Ozone is best known as a component of urban smog, but it is also a potent phytotoxin that costs global agriculture billions of dollars annually and suppresses the land carbon sink worldwide. Tropical forests are particularly exposed. The very traits that make them so productive, including dense canopies and high stomatal conductance, allow a large flux of ozone to enter leaves year-round. Recent experimental work has shown that tropical tree species are as susceptible to ozone damage as, or more susceptible than, many temperate species, meaning earlier studies that assumed lower sensitivity may have underestimated the problem. With the Amazon holding around 120 petagrams of carbon, even modest percentage losses in productivity carry global consequences for climate projections.

The research team employed the Joint UK Land Environment Simulator, JULES, a process-based land surface model that represents photosynthesis, stomatal behaviour, soil moisture dynamics and vegetation competition. Ozone concentrations were supplied by the UK Earth System Model, UKESM1, whose atmospheric chemistry scheme tracks the full oxidation cycles of ozone precursors including nitrogen oxides, carbon monoxide, methane and biogenic volatile organic compounds. The simulations spanned 1997 to 2024, a period punctuated by six major Amazon droughts, and were driven by observed meteorology from the CRU-JRA reanalysis, prescribed atmospheric carbon dioxide and annually varying crop and pasture fractions. By running paired simulations with and without ozone damage, the team isolated the carbon lost to the pollutant in every year of the record.

Validation of the model was a central concern. Simulated gross primary productivity reproduced the general interannual variability captured by two independent satellite products, Fluxsat and GOSIF, and the model’s tropospheric nitrogen dioxide columns matched observations from the OMI instrument on NASA’s Aura satellite, both in magnitude and spatial pattern during drought years. Because long-term surface ozone measurements in the Amazon remain scarce, the team also compared simulated variability against the TES satellite retrieval of lower-tropospheric ozone for 2005 to 2011, finding that the model reproduced interannual patterns with a coefficient of determination of 0.62, rising to 0.73 during August, the peak biomass burning season. The model had previously been evaluated against six Amazonian measurement sites and shown to capture spatial, seasonal and daily ozone variability.

The headline finding is a robust statistical relationship between fire activity and surface ozone. Across the 1997 to 2024 period, fire carbon emissions explained 72 percent of the year-to-year variance in simulated ozone concentrations, and the same linear relationship emerged independently in four additional Earth system models from the CMIP6 archive, despite their differing meteorology and chemistry schemes. Every major drought year except 2015 and 2023 showed above-average fire activity and elevated ozone, with 2024, the year of the most extreme drought and highest fire emissions in recent record, reaching unprecedented levels of both. This confirms that human-ignited fires, concentrated along the Arc of Deforestation where agricultural frontiers press into flammable degraded forest, are the dominant driver of ozone variability across the basin.

Quantifying the carbon consequences, the researchers found that the reduction in net annual carbon uptake caused by fire-related ozone damage equals 24 percent of direct fire carbon emissions, with an uncertainty range of 18 to 35 percent derived from observed variability in tropical tree ozone susceptibility. Put differently, for every tonne of carbon a fire sends directly into the atmosphere, an additional quarter of a tonne is effectively lost because ozone entering leaf stomata suppresses photosynthesis in surviving vegetation. The simulated interannual variability of this indirect loss was substantial, with a standard deviation of 38 teragrams of carbon per year and a total range of 129 teragrams across the study period. The authors caution that this should be read as an order-of-magnitude estimate, since uncertainties in stomatal representation and the ozone damage scheme itself remain unexplored.

A subtler and scientifically intriguing finding concerns the role of stomata, the microscopic pores through which plants both absorb carbon dioxide and take in ozone. Drought triggers stomatal closure, which can shield leaves from rising ozone concentrations, and rising atmospheric carbon dioxide plus warming-driven increases in vapour pressure deficit have produced a long-term declining trend in stomatal conductance across the Amazon. Sensitivity tests attributed this trend roughly equally to carbon dioxide effects and meteorological change. The result has been a gradual weakening of ozone damage over the past decade, with carbon dioxide increases alone avoiding approximately 2 teragrams of carbon loss per year. Yet this apparent protection is fragile. In 2024, record fire emissions overwhelmed the stomatal defence, producing significantly elevated ozone fluxes across the southern and south-western Amazon despite extensive stomatal limitation.

The regional anatomy of drought matters enormously. In the northern Amazon, drought produces only marginal ozone increases during the January to April dry season, and stomatal closure reduces ozone flux, yielding no statistically significant change in ozone damage. In the southern Amazon, by contrast, drought drives much stronger ozone increases during the July to October fire season, accompanied by a significant increase in ozone flux through stomata. During August, ozone damage in drought years reduced net primary productivity by an amount equal to half of the direct drought effect, and in individual years such as 2010 the ozone contribution exceeded the direct drought stress itself. The damage concentrates in the Arc of Deforestation, where almost all fires are human-initiated through clearing or agricultural escapes. The contrast between 2016, when low fire activity spared the region severe ozone damage, and 2024, when record burning erased any protection, illustrates how ignition, not drought alone, determines the outcome.

The study also raises an alarm for remote, previously untouched forest. Simulations of the 2005 and 2024 droughts show elevated ozone fluxes penetrating regions rarely exposed to pollution stress, and the authors hypothesise that vegetation lacking historical ozone exposure may prove especially sensitive. In 2024, forest disturbance from fire reached unprecedented scale even as deforestation rates declined, suggesting that degradation, not just clearance, now drives the fire-ozone-carbon cascade. Because the 2015/16 and 2023/24 droughts have both been partly attributed to anthropogenic climate forcing, and Earth system models project intensifying Amazonian drying through the twenty-first century, the authors argue that reducing deforestation must be paired with reducing forest degradation to cut drought-related fires and the ozone they generate.

The findings carry a clear message for carbon accounting and climate policy. Ozone damage is almost never included in national or global carbon budgets, yet it represents a substantial indirect emission pathway that amplifies the climate impact of every fire. The authors call for sustained ozone monitoring across the Amazon, particularly in remote regions, and for urgent development of more process-based representations of ozone effects on vegetation, including delayed damage, changes to respiration and carbon partitioning, and the sluggish stomatal responses observed in polluted air. As the climate continues to warm and droughts intensify, the invisible chemistry downwind of the flames may prove as consequential for the fate of the Amazon as the flames themselves.

Subject of Research: Fire-derived ground-level ozone damage to Amazon forest carbon uptake under extreme drought

Article Title: Fire-derived ozone intensifies carbon loss in Amazon forests under extreme droughts

Article References: Brown, F., Sitch, S., Folberth, G. A., Mercado, L. M., Cheesman, A. W., Barningham, S., Weber, J., Johnson, B., O’Sullivan, M., & Artaxo, P. (2026). Fire-derived ozone intensifies carbon loss in Amazon forests under extreme droughts. Communications Earth & Environment, 7(1), Article 790. https://doi.org/10.1038/s43247-026-04047-0

Image Credits: AI Generated

DOI: 10.1038/s43247-026-04047-0

Keywords: Amazon rainforest, ground-level ozone, wildfire emissions, drought, carbon cycle, stomatal conductance, land surface modelling, JULES, UKESM1, gross primary productivity, Arc of Deforestation, climate change