Amazon carbon emissions driven by human actions amplified by droughts

Amazonia, the largest tropical forest in the world, has lost substantial carbon stocks due to large-scale deforestation and other land-use changes. Remaining forests in the region have offset some of this carbon loss through carbon sequestration processes. However, this capacity is declining, primarily because of increased tree mortality^6,7,8,18 and changes in regional climatic conditions modified by deforestation^1,3,9,13. The Amazon forest stores approximately 123 ± 23 billion tonnes of carbon (PgC) in vegetation and soils^19,20. The interactions among the forest ecosystem, hydrological cycle and regional climate are complex, with atmospheric moisture recycling by evapotranspiration accounting for up to half of precipitation¹⁶. Direct human activities within Amazonia, including deforestation, forest degradation and biomass burning, have profoundly altered the carbon balance of the region and climate¹¹. Forest loss reduces precipitation, raises temperatures and lengthens the dry season, delaying the onset of the wet season^1,3,4,14,16. During periods of extreme climatic variability, such as El Niño or anomalously high Tropical North Atlantic (TNA) sea-surface temperatures (SSTs), these impacts are amplified through the interaction of accumulated forest loss, global warming, changing circulation and large-scale anomalies^9,14,16.

Many questions remain about the main drivers of carbon emissions and variability in Amazonia. Studies based on modelling and bottom-up techniques show increased emissions during extreme climate (warm and dry) years, pointing to extreme events as the reason for increased carbon emissions. However, our direct atmospheric measurements show a more complex situation. Here, we use our 15-year history of Amazonian carbon fluxes derived from atmospheric CO₂ measurements to answer important questions about the main factors driving increased carbon emissions and variability in the Amazon and whether they are caused by global extreme climate events or direct human activities (deforestation, biomass burning and degradation) in the Amazon.

In 2000, deforestation in the Brazilian Amazon represented a total area of 474,770 km² (12.5% of the Brazilian Amazon, 3.8 × 10⁶ km²). The Brazilian Amazon represents 58% of the Pan-Amazonia region (Fig. 1). By 2004, the PRODES programme²¹ of Brazil for monitoring native vegetation had recorded cumulative deforestation of 612,225 km² across the Legal Amazon, equivalent to 15.5% of the original forest cover of the region. Between 2004 and 2012, the annual rate of deforestation fell 84% as a result of public policies designed to curb clearing^22,23, reaching a cumulative loss of 697,319 km² (18.4% of the total forest area). The rate began to rise again from 2013 to 2018, when total forest loss reached 19.5%, and accelerated sharply after 2019, reaching 791,835 km² (20.9%) in 2022. Although the annual rate of deforestation declined strongly in 2023 and 2024 (30% and 47%, respectively) relative to the 2019–2022 period, cumulative forest loss still reached 807,301 km² (21.3%). From 2010 to 2024, pasture area expanded by 19%, soybean plantation area by 193% and mining by 272% (ref. ²⁴) to estimated values of 561,416 km², 59,368 km² and 3,956 km², respectively (Supplementary Fig. 1).

We present precipitation and temperature time series from 1979 to 2024 in Extended Data Fig. 1. To assess the impact of the past six years, during which deforestation accelerated sharply between 2019 and 2022, on precipitation and temperature, we compared the previously studied period¹ (1979–2018) with the current period from 1979 to 2024, using the same database for both periods (Methods). From 1979 to 2018, the changes in annual precipitation were not statistically significant (27 ± 69 mm); however, from 1979 to 2024, there was a reduction of 174 ± 69 mm, representing an 8% decrease. During the dry-season peak period (August–October), from 1979 to 2018, precipitation decreased by 17% (64 ± 19 mm), and from 1979 to 2024, by 20% (83 ± 18 mm). The increase in annual temperature from 1979 to 2018 was 0.93 ± 0.13 °C, and from 1979 to 2024 was 1.13 ± 0.13 °C, representing a 0.20 °C increase over 6 years. During the dry-season peak period, the increase in temperature was 1.51 ± 0.17 °C from 1979 to 2018 and 1.81 ± 0.18 °C from 1979 to 2024, representing a 0.30 °C increase over 6 years. Between 2018 and 2024, Amazonia lost an additional 60,964 km² of primary forest and 564,532 km² of burned area, representing a significant increase in human activity impacting the region since 2019. Further discussion of changes in climatic conditions in Amazonia and the studied regions is provided in Extended Data Fig. 1, Extended Data Table 1 and Supplementary Discussion 1.

Regional analyses show that the most pronounced temperature increase occurred in the southeast region (ALF), with warming of 2.19 ± 0.21 °C during the dry-season peak and 1.37 ± 0.15 °C in the annual mean between 1979 and 2024. From 2018 to 2024, this region lost 37,930 km² (62% of the total deforestation of the Brazilian Amazon) of primary forest. The most intense decline in precipitation happened in the northeast region (SAN). From 1979 to 2024, annual precipitation decreased by 304 ± 150 mm (12%), and during the dry-season peak, it declined by 37% (77 ± 23 mm). Over the same six-year period, SAN lost 15,023 km² (25% of the total deforestation) of primary forest. More details are presented in Extended Data Fig. 1, Extended Data Table 1 and Supplementary Discussion 3.

