What The RSPB's Air-Quality Paper Leaves Out

✅ KEY TAKEAWAY: The RSPB’s air-quality argument against prescribed burning is flawed; it misinterprets air-quality standards and ignores that controlled winter burns prevent massive PM2.5 emissions from severe summer wildfires.

In the wake of Britain's worst wildfire season on record, the University of Leeds and the RSPB have published modelling work in Environmental Research Letters and are calling for grouse-moor licensing in England on air-quality grounds. The science deserves a fair reading. The policy translation deserves a harder one.

The 2025 wildfire toll, drawn from NFCC and Defra reporting, comes to 47,879 hectares burned, 1.3 million tonnes of CO₂, 8,809 tonnes of PM2.5 fine particulate matter, and an economic cost of £460 million. A subsequent epidemiological assessment puts the modelled upper bound at 65 premature deaths from the season's smoke.

The 2018 Saddleworth fire, which exposed an estimated 4.5 million people to elevated particulate levels, has been linked to around 28 premature deaths. That is the public-health context in which the new paper lands.

The paper itself, lead-authored by Dr Ailish Graham at Leeds, combines satellite-detected burn data with regional air-quality modelling to estimate population exposure to PM2.5 from moorland burning between 2017 and 2022.

It concludes that, on days when prescribed burning occurred during the burn season of October 2017 to April 2018, an additional 0.55 million people on average were exposed to PM2.5 levels above WHO daily guidelines, with a peak of 2.3 million across the period.

The RSPB has translated that finding into a call for the Scottish grouse-moor licensing model to be extended to England. The RSPB is a wildlife charity and has no expertise in wildfire or human health.

The wrong yardstick

The exposure threshold the paper uses is the WHO 24-hour PM2.5 guideline of 15 µg/m³. That is an advisory aspiration. It is not UK law. Britain's binding PM2.5 standards, under the Air Quality Standards Regulations 2010 and England's Environmental Targets (Fine Particulate Matter) Regulations 2023, are framed in annual means and long-run population exposure reduction. No daily statutory limit exists.

Defra's 24-hour Daily Air Quality Index, with a "Moderate" trigger at 35 µg/m³, advises vulnerable individuals; it is not a regulatory ceiling on emitters. The paper measures a wet January plume against a yardstick the UK has not adopted in statute, and the RSPB press release recommends legislation on that basis.

The headline number, in context

The release leads with the line that moorland burning emits the equivalent of "almost a third" of road transport's PM2.5. The paper's own absolute figure is more sober: 1,300 tonnes per year of primary PM2.5, around 1.5% of total British anthropogenic primary PM2.5 emissions.

The "third of road transport" framing is impressive only because primary PM2.5 is itself a small share of road transport's air-quality footprint - most of the road sector's damage to UK air is NOx and secondary particulate, not primary PM2.5.

Choose the smallest fraction of the second-largest source and almost anything looks comparable. The honest scale of the moorland-burning contribution to UK primary PM2.5 is in the release itself: 1.5%.

The modelled exposure figures also need context. The 0.55 million and 2.3 million numbers are population-weighted dispersion outputs for the burn season of October 2017 to April 2018, not annualised exposure.

The methodology is a defensible way of isolating prescribed burning from wildfire - wildfires were less than 5% of the burned area in that window - but it means the headline figures describe the burn season specifically. A non-specialist reader will not necessarily catch that.

Read the fuel load

Rank heather is fuel. Peer-reviewed estimates put fuel loads on unmanaged moorland as high as 62.6 tonnes per hectare. Ignited under dry, windy summer conditions, that produces a fire intensity above 10,000 kilowatts per metre: beyond ground crews to suppress, hot enough to generate its own weather, and hot enough to take the deep peat with it.

A cool burn done in damp winter conditions is the inverse operation. It strips the surface canopy, drops the fuel load to 2–3 tonnes per hectare, and leaves the underlying peat untouched. Same hillside, different fire, by design.

