The Amazon's fires are poisoning the air the forest breathes

Burning the Amazon costs more carbon than the burning itself. That is the finding of a study published on October 7, 2026 in Communications Earth & Environment, which estimates that ozone formed in the smoke of Amazon fires damaged the leaves of surviving trees enough to reduce the forest's carbon uptake by an amount about one quarter the size of the fires' direct carbon emissions over 1997-2024 ^[1]. For every four units of carbon the flames released, in other words, the model puts the lost uptake downwind at roughly one more.

The number comes from simulations, not from a tally of carbon across the whole forest, and the authors say so ^[1]. But the direction of the result matters. Fire has always been counted as a one-time release from the trees that burn. This work suggests it also leaves a lingering injury in the trees that do not, and that the injury gets worse in exactly the dry years when the forest can least afford it.

The quiet loss

A burn scar is easy to see from orbit. What the study describes is harder to see: trees still standing, still green, doing less work.

Ground-level ozone forms when pollutants, including those from fires, react in sunlight. The gas can enter leaves and interfere with a plant's ability to take up carbon dioxide ^[1][2]. It is the same pollutant that makes city air hazardous to breathe, and it is as unwelcome in a leaf as in a lung. Flossie Brown of ETH Zurich, the paper's lead author, said fire pollution damages the leaves of surviving trees and reduces their ability to keep absorbing carbon ^[2].

That is the loop at the center of the paper. Fires release carbon directly. They also create ozone that weakens the forest's capacity to draw carbon back out of the air. A weaker sink leaves more carbon aloft, which feeds the warming that dries the forest and makes it burn more easily. The paper frames the ozone effect as a potential indirect pathway by which fires amplify carbon loss ^[1].

The idea is not entirely new. Earlier research cited in the paper estimated that fire-related ozone damage cut Amazon plant productivity by about 15%. But it had not explicitly worked out how much that mattered for atmospheric carbon dioxide compared with the fires' own emissions ^[1]. The paper says earlier work was held back by a lack of data on how susceptible tropical forests are to ozone ^[1]. This study sets the two quantities side by side, and that comparison is its contribution.

How the team did it

The researchers used a land-surface model driven by observed weather, and ran it over the Amazon for 1997 through 2024. They compared simulations with and without ozone damage, and they explored how fire emissions, weather and rising atmospheric carbon dioxide each shaped the outcome ^[1]. They also assessed parts of the model against three yardsticks: how wide the pores on leaves open, how much carbon plants take up through photosynthesis, and how much nitrogen dioxide, a pollutant gas that helps form ozone, sits in the lower atmosphere ^[1].

Those leaf pores are called stomata, and they matter here because they are the door ozone comes through. In a drought, trees close the pores to save water, which should shut some ozone out. The paper finds that simulated ozone damage nonetheless remained important during extreme drought, despite those limits on how far the pores open ^[1]. The forest cannot close the door entirely, because it still has to breathe.

The team looked at drought years in 1997/98, 2005, 2007, 2010, 2015/16 and 2023/24, and the paper identifies 2015/16 and 2023/24 among the most extreme of the period ^[1]. It associates the 2023/24 drought with the highest fire emissions in recent times ^[1]. The University of Exeter, whose researchers are among the authors, says that in the model the ozone damage during that drought was nearly twice the average for the preceding decade ^[2].

Where the damage sits

The worst ozone damage is not spread evenly. According to the University of Exeter's account of the study, it is concentrated in the Arc of Deforestation, the belt where agricultural land is expanding into rainforest ^[2]. The same account says almost all Amazon fires are started by people, through land clearance or agricultural burning that escapes into standing forest ^[2].

This is what gives the finding its practical edge. A fire started by lightning in an untouched forest is a natural hazard. A fire lit to clear pasture, or one that escapes from a field, is a decision, and it is made by someone who does not pay for the ozone. Stephen Sitch of the University of Exeter, a co-author, said that reducing deforestation and forest degradation would address both the direct fire emissions and the additional ozone-related carbon loss ^[2]. Alexander Cheesman of James Cook University, also a co-author, said air pollution, land clearing and climate change should be considered together when judging the Amazon's future ^[2].

I think the second of those remarks is the more important, and the harder to act on. Policy tends to divide the problem into departments. Deforestation sits in one, air quality in another and climate in a third. A leaf does not know which department it belongs to. The ozone that harms it is made of smoke from a land-clearing fire, and the carbon it fails to absorb ends up in the same atmosphere as everything else.

What the study does not show

The central figure is a model estimate. It is not a field measurement of the carbon taken up by every forest in the Amazon, and the paper is explicit about that ^[1]. The authors describe the ozone-damage scheme they used as simplistic. It does not capture some poorly observed plant responses, including possible ozone effects on respiration and on how plants allocate carbon, and the authors say more observations of tropical trees and further model development are needed ^[1].

There is a second limit, on how the result should be read. The paper notes that ozone is rarely included in carbon budgets or future scenarios, despite earlier estimates that rising ozone precursors could reduce the carbon tropical forests sequester today ^[1]. It presents its figure as a way to quantify an indirect loss from fire. It does not claim that every climate model leaves the process out ^[1].

Several questions remain open. No one yet knows how sensitive the one-quarter ratio is to the assumed ozone susceptibility of tropical trees. It is not settled how much of the lost uptake comes from ozone made by fires specifically, as opposed to ozone from other sources. And nobody has put a number on how much ozone damage would fall under particular cuts in clearing or escaped burns. Field measurements of ozone exposure, leaf damage and carbon uptake during and after severe droughts would be the obvious way to test the model, and the authors' call for more tropical-tree observations points the same way ^[1].

The work was funded by the Natural Environment Research Council, partly through the GW4+ Doctoral Training Partnership ^[2]. The Exeter account is an institutional release and not an independent assessment, and the study itself is the primary source for the findings above.

Why it matters anyway

None of those caveats changes the shape of the argument. The model may be crude, but the authors name processes it leaves out, and they do not say those would shrink the effect ^[1]. Even a rough figure of a quarter makes the point that the accounting convention, which counts what burns and stops there, understates what fire does to the forest.

That matters because the Amazon is one of the planet's great carbon stores, and its role as a sink is part of what holds the climate where it is. A forest that is merely wounded around the edges can be argued over. A forest whose surviving trees take up less carbon in the worst years is a different kind of problem. It means the loss is not finished when the smoke clears.

There is also a plain human reason to care. Ozone harms people as well as plants. The measures that would spare the leaves, keeping fire out of the forest and slowing clearing, would also spare lungs downwind. That is a rare case where the carbon argument and the public-health argument point to the same place.

What to watch

The nearest test is the weather. The University of Exeter's October 7 account said forecasters saw a strong El Niño developing that year, with the potential to become one of the most intense on record, and it attributed concern about a possible repeat of the 2023/24 damage to the study's authors ^[2]. That is a warning from the researchers, not a forecast that the Amazon will burn again. How strong the El Niño proves to be, and what it means for Amazon drought and fire, is the first thing to follow.

If a drought arrives and the fires follow, the Arc of Deforestation is where this study predicts the ozone will bite hardest. The next step is for the field measurements to catch up with the model. If they confirm it, the carbon budgets that govern how much the world thinks it can emit will have to carry a line they mostly lack today.