This study includes 1,017 vertical profiles conducted between 2010 and 2024 at four Amazonian sites, whose upwind areas of influence represent nearly the entirety of Amazonia (Fig. 1 and Extended Data Fig. 2). Vertical profiles were typically performed twice per month, from 4.4 km above sea level (a.s.l.) to near the surface (about 0.3 km a.s.l.), sampling air at 12 heights in a descending helical trajectory at each of the four sites between 12:00 and 13:00 local time. Air samples were analysed for CO₂ and CO at the Greenhouse Gas Laboratory (LaGEE/INPE), São José dos Campos, Brazil. The Amazon exhibits a dominant lower-tropospheric circulation pattern, with air masses originating over the tropical Atlantic Ocean, crossing the Amazon towards the Andes, then to the southern regions, and returning to the Atlantic in the southern regions. This circulation enables integration of both natural and anthropogenic processes by computing atmospheric mass balance across vertical profiles of sites. Here, we present Amazonian annual mean CO₂ vertical profiles (VP; Fig. 2a), and the profiles differenced from the CO₂ background (∆VP; Fig. 2b) for 2010–2024, calculated from the annual mean profiles at each site (Extended Data Fig. 3), using the same weighting methodology used for Amazonia carbon balance calculations^1,13 (Methods). Most years show a positive ∆VP near the surface (Fig. 2b), indicating that Amazonia is a net carbon source rather than a sink and cannot offset regional anthropogenic emissions.

Carbon fluxes were calculated using the atmospheric column budget technique¹ for each vertical profile and site over a 15-year study period. Total carbon flux  (TCF), fire carbon flux (FCF) and net biome exchange (NBE) monthly mean carbon fluxes (TCF subtracted by FCF = NBE) for each site were derived from the two vertical profiles collected per month, and Amazonia-wide monthly and annual mean fluxes were computed using a weighted mean approach that accounts for regional contributions^1,13 (Methods). Figure 3 shows the Amazon carbon balance over 15 years, alongside some climatic and anthropogenic drivers. Examining the annual mean vertical profiles (Fig. 2) and the annual carbon balance (Fig. 3, blue bars) shows strong variability over the 15-year period.

To better understand the variability in Amazonia annual carbon balance, we use k-means clustering analysis to group years, taking into account the following parameters: annual carbon balance (TCF), precipitation, temperature, deforestation (annual increment and accumulated), burned area (integrating all vegetation types)²⁵, fire spots, Oceanic Niño Index²⁶ (ONI) and TNA warming²⁷. The cluster analysis identified the 15 years as belonging to three clusters: group 1 (2011, 2012, 2013, 2014, 2017 and 2018); group 2 (2019, 2020, 2021 and 2022); and group 3 (2010, 2015, 2016, 2023 and 2024). See the statistical analysis in Methods, Extended Data Fig. 4, Extended Data Tables 2 and 3, Supplementary Discussion 2 and Supplementary Table 1. Group 1 is characterized by wetter years and low human activities. Group 2 reflects a period with intense human activity—deforestation and degradation (biomass-burning, selective logging, mining and edge effects), and low climatic stress. Group 3 comprises years characterized by intense global climate events, such as El Niño and TNA events with associated biomass burning.

Group 1 (Fig. 2c, Extended Data Table 3a and Extended Data Fig. 5), representing the wettest years with low levels of human disturbance (deforestation and degradation), shows a near-neutral annual mean total carbon flux (TCF) of 0.11 ± 0.08 PgC yr⁻¹ and ∆VP (VP – BKG) of 0.14 ppm, resulting from a carbon sink (−0.20 ± 0.12 PgC yr⁻¹) that offsets 64% of fire emissions (0.32 ± 0.07 PgC yr⁻¹). The annual deforestation and burned area were 6,389 km² and 55,115 km², respectively. This result indicates that Amazonia did not significantly contribute to the global carbon balance during wet years and with low human disturbance.

Group 2 years (Figs. 2c and 3, Extended Data Table 3b and Extended Data Fig. 5) were characterized by a period with higher levels of human disturbance and weak governance¹³. During this period, the annual deforestation rate rose by 97%, whereas the burned area increased by 25%, resulting in a TCF of 0.43 ± 0.09 PgC yr⁻¹ and a ∆VP of 0.53 ppm, representing increases of 291% and 379% relative to group 1, respectively. The annual mean temperature was the same as in group 1, and precipitation was 134 mm (6%) lower (mainly due to the 2020 wet season), indicating low climatic stress during this period. In August 2020, a strong and prolonged La Niña event began and lasted until February 2023 (Fig. 4). During this period, carbon uptake gradually increased in the first quarter of each year (Fig. 4), offsetting an ever-greater proportion of anthropogenic emissions. As climate stress was very low during this period, it mainly represents the carbon emissions from human activity in Amazonia.

