Science1 distinct publisher3 min readPublished
A study of 788 fires that burned into 499 western communities argues the fix is fire-resistant land next to town, and that a decade would do it if fuel-treatment money moved.
The Scientist · Science desk

Compiled by The ScientistSomething wrong?How this is made
The ranking is where the argument sits. Thirty-seven land uses, sorted by how often they actually burned between 2000 and 2021, fall into five bands: the safest burned once every 1,400 years or more, the Very Dangerous ones once every 150 years or less [3][4]. Wildland vegetation sits in the Dangerous band at one fire per 190 years [5], so the safest ground has a return interval roughly 7.4 times longer [1]. None of that is a fuel-load measurement. It is a record of what the ground was being used for when fire arrived.
Two of the headline percentages do not nest. The study reports that the fires reaching those 499 communities burned 95% in wildland vegetation and 88% in the Dangerous and Very Dangerous categories combined [6]. Wildland vegetation is itself classed as Dangerous [5], which means the combined share should be at least 95%, not seven points below it [3]. The phys.org summary does not state the denominator behind either figure, and anyone being asked to rezone land on the strength of the number needs that spelled out.
The 2025 Los Angeles fires are offered as independent support, and they are weaker support than the average. The Palisades Fire burned 84.6% in wildland vegetation, the Eaton Fire 81.9% [8][9], which puts them 10.4 and 13.1 points below the long-run figure for community-reaching fires [2]. The author's reading is that even under strong Santa Ana winds, fires mostly stayed in wildland vegetation and dangerous uses [10]. The other reading is that on the windiest days a larger slice of the burn happens somewhere else, and the windiest days are the ones that destroy towns.
Jon Keeley of the U.S. Geological Survey supplies the mechanism in its plainest form: fifty years ago parts of the basin were ringed by citrus orchards, and fires coming out of the mountains burned out in the orchards before reaching homes [11]. That is the operative difference between this proposal and thinning. But look at the safe list. Cultivated crops, freeways, lakes, rivers and deserts [7]: three of those are pre-existing physical features, which leaves irrigated agriculture and paved corridors as the only safe uses a county can actually install [5].
Then the footprint. Communities plus their 5,000 m buffers cover 20.45% of the eleven western states [16], about one acre in five [4]. Buffers of neighbouring towns merge into continuous belts, as on the Nevada-California line [20], and the author's conclusion is that communities have to combine and go to the federal government for major funding [18]. The specific ask is that most fuel-reduction treatments under the roughly $7 billion Confronting the Wildfire Crisis program move soon to ground very near communities [17][19]. That program exists to reduce fire severity [17]. Reducing severity in the backcountry and converting land use next to town are not the same job, and the request is to spend one budget on the other.
Ranked by verification strength, evidence, and original report placement.
A new study in Environmental Research: Ecology, summarised by its author for phys.org, analysed how fires burned into communities and how fire-resistant land uses combined with fuel treatments could reduce fires burning into communities.
Based on fires from 2000-2021, 37 land uses define five land-use safeness categories from Safest to Very Dangerous.
Safest land uses burned once every 1,400 years or more; Very Dangerous uses burned once every 150 years or less.
Wildland vegetation was rated Dangerous, burning once every 190 years.
The 788 fires from 2000-2021 that burned into 499 communities burned 95% in wildland vegetation, 88% in Very Dangerous and Dangerous land uses, and only 6% in safe land uses.
The study found grasslands, chaparral and wildland vegetation vulnerable to wildfire, while cultivated crops, freeways, lakes, rivers and deserts were fire-resistant.
Follow any of these and your For You feed starts watching them — no settings page required.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Quantified study findings, single self-authored account
The empirical core is specific and checkable: 37 land uses binned into five safeness categories with fire return intervals, 788 fires burning into 499 communities from 2000-2021, per-fire composition for the 2025 Palisades and Eaton fires, and a 20.45% community-plus-buffer footprint, all traced to a named peer-reviewed article in Environmental Research: Ecology. It is pulled down by the fact that the only supplied source is the author's own summary of his own paper, with no independent verification, one unreconciled internal arithmetic conflict between the 95% and 88% shares, and an explicit concession that the proposed buffer system as a whole could not be tested.
No adoption evidence supplied
The supplied source describes a proposed framework and a list of steps communities could take, but records no community, agency or program that has adopted the buffer system, no relocation of Confronting the Wildfire Crisis treatments, and no dataset usage. The existing ~$7 billion program funding is prior spending the author wants redirected, not evidence of uptake of this proposal, so no adoption level can be measured.
Decade-scale promise outruns an untested framework
The framing -- protecting all western US communities and restoring fire to wildlands within a decade -- is materially stronger than what the source can support. The retrospective fire-composition analysis is substantive, but the prescriptive system that carries the timeline is acknowledged as untested, has no adoption, no cost or land-acquisition mechanism, and rests on figures with an unreconciled 7-point inconsistency. The 2025 Los Angeles fires are presented as independent confirmation even though their wildland-vegetation shares run 10.4 and 13.1 points below the study's own baseline.
Author-written advocacy for redirecting his own program of work
The piece is published through Science X Dialog explicitly so researchers can report their own published findings, and the author uses it to argue for redirecting roughly $7 billion of existing federal fuel-treatment money, for federal agencies to donate land to communities, and for communities to adopt the datasets from his paper. That is a clear promotional and policy-advocacy incentive structure with no counterparty voice in the cluster; no undisclosed financial interest is evidenced, which keeps the reading short of the top of the scale.
Single publisher, single voice, quantified but unreplicated
Confidence is limited by cluster structure: one publisher, one author, no independent expert or agency reaction, and no adoption signal to triangulate. The retrospective quantitative findings are stated precisely and attributed to a named journal article, which supports moderate confidence in the descriptive claims, but the prescriptive system, the decade timeline and the funding-reallocation case cannot be corroborated from the supplied material, and one internal arithmetic conflict remains unexplained.
science
Tides vary by nearly a metre inside one bay, and gauge-based flood maps smooth it out1 distinct publisher
science
A Vermont catfish cancer turns out to be contagious, not pollution damage1 distinct publisher
product
Geologic hydrogen joins the clean-fuel race with no cost curve and a leak problem1 distinct publisher
product
Infleqtion points quantum gravity sensing at mining's worst line item: drilling blind1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
phys.org
1 article · August 25, 2026