And the predictions made before a wind farm is consented are almost never tested in public afterwards.

What the eagles did

The two sites are Dunmaglass and Stronelairg. Dunmaglass has 33 turbines of 3 MW and Stronelairg 66 of 3.6 MW. Stronelairg was sought at up to 83 turbines and consented at 67 by the Scottish Energy Minister in June 2014.

The tracking study drew on “detailed data from 99 turbines at two wind farms in central Scotland and thousands of GPS-telemetry data from dispersing golden eagles”, using 19,629 location fixes to measure distance from a turbine. Once the turbines were running, the eagles kept their distance: “birds rarely approached closer than 80 m to the tip of turbine blades.” The study found one collision, and found it incidentally: “at Stronelairg a collision casualty of an untagged subadult golden eagle was found under an outer turbine in early May 2020.”

The eagles in that study were young, non-territorial birds. A wider study by the same research group covered 569 turbines at 80 wind farms across Scotland and 59 GPS-tagged eagles with 28,284 records during natal dispersal. After the turbines began operating, “Eagles were eight times less likely to be within a rotor diameter’s distance of a hub location after turbine operation, and modelled displacement distance was 70 m.” A 2023 study of birds that do hold territories, at 11 operational Scottish wind farms, reached the same place: “territorial eagles rarely approached turbines.”

These are resident birds, not passing migrants. The UK’s golden eagles number 400 to 500 breeding pairs, and the BTO says the population “appears to be self-contained, with no evidence of interchange with those elsewhere.”

An eagle that keeps 80 m from the blade tips does not hit them, which is part of why the collision count is low. It has also given up the ground among the turbines. A body count measures the first. The tracking measured the second.

Two incentive notes belong here. SSE, the developer of Stronelairg, financially assisted the production of the manuscript of the tracking paper about its own wind farm, and was among the funders of the tags; the authors state that SSE had “no influence or commentary” on it. And the consultancy behind the Scottish tracking studies, Natural Research, also wrote the reports on which NatureScot’s golden eagle and red kite avoidance rates rest (Whitfield, 2009; Urquhart and Whitfield, 2016). Neither fact makes a figure wrong. Both are reasons to read the method.

Fewer birds near the turbines

A study of 12 UK upland wind farms, Pearce-Higgins et al. 2009, found that “breeding bird densities may be reduced within a 500‐m buffer of the turbines by 15–53%”, and that “Seven of the 12 species studied exhibited significantly lower frequencies of occurrence close to the turbines, after accounting for habitat variation.”

That 15 to 53% needs its scope kept. It is a range across several upland species, not a raptor figure. The species most affected included buzzard and hen harrier, both birds of prey, and golden plover, which is a wader. The source says densities “may” be reduced, and the same study found “no evidence that raptors altered flight height close to turbines.”

A 2012 follow-up is often quoted for the finding that construction harms birds more than operation does. It cannot be applied to birds of prey: “Data were available for ten species although none were raptors.”

For what displacement costs an eagle, the evidence is Norwegian. On the island of Smøla, where the wind power plant has consisted of 68 turbines since 2005, satellite-tracked “non-territorial subadult white-tailed eagles”, followed from 2004 to 2009, showed “Displacement (indicated by an overall 40% proportional reduction in utilization)”, and “territories within 500m from the turbines in the post-construction period experienced significantly lower breeding success than the same territories before construction.” Smøla also kills eagles: on average, 7.8 white-tailed eagles collide with its turbines each year. At Smøla, displacement and collision were both measured, and both were found. In Scotland, displacement of golden eagles has been measured; what it costs them in breeding success has not been measured in any source we read.

How the models count it

Before a wind farm is consented, its effect on birds of prey is usually predicted with a collision risk model. The model assumes most birds will avoid the turbines, and the avoidance rate says how many. NatureScot’s guidance for Scotland, in the version updated in September 2025, sets the rate at 99% for golden eagle, red kite and hen harrier, 95% for white-tailed eagle, with “a default value of 98%” for species not listed. Its September 2018 version gave the same figures for these species.

The white-tailed eagle’s lower rate rests on “Sufficient evidence from flight behaviour and collision monitoring studies in Norway for vulnerability to collisions”. The evidence for that figure is Norwegian, not British.

In a collision model, avoidance is the good news: the higher the rate, the fewer predicted deaths. The guidance says what the rate covers: “As birds may avoid a wind farm (for example some may be displaced from the area, while others may avoid turbines or take other evasive action to prevent a collision), the CRM accounts for this by applying an avoidance rate.” So in the model, a bird displaced from the area is a collision avoided. The prediction counts deaths. It does not count the ground lost, which is the harm the tracking measured.

Predictions rarely tested in public

The research behind this article gathered 18 British wind energy cases in which birds were raised as an issue, most of them birds of prey. For only one could we read both the prediction made before construction and an outcome after it. That count is of what we could read, not of what exists. Most environmental statements and monitoring reports are PDF documents, which this research did not extract, and some monitoring reports, such as Edinbane’s below, we could not find at all. A reader with time and the PDFs might match more predictions to outcomes. What we found nowhere is a systematic published comparison of the two, or a register that collects them, in any UK nation.