Group 3 (Extended Data Table 3c and Extended Data Fig. 5) comprises years characterized by intense global climate events, such as El Niño and TNA warming. The mean ONI and TNA values for group 3 were higher than those for groups 1 and 2. Any years from groups 1 and 2 presented El Niño and/or warming TNA events. The El Niño events occurred in 2015–2016 (strong–very strong) and 2023 (strong; Fig. 3c), and the warming TNA events occurred in 2010 (moderate) and 2023–2024 (moderate–strong). We used the same ONI classification criterion²⁸ for the TNA events (Methods). We also observed in Fig. 2c that the group 3 vertical profile showed an increase in the free-troposphere layer, unlike the other two; this is due to convective transport induced by the fire. Both El Niño and TNA warming contribute to reduced precipitation and increased temperatures in Amazonia. On average, group 3 received 308 mm (13%) less annual precipitation than group 1, with 20% less during the wet-season peak (January–March), and 17% less during the dry-season peak (August–October). The annual mean, wet-season peak and dry-season peak temperatures were higher than those of group 1 by 0.5 °C, 0.4 °C and 0.8 °C, respectively. These drought and warming conditions in Amazonia increased the flammability of the forest. We observed a 92% increase in burned area²⁵ compared with group 1, whereas deforestation²¹ was 22% higher. TCF was 0.49 ± 0.07 PgC y⁻¹, and ∆VP was 0.61 ppm, both of which were 325% higher than Group 1. However, for this group, NBE was −0.06 ± 0.10 PgC y⁻¹ (nearly neutral sink), indicating that the main source of carbon is not the forest itself, but, again, human influence that promotes fires. Furthermore, recent climatic conditions interfere with the consequences of an intense El Niño year on the forest, as can be observed in 2023 and 2024 (Extended Data Tables 2 and 3 and Extended Data Fig. 7). FCF was 0.55 ± 0.10 PgC yr⁻¹, 72% higher than group 1 and 53% higher than group 2, showing that fire emissions are an important characteristic of group 3. This finding highlighted the importance of clarifying the origin of fires, whether natural or anthropogenic. Therefore, a study was conducted to identify the relationship between lightning and forest fires in group 3. During the years of this group, it was found that lightning-induced ignition for fire events represented 1.6% and 1.0%, respectively, for the years 2023 and 2024 (Methods and Extended Data Fig. 6), allowing us to conclude that the emissions from group 3 originated primarily from human-induced fires, which were amplified by drought conditions. The years 2023 and 2024 were chosen to ensure gathering a more accurate cloud-to-ground lightning dataset, as higher-quality and more reliable data are only available from 2017 onwards.

These findings underscore the dominant role of direct human disturbance in Amazonia (Fig. 5, a schematic grouping summary), driving year-to-year changes in carbon fluxes and highlighting the urgent need for mitigation to prevent the approach of the region to a climatic tipping point^29,30. Climatic stress amplifies the effects of human-induced fire, because drought conditions substantially increase forest flammability^{1,9,10,11,13,15,31,32,33}.

The 2023 wet season exhibited the largest carbon sink over our study period of 15 years (Fig. 4 and Extended Data Fig. 7), which occurred after three consecutive La Niña years (Fig. 4). The 2023 Amazonia carbon balance was similar to that of group 1 (0.11 ± 0.09 PgC yr⁻¹), because the wet-season sink offset 75% of the carbon emissions from deforestation and burned area, which were, respectively, 38% and 52% higher than in group 1. The annual mean temperature was similar to that of group 1, starting colder than normal and warming progressively as strong TNA and El Niño events became unprecedented drought and heat wave^34,35 (Fig. 4).

While 2023 began after 3 years of La Niña, 2024 began in the middle of strong El Niño and TNA events, which made its outcome completely different. The year 2024 exhibited the highest carbon emissions (0.81 ± 0.09 PgC yr⁻¹) in the 15-year record (Extended Data Tables 2 and 3 and Extended Data Fig. 7), 605% higher than that of group 1. It was also the hottest, driest and most strongly degraded by fire over the past two decades³⁶. Annual temperatures were 1.2 °C higher than in group 1, and annual precipitation was 462 mm lower than in group 1. The burned area was 182% higher than in group 1, producing the most extensive burned area observed since 2010 in Amazonia (Fig. 3a and Extended Data Tables 2 and 3). Following the reinstatement of public policies to curb deforestation in 2023 (ref. ¹³), fire became a widespread tool for deforestation. In 2024, PRODES identified that 38% deforestation was caused by fire³⁷. Most Amazonian wildfires are intentionally set, and severe droughts exacerbate their extent and intensity^15,29,31,37.

More detailed subregional (RBA, ALF, SAN, TTC) analyses are provided in Extended Data Tables 2 and 3, Extended Data Figs. 2 and 3, Supplementary Discussion 3 and Supplementary Figs. 3–5.

Sites, air sampling and analysis

Here, we report CO₂ and CO measurement results from the four Amazonian aircraft vertical profile sites of the CARBAM project³⁸ (SAN: 2.86° S 54.95° W; ALF: 8.80° S 56.75° W; RBA: 9.38° S 67.62° W; in 2010–2012 for TAB: 5.96° S 70.06° W; and since 2013 for TEF: 3.39° S 65.60° W, and since 2022 for CZS: 7.00° S 72.50° W), detailed in refs. ^1,13. The regions of influence for the four sites representing the northeast (SAN), southeast (ALF), southwest (RBA) and northwest Amazon region (TAB, 2010–2012; TEF, 2013–2021; CZS, 2022–2024) are shown in Extended Data Fig. 2c–h. Our samples were typically collected twice per month, totalling 1,017 vertical profiles over 15 years, in a descending spiral from 4,420 m to about 300 m a.s.l. In 2015, the vertical profile sampling was stopped in April at all sites and resumed in November at RBA. In 2016, vertical profiles were performed only at RBA and ALF. The vertical profiles were usually taken between 12:00 and 13:00 local time. Air samples were analysed by a non-dispersive infrared analyser for CO₂ and by gas chromatography with HgO reduction detection for CO. The detailed analytical and sampling methods were presented in previous studies^1,13.

Amazonia carbon study area

We defined the Amazonia study area according to subregions from ref. ³⁹ and biomes from ref. ⁴⁰, in which the studied area in Amazonia was determined to include only the forest ecosystems subregions: Amazonia stricto sensu, Guianas and Gurupi, with a total area of 6,700,798 km².