The emissions arithmetic follows. Uncontrolled wildfires emit two to four times the PM2.5 of a prescribed burn per unit of biomass consumed, and far more in absolute terms because they consume far more biomass. A severe wildfire can release up to 100 tonnes of carbon per hectare once the peat catches. A controlled winter burn releases 1 to 2 tonnes.

The 1,300 tonnes of PM2.5 the Leeds paper attributes to a typical year of moorland burning has to be set against the 8,809 tonnes of PM2.5 generated by the wildfires of 2025 alone.

"Not naturally fire-adapted"

The stronger argument in the press release is the claim that British landscapes are not fire-adapted, and that prescribed burning is therefore an imported solution to an imported problem. Rewetting, on this account, can deliver fire resilience without smoke.

The Holocene record is not on its side. Charcoal and pollen sequences from northern English and Scottish peats show fire as a recurrent feature of these landscapes for thousands of years, long predating gamekeeping.

The recent fire on the rewetted lowland bog at Danes Moss in Cheshire makes the same point in real time. A wet peat layer beneath shoulder-high tinder is still a wildfire waiting for a spark. Rewetting matters. It does not, on its own, defuel.

The peer-reviewed direction of travel is sequencing rather than substitution. Research on a "burn-to-rewet" approach - removing surface vegetation on degraded peat before raising the water table - reports methane emissions falling by more than 95% within 90 days, and full Sphagnum recovery within five years.

The mosaic of vegetation heights that approach leaves behind is also the habitat structure on which curlew, lapwing and golden plover depend. Uniform wet sponge is not a wader landscape.

What licensing actually has to weigh

The specific policy ask is the extension of the Scottish grouse-moor licensing model to England. Whatever one thinks of that proposal on its merits, a regulator weighing it on air-quality grounds has to do three things the Leeds/RSPB release does not.

1. It has to compare prescribed-burn emissions against the wildfire emissions they avoid, not against an unmodified counterfactual.

2. It has to benchmark against UK statutory air-quality standards, not against advisory guidelines the UK has chosen not to legislate.

3. And it has to engage with the keepers, land managers and fire and rescue services who do the actual work of keeping summer fuel loads in check, rather than treating their experience as a category of error.

Until that analysis is on the table, the public-health case for restricting cool burns has not been made. It has been asserted.

Technical concerns with this paper

Consulting academic expertise has raised five technical concerns about the paper. All five point in the same direction: the paper likely overstates the air pollution emissions.

Starting with the model. The authors openly report a 36% positive bias against observed PM2.5 concentrations. That is not a rounding error requiring explanations, which likely relate to concerns about two model inputs which push the same way. The fraction of biomass burned (FB) is set at 0.9 - as stated a value for European grasslands.

However, heather moorland comes in at 0.79 in published UK fieldwork, roughly 13% lower. The PM2.5 emission factor (EF) is set at 7.1 g/kg but another published standard Andreae & Merlet inventory gives 5.4, around 24% lower. Both should have been carried as ranges but coincidentally add up to 37%, near identical to the positive bias.

Then the regional fuel load. The drier moorlands of the North York Moors and East Scotland often carry younger, smaller heather with far less biomass per burn. Burn rotation data should exist for such areas. They are not in the paper.

Stack the three quantitative overestimates together - 36% modelled bias, 13% on FB, 24% on EF - and the cumulative inflation is roughly 70%. Strip it out, and prescribed moorland burning falls below 0.5% of UK PM2.5 emissions. The 0.8%–1.9% headline does not survive its own assumptions.

Then the Saddleworth Moor comparison. The paper sets a full season of controlled burns against one wildfire. The much larger Winter Hill fire, burning the same weeks, is absent from the table. A like-for-like comparison would aggregate every wildfire across the study periods and benchmark that total against the prescribed one. Comparing apples vs apples is important when making policy relevant comparisons.

The policy citation is also out of date. The paper leans on the Climate Change Committee's 2020 land-use report to support a ban on rotational burning. The CCC's 2023 progress report to Parliament (Box 2.3) is clear: "The CCC is in the process of reviewing its previous recommendations on the practice of rotational burning as a land management technique." Three years on, the older line is still being cited as settled.