The one case with both sides is Braes of Doune, near Stirling. It was applied for at 49 turbines and built as “36 V80 2.0 MW (megawatts) turbines with a total nominal output of 72 MW”, commissioned in February 2007. Its developer, Airtricity, describes the prediction this way: “All (agencies) agreed that while there was some risk of collisions, the effects were likely to be small and not significant to the Central Scotland red kite population.” Airtricity “was required to commission ongoing research and monitoring of the red kites”. Results from monitoring at the site were published in 2016 by Natural Research, the consultancy behind the Scottish eagle tracking. Searches under the turbines ran from February 2007 to March 2012, and “Three red kites were documented as collision fatalities during 50 months’ monitoring.” The first, found in July 2007, was, the BBC reported, “killed after getting caught in the blade”. Corrected for carcasses missed by searchers, taken by scavengers or thrown outside the search area, the report’s estimate is “1.28 (90 % CI = 1.11, 1.55)” red kites a year.

The report calls itself “the only known study of avoidance rate, incorporating both pre- and post- construction data, to have been completed in the UK, and possibly in Europe.” By its own account, then, Britain has one published before-and-after test, at one site. It tests the model, not the promise: it uses the deaths to work out the avoidance rate that would have predicted them, and recommends “an avoidance rate of 99 % should be used and not 98 %” for red kites, the figure NatureScot now cites it for. It does not set the environmental statement’s prediction against the outcome, and nor does any source we read. Whether about 1.3 kites a year is “not significant to the Central Scotland red kite population” is a population question the report does not address. Nothing in it shows the prediction wrong. The report notes that it “has not undergone peer-review”, and its authors thank “Scottish & Southern Energy for funding the carcass persistence field trials”. That is the whole of the British prediction-and-outcome evidence we could assemble.

The other cases stop halfway. At Pairc, on Lewis, a 57-turbine scheme of 205 MW, the developer’s own environmental statement predicted that “Three of the predatory birds a year could be killed in collisions with turbine blades”, with three to four golden eagle pairs nearby and one pair nesting at the heart of the site. No source we read establishes whether it was built, let alone how many eagles died. (The figure of 76 eagles sometimes quoted for Pairc is a campaigners’ extrapolation over the scheme’s 25-year life; three a year for 25 years is 75. The environmental statement’s figure is three a year, and it is a prediction.)

On Skye, the Edinbane scheme was cut from 27 to 18 turbines, partly “to answer concerns about eagles”. A consultant’s 2012 report, “Edinbane Windfarm: ornithological monitoring. A review of the spatial use of the area by birds of prey.”, exists. We found no public copy. At Beinn an Tuirc in Kintyre, 46 turbines came with 450 hectares of conifer plantation cleared to give eagles hunting ground away from the turbines. A before-and-after study of the resident eagles’ ranging was published in 2005; we could not obtain it, and know its result only as later researchers characterise Scottish findings, as “substantially finding macro-avoidance”.

The guidance is clear about what should happen. Natural England’s advice for onshore wind, which applies to England, Wales and Northern Ireland and was published on 9 June 2015, says: “You should gather information about wild bird activity for at least 5 years after a wind farm’s built.” It asks for carcass searches “for at least 3 years after the wind farm’s built”, and gives the reason: “to prove that mitigation has been effective and to provide information for future plans.”

The operative word is “should”. Nothing we read asks developers to publish what they find, and we found no public register of post-construction bird monitoring results in any UK nation. Asked in Parliament in July 2026 whether it had estimated how many seabirds have died colliding with offshore wind turbines, the government replied that it had “commissioned the Joint Nature Conservation Committee to undertake a project to calculate estimated mortalities from offshore wind farms”. That is a model, not a count. Information “for future plans” cannot inform a future plan if it sits in a consultant’s file.

In Spain, the predictions were tested

The strongest evidence that pre-construction assessments can miss is Spanish. A team led by Ferrer, publishing in 2012 under the title “Weak relationship between risk assessment studies and recorded mortality in wind farms”, wrote: “We analysed data from 53 EIAs in relation to the actual recorded bird mortalities at 20 fully installed wind farms.” The sites were “all located in Tarifa, Andalusia region, southern Spain, near the Strait of Gibraltar”. The result: “No relationship between variables predicting risk from EIAs and actual recorded mortality was found.” The deaths were not few. The same paper reports that “The collision rate of birds with turbines was one of the highest ever recorded for raptors, and the griffon vulture was the most frequently killed species.” An earlier count at two Tarifa wind farms recorded 151 collisions between 1993 and 2003, 73% of them griffon vultures.