Annual mean vertical profiles

The annual mean VP for each site was calculated by first producing the monthly mean for each height and then the annual mean for each height. The annual mean ∆VP for each site and for Amazonia was calculated by subtracting the corresponding background (BKG) for each VP and each height (Fig. 2 and Extended Data Fig. 3). To calculate the Amazonia annual mean VP, we applied the same method used to obtain the Amazonia annual mean flux and balance. To scale to the whole of Amazonia, we divided it into three regions (see ref. ¹³ and Supplementary Fig. 5). To compose the VP for region 1 (SAN + ALF), we calculated the weighted mean CO₂ concentration at each height, proportional to the area of each region. To calculate the VP for region 2 (RBA + TAB, 2010–2012; RBA + TEF, 2013–2021; RBA + CZS, 2022–2024), the same procedure was used as in region 1. And for VP of region 3, the remaining Amazonia area not covered by the regions of influence of the vertical profile, we used the same VP as region 2. To compose the VP for Amazonia, the weighted mean for each height CO₂ concentration was calculated, considering the three regions for each height (Supplementary Fig. 5).

Estimation of carbon fluxes

We used a column budget technique to estimate TCF, FCF and NBE, which consists of the difference between CO₂ mole fraction measured in the vertical profile and the estimated background mole fraction⁴¹ (∆CO₂), considering the travel time of air parcels along the trajectory from the coast to the site (equation (1)), following the methodology in refs. ^1,13,42.

Background

Background mole fractions were estimated according to the methodology described in ref. ⁴¹, using the geographical position of each air-mass back-trajectory from the HYSPLIT model⁴³ when it intersects two virtual limits: (1) a latitude limit, from the Equator southwards at 30° W and (2) a line between two observation sites at Ragged Point, Barbados (RPB), and Ascension Island, the United Kingdom (ASC), from NOAA/GML. We used CO₂ (ref. ⁴⁴) and CO (ref. ⁴⁵) concentrations from the NOAA/GML Carbon Cycle Cooperative Global Air Sampling Network to determine the background.

Carbon fire fluxes

We calculated fire emissions according to the methods published in refs. ^1,13. We used observed CO/CO₂ ratios by site from 2010 to 2024 (CO/CO₂, ppb CO per ppm CO₂) from measured CO mole fractions. The CO/CO₂ ratios were calculated from clearly identifiable plumes in the vertical profiles during the dry season along period study (39 VP ALF CO/CO₂ = 60.7 ± 30.7 ppb ppm⁻¹; 35 VP SAN CO/CO₂ = 56.1 ± 20.4 ppb ppm⁻¹; 44 VP RBA CO/CO₂ = 67.7 ± 24.1 ppb ppm⁻¹; and 19 VP TTC (TAB_TEF_CZS) CO/CO₂ = 74.2 ± 29.1 ppb ppm⁻¹), which represents how many ppb CO is emitted for each 1 ppm CO₂ during burning process. To calculate CO₂ from fire, we used the CO flux and subtracted it from natural CO emissions (equation (2)). After the step \({F}_{\mathrm{CO}}-{F}_{\mathrm{CO}}^{\mathrm{Natu}{\rm{r}}\mathrm{al}}\), we used the CO/CO₂ ratio to estimate the amount of CO₂ from fire. NBE represents all processes from natural sources and sinks and all anthropogenic non-fire emissions (or after fire, without CO emissions) (NBE = total − fire).

Flux uncertainty analysis by Monte Carlo error propagation

Flux uncertainties were estimated using Monte Carlo error propagation (1,000 iterations). CO₂ measurement uncertainty was 0.03 ppm, and background (BKG) uncertainties were derived from residuals to the fitted CO₂ smooth curve, according to ref. ¹. Uncertainties in back-trajectory travel times¹³ were also included. For carbon fire emissions and NBE, uncertainties in CO/CO₂ ratios, CO fluxes and a 25% uncertainty in natural CO emissions were considered. VP flux uncertainties were derived from the Monte Carlo ensemble. Monthly uncertainties were estimated using pooled standard deviation to account for intra-month variability within the annual mean flux, and period uncertainties were calculated using the quadratic mean.

An additional source of uncertainty is the limitation on sampling to 4.4 km a.s.l. Transit times of air parcels from the coast to sampling sites, based on HYSPLIT, can vary, with mean trajectory times of 2.6 days for SAN, 5.3 days for ALF, 6.6 days for RBA, 5.7 days for TAB, 5.0 days for TEF and 7.6 days for CZS. The convective process, together with long travel times of air masses, can represent loss of carbon sources and sink contributions from the surface, because our vertical profiles start at 4.4 km. Comparing the background concentration and the top of vertical profiles is one way to verify potential information loss^1,13. According to the method we use, the flux is obtained by subtracting the measured CO₂ mole fraction in the VP from the background concentration, accounting for the integration time (equation (1)). During the dry season, the loss of information is larger to positive (VP > BKG), because during the burning season (peak of the dry season), the top of VP starts with a higher CO₂ concentration than BKG due to convective processes driven by biomass burning.

A previous study⁴⁶ proposed a correction of our gas concentration measurements based on a comparison of laboratory analyses of samples collected on 12 vertical profiling flights above Manaus in 2017–2019. The in situ measurements were made with a cavity ring-down spectrometer (CRDS), and discrete flask samples (PFPs) collected on the same flights were analysed in the laboratory. Some of the continuous CRDS measurements differed from those determined in the laboratory from flask samples (PFP) collected on the same flights (Figure A4 in ref. ⁴⁶). Based on linear regression of N = 158 paired CRDS-PFP measurements, ref. ⁴⁶ reported that water vapour content explained only 13% of the variance in the CRDS-PFP differences (Figure A7 in ref. ⁴⁶). Nevertheless, ref. ⁴⁶ proposed applying this regression relationship to adjust all laboratory measurements. Applying such an adjustment would conflate multiple methodological differences between the two sampling and analytical systems. The regression in ref. ⁴⁶ compares independent instruments sampling distinct air parcels under different pressures, flow rates and materials. The proposed correction assumes that the in situ CRDS measurements are accurate, that the laboratory analyses are biased, that discrepancies arise solely from water vapour, and that the regression relationship can be extrapolated to approximately 13,319 laboratory measurements collected over 15 years. For these reasons, the proposed adjustment was not applied in this study. LaGEE measurements used in this study are internally consistent across the full temporal and spatial domain.

