Russell Buzby

In June 2025, the case for prescribed burning got the number it had been missing. A Stanford-led team reported in AGU Advances that recent prescribed burning in the western United States had reduced the severity of subsequent wildfire by about 16 per cent, and, counting the smoke from the burns themselves, had cut fine-particle emissions by a net 14 per cent (Kelp et al., 2025). The study was widely covered and quickly became the stock answer to the hardest objection hazard reduction faces: that the burns make smoke too. In the coverage that followed, the smoke ledger appeared to balance.

On 6 August, the challenge went public in earnest. Mark Kreider of the US Forest Service’s Pacific Southwest Research Station released the latest version of the code and data behind a comment on the paper, first deposited in September 2025. The deposit’s description states the problem plainly: the original analysis classifies an entire prescribed-fire treatment as “encountered” by wildfire whenever any smoke-emissions point falls within a buffer around a single treatment point. That buffering multi-counts the same burned area, crediting more ground as treated-and-burned than was ever treated. “Correcting this error reverses a main conclusion of the paper: prescribed fire does not reduce net emissions” (Kreider, 2026). Precision about the status of this claim matters. No journal has published the comment yet, the original paper stands without correction or retraction, and the original authors had not responded publicly as this was written. What exists is a reproducible challenge, sitting in the open, that the field will now have to resolve.

The deposit does not stand alone. A broader framework had already appeared in June, when Kreider published in PNAS with Shawn Urbanski and Joseph Fargione, asking when prescribed fire can reduce total fine-particle emissions at all. Their answer: under most current conditions, rarely. Reported reductions are “often based on unrealistic assumptions”, such as every treatment being encountered by wildfire within its effective lifespan; using empirically based encounter rates, prescribed fire increased net emissions at 99 per cent of the 73 study sites they analysed, with a median ten-year increase of 210 per cent and only 0.06 tonnes of wildfire emissions avoided per tonne of prescribed-fire emissions (Kreider et al., 2026).

Before anyone reads this as ammunition against burning, look at who is writing it and what they conclude. Kreider and Urbanski work inside the US federal fire research establishment. Fargione is The Nature Conservancy’s North America science director, and his organisation runs some of the continent’s biggest prescribed-fire programs. Their paper’s closing position is that prescribed fires “remain essential for forest management and hazard reduction”. They even flag the reason the health question stays open after the emissions question closes: burns are lit in weather chosen so smoke disperses, so more emissions do not automatically mean more exposure. This is a dispute about one number, conducted by people who all want burning to continue.

That is why the discipline of the distinction matters. The contested ground is the net emissions claim, and only that. The 16 per cent severity reduction stands unaddressed by the deposit, and the risk-to-life-and-property case never rested on smoke arithmetic in the first place. The operational record assembled in A Fire That Stopped at the Burn Linebelongs to that second category. So does new Australian evidence published the same week as the Kreider deposit: an analysis of 2,580 houses lost in NSW in the 2019-20 season found 81 per cent of destroyed homes sat within mapped flammable vegetation, and risk became minimal beyond 200 metres, validating the NSW risk mapping that land-use planning relies on (Price & Bowman, 2026). Property protection runs through distance and development controls, alongside the severity that burning itself reduces. None of those levers moved this month.

The timing for Australia is what turns an academic exchange into an operational question. AFAC’s winter outlook already has authorities watching south-eastern fuels closely, with the spring outlook due (AFAC, 2026). The NSW parks service’s enhanced program commits it to treating an average of 135,000 hectares a year, a figure that includes mechanical works alongside burning (NSW DCCEEW, n.d.). Victoria manages to a statewide residual-risk target at or below 70 per cent (Forest Fire Management Victoria, 2022). Burn plans for the spring window are being finalised while the most quoted number beneath the “burning also reduces smoke” line sits in open dispute. Importing the reversal uncritically would be as lazy as importing the original finding was: western US conifer systems are not Australian eucalypt systems, the fuels and burn windows differ, and so do the emission factors. The right response is neither panic nor dismissal. It is knowing which of your program’s public claims just moved.