This does not show that British assessments fail; Spanish wind farms, birds and assessment practice are not Scottish ones. It shows what a test looks like, and that where one was run, across 20 wind farms, the predictions did not hold. Britain’s nearest equivalent is the single Braes of Doune study, which calibrated the model rather than checking the environmental statement. Until a test like Tarifa’s is published, nobody can say whether a Scottish environmental statement is a better guide to eagle deaths than the Tarifa assessments were to vulture deaths.

Who decides, nation by nation

A common belief is that Natural England licenses wind farms to kill birds. It says plainly that it “cannot issue a wild bird licence to allow development”, and that “There are no licensing purposes to permit development or construction”. Its remit is England, and most of the cases in this article are in Scotland and Wales.

In England and Wales, killing a wild bird is an offence if done “intentionally”; Scotland adds “or recklessly”. A turbine strike falls to the defence that “the act was the incidental result of a lawful operation and could not reasonably have been avoided”. Scotland’s version of that defence requires that whoever relies on it either “took reasonable precautions for the purpose of avoiding carrying out the unlawful act” or “did not foresee, and could not reasonably have foreseen” that the act would result. So no regulator licenses a wind farm to kill a red kite. The question is settled at consent.

Who gives that consent depends on the nation:

The nature bodies advise. Ministers and planning authorities decide, and in Scotland and England their own policy tells them to give renewable energy significant or substantial weight.

The case that the system works

Predictions change outcomes. Scottish Ministers refused Stacain, a 14-turbine scheme near Inveraray, in 2009 after a public inquiry, agreeing that “wind turbines would introduce a risk of collision for golden eagles”. They refused Lewis Wind Power, 181 turbines, in 2008 because of its “significant adverse impacts on the Lewis Peatlands Special Protection Area”. At Stornoway the developer cut its scheme from 42 to 36 turbines, and RSPB Scotland and Scottish Natural Heritage withdrew their objections. In January 2026 councillors on Skye voted eight to five against the Balmeanach wind farm, with the impact on white-tailed eagles among the concerns raised; that decision is recent and may yet be appealed. The system does act on predictions. What it rarely does is check them in public.

Collisions are rare. Natural England’s own guidance says “It’s usually rare for a wild bird to collide with a turbine”. The British record we found fits that: the one published carcass search we found recorded three red kites in 50 months. The eagles’ avoidance, measured above, is part of the reason.

Red kites in Wales are recovering. Wales now has about 2,000 to 2,500 pairs. A 2024 BTO study modelled the effect of existing and planned wind farms and found it “very unlikely to lead to national population declines in Wales”. That study is a population model, not a measurement, and it was carried out with “funding and data provided by the nine wind farm development companies involved in this study, coordinated by RenewableUK Cymru”, the industry body. It carries its own caveat: a study author told the BBC that in the Elenydd-Mallaen SPA in mid Wales, where red kites are dense, there was a small possibility of the population declining if all planned wind farms were built. It is a reasoned prediction. Like the others, it has not yet been tested against an outcome. Our earlier piece on red kites and wind turbines covers the wider mortality debate.

Where people have tested fixes, some work. At Smøla, painting the rotor blades of four turbines, against four unpainted controls, produced “an average 71.9% reduction in the annual fatality rate after painting at the painted turbines relative to the control turbines”. In Wyoming, automated curtailment of turbines was linked to “an 82% (75%–89%) reduction in the fatality rate at the treatment site relative to the control site”.

What this does not show

It does not show that any British wind farm was a proven siting failure. Of the cases we could find, one could be tested from what we read, and its published outcome, about 1.3 red kites a year, does not show its prediction wrong. It does not show a raptor migration corridor in Britain: the golden eagle is resident, and the one official review of bird migration we read, a Scottish Government study for offshore wind, says “it is unclear whether birds migrate across a broad front, or within a narrow corridor”. It does not show that displacement has reduced Scotland’s golden eagle population; no source we read measured that.

This site is critical of poorly sited renewables, and so has an interest in finding siting failures. That is why the claim here is kept to what the sources carry. Two things are measured: birds of prey give up ground around operating turbines, and breeding densities of several upland species, at 12 UK upland wind farms, may be 15 to 53% lower within 500 m of turbines. One thing is missing: beyond one site’s calibration study, a public test of whether the predictions made before consent were right.

What would settle it

The fix is publication. Natural England’s guidance already says monitoring is there “to provide information for future plans”. A public register of post-construction bird monitoring, in each nation, held against the prediction in each environmental statement, would let anyone run the test Ferrer’s team ran in Tarifa. Until that exists, consents rest on predictions that, with one exception among the cases we could find, have not been checked where the public can see them.

Sources

About This Analysis

This article is part of hostile.eco’s evidence-based environmental advocacy. All claims are sourced and all data is cited. When we get something wrong we publish the correction rather than quietly editing it. See Corrections. If you find an error, please let us know.

Subscribe for Updates More Analysis

From the Sett

Weekly evidence on energy independence, wildlife siting, and what the data actually shows. No slogans. No spin. Just the source material.