An additional source of uncertainty is addressed in the calculation of fire carbon emissions and NBE. Over time, the forest is becoming drier and hotter. These changes promote a lower CO/CO₂ ratio. The use of a single mean CO/CO₂ ratio is another source of uncertainty, leading to lower fire emissions than those observed in recent years because the ratio tends to decline over time. It is likely that carbon emissions from fires were higher than our averages for 2023 and 2024 indicate, because both were drought years.

Missing data imputation

We used the missForest algorithm to impute missing monthly TCF and FCF data at the ALF, SAN, RBA and TAB_TEF_CZS (TTC) sites. These gaps resulted from sampling and logistical issues in the laboratory. The missForest method is a non-parametric imputation approach based on random forests^47,48, and we implemented it in R language⁴⁹ through the missForest package⁵⁰. To estimate missing values for TCF and FCF^1,13, we included the following monthly variables in the imputation process: temperature (T), precipitation (P), vapour pressure deficit (VPD), ONI, TNA index, EVI, burned area (BA) and fire spots (FS). The missing data gaps were imputed using average values calculated from 1,000 runs of missForest. The normalized root mean square errors obtained for FC_Total and FC_Fire were 0.0014 and 0.0012 for ALF, 0.0011 and 0.0010 for RBA, 0.0055 and 0.0053 for SAN and 0.0018 and 0.0016 for TTC, respectively. NBE missing data were calculated by subtracting the fire from TCF.

Regions of influence

We define regions of influence as those areas covered by the density of back-trajectories⁵¹. We updated the calculations for the regions of influence used in our published CO₂ papers¹. In this study, the regions received two treatment steps. In the first step, we calculated back trajectories for altitudes lower than 1,300 m. As we have only four heights below 1,300 m, the density of trajectories could be reduced, compromising the representativeness. To better characterize atmospheric circulation over the quarters and years, we calculated trajectories at each sampling height (>1,300 m) twice a week, at the same locations and time of day, over the 15 years for each VP sampling site. A second treatment in the grid cells was done to increase the weight for the grid cells near the sample site. We normalized the weight for the region of influence to be from 0 to 1. First, the grid with a value of 1 was calculated by disregarding 2.5% of the grids with the highest values, thereby increasing the area of the grid region with a value of 1. It was considered a regular grid of 1° × 1° generated in the WGS 84 georeference system (EPSG:4326).

Precipitation, temperature, VPD and EVI data

We used the quarterly regions of influence maps as spatial weighting functions for all studied parameters to determine how each parameter influenced the carbon flux and for the missing data imputation process (missForest), following ref. ⁴. We used the GPCP database (http://eagle1.umd.edu/GPCP_ICDR/GPCP_Monthly.html), v.1.3, for precipitation analysis (described in refs. ^52,53,54), which contains daily data at a 1° × 1° latitude–longitude resolution. We used 2-m temperatures from ERA-5 that are monthly means of daily means since 1959, and which were used with a resolution of 0.25° × 0.25° latitude–longitude, obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF)^27,55. The VPD product is a measure of the indirect vapour pressure deficit in kPa (resolution of 2.5 arcmin) of monthly means of temperature and humidity, provided by the Climatic Research Unit Ts4.0 (ref. ⁵⁶). The dataset was resampled to a 1° × 1° spatial resolution using the monthly mean. The EVI is a vegetation index that aims to estimate the fraction of photosynthetically active radiation (fPAR) from terrestrial vegetation. The EVI product used⁵⁷ was the MANVI: Moderate Resolution Imaging Spectroradiometer (MODIS) multiangle implementation of atmospheric correction nadir-solar (see the methods in ref. ¹³).

Burned area

Two products of burned area (BA) were used: (1) for all annual burned areas, the BA from MapBiomas was used, integrating all types of vegetation that were burned (primary and secondary forest, pasture, agriculture and others), from the MapBiomas Fire Collection 4 (ref. ²⁵; https://brasil.mapbiomas.org). MapBiomas Fire was validated with a global accuracy of 89% (ref. ³³); (2) for monthly BA (used only for missForest calculation), MODIS products were used: MODIS Collection 6 MCD64A1 burned area product^13,58.

Deforestation

The procedures to retrieve deforestation as geographic data built by PRODES/INPE²¹ based on a historical series of Landsat images provide deforestation annual increments in the Brazilian Amazon. Detailed information on the PRODES methodology is available and can be accessed at http://urlib.net/rep/8JMKD3MGP3W34R/443H3RE (ref. ⁵⁹). The dataset accuracy is 93.5 (ref. ⁶⁰). We used the data period from 2010 to 2024. Using QGIS software, a 0.25° × 0.25° grid was generated for the entire Brazilian Amazon, which was filled with the absolute values of deforested area calculated for each cell and each year of the series. The mean deforestation area was calculated for the period 2010–2024 within each grid cell. Annual deforestation data correspond to the deforested area mapped by the PRODES project in a given year for Amazon forests.

Monthly deforestation and degradation

Monthly deforestation and degradation data correspond to either the deforestation alerts (clear cut, deforestation with vegetation, and mining) or degradation alerts (ordered selective logging, disordered selective logging, burns and degradation) captured by the DETER⁶¹ Amazon system in a given month. This dataset is produced daily by the visual interpretation of medium spatial resolution optical imagery of National Institute for Space Research (INPE) and, although not a full estimate of deforestation and degradation in the Amazon, can indicate variations in deforestation patterns over time^33,62. We used the raw version of the dataset provided by INPE that included both password-protected data and data collected from August 2015.