Here is the strategic error worth naming: the case for hazard reduction never needed the smoke-neutrality claim, and any program that leaned its public justification on it made a choice. The most contested number in a field is the one that attracts reanalysis, and an argument that depends on it is the argument most likely to break. Smoke remains a real and rising health cost either way. Missoula modelling published this week estimates that a 2020-style smoke season recurring annually would push excess lifetime cancer risk to roughly seven times the non-smoke baseline (Jin et al., 2026), and the cardiovascular half of that ledger is the subject of The Smoke Reaches the Cardiac Ward. Agencies can hold both ideas at once: burn because the severity evidence and the operational record justify it, and treat the smoke as a cost to schedule around, never a benefit to claim.

For readers outside fire management, the transferable lesson is about evidence governance. A headline finding entered the public conversation within months of publication. The reanalysis first appeared as a code deposit within three months of publication, and fourteen months on, no journal has ruled. Between those two moments, how many strategies and business cases absorbed the number? Most organisations have no way to answer, because nobody owns the watch. Citations enter a program’s public case the way assumptions enter a financial model, except that finance reviews its assumptions every cycle and program owners almost never re-open a settled reference. Assigning someone to track the handful of claims a program’s legitimacy actually rests on costs little, and this week shows what it buys.

The counterweights complete the picture. Research on Kalunga fire practice in Brazil’s Cerrado published this week argues that restrictive zero-fire policies are eroding locally adapted systems that may help limit wildfire spread (Borges et al., 2026). Work on Texas finds that what actually constrains burning in practice is liability gaps and an ageing practitioner workforce, well before anyone reaches the emissions arithmetic (Brooke et al., 2026). Prescribed fire will survive this dispute, because the case for it stands on ground the dispute does not touch. Agencies just need their spring paperwork to stand on that same ground, with the smoke number left to the journals.

References

• AFAC. (2026, May 28). Seasonal Bushfire Outlook: Winter 2026. Australasian Fire and Emergency Service Authorities Council. https://www.afac.com.au/public-resources/seasonal-bushfire-outlook-winter-2026
• Borges, S. L., dos Santos, A. C., Silva, P. S., & Schmidt, I. B. (2026, August 5). Fire as a productive and preventive technology: Traditional knowledge and risk governance in Kalunga agropastoral mosaics. Fire Ecology. https://doi.org/10.1186/s42408-026-00522-3
• Brooke, D., Taylor, C. A., & Treadwell, M. L. (2026). Rekindling fire: The resurgence and challenges of prescribed fire culture in Texas. Rangelands. https://doi.org/10.1016/j.rala.2026.07.001
• Forest Fire Management Victoria. (2022). Victoria’s bushfire risk management report 2021-22. State of Victoria.
• Jin, L., Tan, L., Ketcherside, D. T., Selimovic, V., Nauman, K., Yokelson, R. J., & Hu, L. (2026, August 7). Characterizing emissions, chemistry, and health impacts of aged wildfire smoke in a western US city.Atmospheric Chemistry and Physics, 26. https://doi.org/10.5194/acp-26-11047-2026
• Kelp, M., Burke, M., Qiu, M., Higuera-Mendieta, I., Liu, T., & Diffenbaugh, N. S. (2025). Effect of recent prescribed burning and land management on wildfire burn severity and smoke emissions in the western United States. AGU Advances, 6(3). https://doi.org/10.1029/2025AV001682
• Kreider, M. (2026, August 6). Code and data for Kreider et al. (2026): Comment on ‘Effect of recent prescribed burning and land management on wildfire burn severity and smoke emissions in the western United States’ by Kelp et al. [Code and data deposit]. Zenodo. https://doi.org/10.5281/zenodo.21829626
• Kreider, M. R., Urbanski, S. P., & Fargione, J. (2026). Prescribed fire is unlikely to reduce net PM2.5 emissions in most locations. Proceedings of the National Academy of Sciences, 123(28). https://doi.org/10.1073/pnas.2613722123
• NSW Department of Climate Change, Energy, the Environment and Water. (n.d.). Enhanced Bushfire Management Program. https://www.environment.nsw.gov.au/topics/fire/managing-fire/bushfire-management-program
• Price, O. F., & Bowman, D. M. J. S. (2026, August 7). House loss patterns in the Australian 2019-20 fire disaster confirm the reliability of official risk mapping. International Journal of Wildland Fire. https://doi.org/10.1071/wf25197