Fire spots

Fire spots in Pan-Amazonia between 2010 and 2024 and burned area in Brazilian Amazon were retrieved from the BDQueimadas wildfire monitoring program of INPE⁶³ (see the methods in ref. ¹³).

ONI

Warm (El Niño) and cold (La Niña) periods based on a threshold of ±0.5 °C for the ONI, 3-month running mean of ERSST v.5 SST anomalies in the Niño 3.4 region (5° N–5° S, 120° W–170° W), based on centred 30-year base periods updated every 5 years (ref. ²⁶). To analyse the intensity of the events, we used the ONI classification anomaly²⁸—for El Niño events, weak (0.5–0.9 °C), moderate (1.0–1.4 °C), strong (1.5–1.9 °C), and very strong (≥2.0 °C).

TNA index

The TNA index was computed as the mean sea-surface temperature (SST) anomaly over the region 23.5° N–5.5° N and 57.5° W–15° W (ref. ⁶⁴). Monthly SST fields were obtained from ERA5 (0.25°), and anomalies were derived relative to a 30-year climatological baseline (1971–2000), consistent with standard TNA calculations²⁷. We used the same ONI anomaly classification to analyse the intensity of the TNA warming events.

Statistical analysis

Annual data (2010–2024) for the Amazon included TCF, precipitation (P), temperature (T), ONI, TNA index, fire spots (FS), burned area (BA), annual deforestation (DEF) and accumulated deforestation (DEF_acc). Variable distributions and homogeneity of variance were assessed using the Shapiro–Wilk and Levene’s tests (Supplementary Table 1), respectively. Although most variables met normality assumptions, robust statistical approaches were adopted because of the small sample size (n = 15) and potential deviations from Gaussian behaviour typical of environmental datasets (Supplementary Discussion 3). Principal component analysis was applied to standardized data, with the first two components considered to represent the dominant variability. The k-means clustering (k = 3) was performed in the principal component analysis space, with cluster characteristics summarized in a standardized heatmap. The optimal number of clusters was determined using the average silhouette width (Extended Data Fig. 4).

Group differences among clusters were assessed using Welch’s analysis of variance (Supplementary Table 1), which is robust to heteroscedasticity and unequal sample sizes. When significant differences were detected (P < 0.05), pairwise comparisons were conducted using the Games–Howell post hoc test (Supplementary Table 1), which does not assume equal variances and appropriately controls the type I error rate under these conditions. This approach provides a consistent and reliable framework for identifying differences among clusters in environmental variables (Supplementary Discussion 2). All analyses were conducted in R language⁴⁹.

Amazonia pasture, soybean and mining area

We obtained annual estimates of pasture, soybean cropland and mining area across Amazonia from the MapBiomas Collection 10²⁴ dataset (30 m spatial resolution). Annual land-cover maps were used to isolate class-specific pixels (pasture, soybean and mining) and to compute their corresponding areas (km²) within the Amazonia carbon study area. This approach provided a consistent time series of land-use expansion and persistence for 2000–2024.

Lightning-associated wildfires

After 2017, in group 3, the years with accurate data were 2023 and 2024, when a more accurate cloud-to-ground lightning dataset was available. We used cloud-to-ground strokes from the Earth Networks Total Lightning Network (ENTLN)⁶⁵. In South America, particularly in Amazonia, ENTLN lacks a dense sensor network comparable to that in other regions of Brazil. Even so, it operates with a stroke detection efficiency above 50% (corresponding to a flash detection efficiency above 80%) and a mean location accuracy below 1 km. Just for reference, in regions with high sensor densities, after the 2021 upgrade, the ENTLN reaches cloud-to-ground flash detection efficiency levels above 95% and median location accuracy below 100 m (ref. ⁶⁶). Although performance depends on regional sensor coverage, ENTLN is considered one of the most advanced and reliable lightning networks globally⁶⁷. Active fires used in this study were provided by the Visible Infrared Imaging Radiometer Suite on the Suomi National Polar-Orbiting Partnership satellite, with a spatial resolution of 375 m. The Visible Infrared Imaging Radiometer Suite 375 m active fire product provides geolocation accuracy of the order of one pixel (about 375 m), with typical sub-pixel geolocation uncertainties (about 70 m), ensuring reliable spatial representation of fire locations. All overpasses, day and night, were used from the Wildfire Monitoring Program of the INPE, available at http://www.inpe.br/queimadas/ (ref. ⁶³). To assess lightning-associated wildfire among all cloud-to-ground lightning strokes, we identified probable lightning candidates that occurred up to 72 h (3 days) before active fire detection, using a fixed buffer radius of 1 km around each active fire and accounting for the location accuracy of both datasets. A detailed description of lightning candidate characteristics in Brazil is provided in ref. ⁶⁸.

1. Gatti, L. V. et al. Amazonia as a carbon source linked to deforestation and climate change. Nature 595, 388–393 (2021).

   Article  ADS  CAS  PubMed  Google Scholar 

2. Leite-Filho, A. T., Costa, M. H. & Fu, R. The southern Amazon rainy season: the role of deforestation and its interactions with large-scale mechanisms. Int. J. Climatol. 40, 2328–2341 (2020).

   Article  Google Scholar 

3. Leite-Filho, A. T., Soares-Filho, B. S., Davis, J. L., Abrahão, G. M. & Börner, J. Deforestation reduces rainfall and agricultural revenues in the Brazilian Amazon. Nat. Commun. 12, 2591 (2021).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

4. Fu, R. et al. Increased dry-season length over southern Amazonia in recent decades and its implication for future climate projection. Proc. Natl Acad. Sci. USA 110, 18110–18115 (2013).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

5. Franco, M. A. et al. How climate change and deforestation interact in the transformation of the Amazon rainforest. Nat. Commun. 16, 7944 (2025).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

6. Brienen, R. J. W. et al. Long-term decline of the Amazon carbon sink. Nature 519, 344–348 (2015).

   Article  ADS  CAS  PubMed  Google Scholar 

7. Phillips, O. L. & Brienen, R. J. W. Carbon uptake by mature Amazon forests has mitigated Amazon nations’ carbon emissions. Carbon Balance Manage. 12, 1 (2017).

   Article  Google Scholar 

8. Hubau, W. et al. Asynchronous carbon sink saturation in African and Amazonian tropical forests. Nature 579, 80–87 (2020).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

9. Feron, S. et al. South America is becoming warmer, drier, and more flammable. Commun. Earth Environ. 5, 501 (2024).

   Article  ADS  Google Scholar 

10. Staal, A. et al. Hysteresis of tropical forests in the 21st century. Nat. Commun. 11, 4978 (2020).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

11. Longo, M. et al. Degradation and deforestation increase the sensitivity of the Amazon Forest to climate extremes. Environ. Res. Lett. 20, 054024 (2025).

    Article  Google Scholar 

12. Doughty, C. E. et al. Tropical forests are approaching critical temperature thresholds. Nature 621, 105–111 (2023).

    Article  ADS  CAS  PubMed  Google Scholar 

13. Gatti, L. V. et al. Increased Amazon carbon emissions mainly from decline in law enforcement. Nature 621, 318–323 (2023).

    Article  ADS  CAS  PubMed  Google Scholar 

14. Marengo, J. A. Long-term variability, extremes and changes in temperature and hydrometeorology in the Amazon region: a review. Acta Amaz. 54, e54es22098 (2024).

    Article  Google Scholar 

15. Aragão, L. E. O. C. et al. 21st Century drought-related fires counteract the decline of Amazon deforestation carbon emissions. Nat. Commun. 9, 536 (2018).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

16. Costa, M. H. et al. The physical hydroclimate system of the Amazon. In Amazon Assessment Report 2021 (eds Nobre, C. et al.) Ch. 5 https://doi.org/10.55161/HTSD9250 (UN Sustainable Development Solutions Network, 2021).

17. Costa, M. H. et al. Biogeophysical cycles: water recycling, climate regulation. In Amazon Assessment Report 2021 (eds Nobre, C. et al.) Ch. 7 https://doi.org/10.55161/KKHX1998 (UN Sustainable Development Solutions Network, 2021).

18. Esquivel-Muelbert, A. et al. Tree mode of death and mortality risk factors across Amazon forests. Nat. Commun. 11, 5515 (2020).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

19. Malhi, Y. et al. Biogeochemical cycles in the Amazon. In Amazon Assessment Report 2021 (eds Nobre, C. et al.) Ch 6 https://doi.org/10.55161/TAKR3454 (UN Sustainable Development Solutions Network, 2021).

20. Malhi, Y. et al. Comprehensive assessment of carbon productivity, allocation and storage in three Amazonian forests. Glob. Chang. Biol. 15, 1255–1274 (2009).

    Article  ADS  Google Scholar 

21. PRODES. Amazon Deforestation Monitoring Project (PRODES/INPE) https://terrabrasilis.dpi.inpe.br/app/map/deforestation (2025).

22. MMA GOV BR. PPCDAm http://redd.mma.gov.br/pt/acompanhamento-e-a-analise-de-impacto-das-politicas-publicas/ppcdam (2017).

23. West, T. A. P. & Fearnside, P. M. Brazil’s conservation reform and the reduction of deforestation in Amazonia. Land Use Policy 100, 105072 (2021).

    Article  Google Scholar 

24. Mapbiomas_Amazonia. The MapBiomas Amazon Collection 10 includes annual land use and land cover data for the period 1985-2024 https://plataforma.brasil.mapbiomas.org/ (2025).

25. MapBiomas. MapBiomas Fogo. Annual and Cumulative Burned Area and Coverage (Collection 4) https://brasil.mapbiomas.org/en/mapbiomas-fogo (2025).

26. ONI. Oceanic Nino Index, Climate Prediction Center, NOAA https://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/detrend.nino34.ascii.txt (2025).

27. ERA5. European Centre for Medium-Range Weather Forecasts, (ECMWF) - Copernicus Climate Data Store (CDS) https://cds.climate.copernicus.eu/datasets (2025).

28. Climate Prediction Center. Official NOAA CPC ENSO Strength Probabilities. NOAA Climate Prediction Center https://cpc.ncep.noaa.gov/products/analysis_monitoring/enso/roni/strengths/ (2026).

29. Brando, P. M. et al. Tipping points of Amazonian forests: beyond myths and toward solutions. Annu. Rev. Environ. Resour. 50, 97–131 (2025).

    Article  Google Scholar 

30. Flores, B. M. et al. Critical transitions in the Amazon forest system. Nature 626, 555–564 (2024).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

31. Alencar, A. A., Brando, P. M., Asner, G. P. & Putz, F. E. Landscape fragmentation, severe drought, and the new Amazon forest fire regime. Ecol. Appl. 25, 1493–1505 (2015).

    Article  PubMed  Google Scholar 

32. Anderson, L. O. et al. Vulnerability of Amazonian forests to repeated droughts. Philos. Trans. R. Soc. B Biol. Sci. 373, 20170411 (2018).

    Article  ADS  Google Scholar 

33. Alencar, A. A. C. et al. Long-term Landsat-based monthly burned area dataset for the Brazilian biomes using deep learning. Remote Sens. 14, 2510 (2022).

    Article  ADS  Google Scholar 

34. Santos Fleischmann, A. et al. Extreme warming of Amazon waters in a changing climate. Science 390, 606–611 (2025).

    Article  ADS  CAS  PubMed  Google Scholar 

35. Espinoza, J.-C. et al. The new record of drought and warmth in the Amazon in 2023 related to regional and global climatic features. Sci. Rep. 14, 8107 (2024).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

36. Bourgoin, C. et al. Extensive fire-driven degradation in 2024 marks worst Amazon forest disturbance in over 2 decades. Biogeosciences 22, 5247–5256 (2025).

    Article  ADS  Google Scholar 

37. Messias, C. G. et al. Unaccounted for nonforest vegetation loss in the Brazilian Amazon. Commun. Earth Environ. 5, 451 (2024).

    Article  ADS  Google Scholar 

38. Gatti, L. V. et al. CO₂ vertical profiles on four sites over Amazon from 2010 to 2024. PANGAEA https://doi.org/10.1594/PANGAEA.988417 (2026).

39. Eva, H., Huber, O. (eds). A Proposal for Defining the Geographical Boundaries of Amazonia. JRC68635 (Office for Official Publications of the European Communities, 2005).

40. Olson, D. M. et al. Terrestrial ecoregions of the world: a new map of life on Earth: a new global map of terrestrial ecoregions provides an innovative tool for conserving biodiversity. BioScience 51, 933–938 https://doi.org/10.1641/0006-3568(2001)051[0933:TEOTWA]2.0.CO;2 (2001).

    Article  Google Scholar 

41. Domingues, L. G. et al. A new background method for greenhouse gases flux calculation based in back-trajectories over the Amazon. Atmosphere 11, 734 (2020).

    Article  ADS  CAS  Google Scholar 

42. Miller, J. B. et al. Airborne measurements indicate large methane emissions from the eastern Amazon basin. Geophys. Res. Lett. 34, L10809 (2007).

    Article  ADS  Google Scholar 

43. Stein, A. F. et al. NOAA’s HYSPLIT atmospheric transport and dispersion modeling system. Bull. Am. Meteorol. Soc. 96, 2059–2077 (2015).

    Article  ADS  Google Scholar 

44. Lan, X. et al. Atmospheric Carbon Dioxide Dry Air Mole Fractions from the NOAA GML Global Greenhouse Gas Reference Network, Carbon Cycle Cooperative Global Air Sampling Network: 1967 - Present, Version: 2025-08-15 https://doi.org/10.15138/wkgj-f215 (2025).

45. Petron, G. et al. Atmospheric Carbon Monoxide Dry Air Mole Fractions from the NOAA GML Carbon Cycle Cooperative Global Air Sampling Network, 1988-2024, Version: 2025-08-11 https://doi.org/10.15138/33bv-s284 (2025).

46. Botia, S. et al. Combined CO2 measurement record indicates Amazon forest carbon uptake is offset by savanna carbon release. Atmos. Chem. Phys. 25, 6219–6255 (2025).

    Article  ADS  CAS  Google Scholar 

47. Jiang, N. & Riley, M. L. Exploring the utility of the random forest method for forecasting ozone pollution in SYDNEY. J. Environ. Prot. Sustain. Dev. 1, 245–254 (2015).

    Google Scholar 

48. Stekhoven, D. J. & Buhlmann, P. missForest–non-parametric missing value imputation for mixed-type data. Bioinformatics 28, 112–118 (2012).

    Article  CAS  PubMed  Google Scholar 

49. R Core Team. R: A Language and Environment for Statistical Computing https://www.r-project.org (R Foundation for Statistical Computing, 2024).

50. Stekhoven, D. J. missForest: Nonparametric missing value imputation using Random Forest. R version 1.4. https://doi.org/10.32614/CRAN.package.missForest (2013).

51. Cassol, H. L. G. et al. Determination of region of influence obtained by aircraft vertical profiles using the density of trajectories from the HYSPLIT model. Atmosphere 11, 1073 (2020).

    Article  ADS  Google Scholar 

52. Adler, R. F. et al. The global precipitation climatology project (GPCP) monthly analysis (new version 2.3) and a review of 2017 global precipitation. Atmosphere 9, 138 (2018).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

53. Huffman, G. J. et al. Global precipitation at one-degree daily resolution from multisatellite observations. J. Hydrometeorol. 2, 36–50 (2001).

    Article  ADS  Google Scholar 

54. Physical Sciences Laboratory. Global Precipitation Climatology Project (GPCP) Monthly Analysis Product https://psl.noaa.gov/data/gridded/data.gpcp.html (2025).

55. Hersbach, H. et al. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc. 146, 1999–2049 (2020).

    Article  ADS  Google Scholar 

56. Abatzoglou, J. T., Dobrowski, S. Z., Parks, S. A. & Hegewisch, K. C. TerraClimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958–2015. Sci. Data 5, 170191 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

57. Dalagnol, R., Wagner, F. H., Galvão, L. S., Oliveira, L. E. & Aragao, L. E. O. e. C. The MANVI product: MODIS (MAIAC) nadir-solar adjusted vegetation indices (EVI and NDVI) for South America. Zenodo https://doi.org/10.5281/zenodo.3159487 (2019).

58. Giglio, L., Boschetti, L., Roy, D. P., Humber, M. L. & Justice, C. O. The Collection 6 MODIS burned area mapping algorithm and product. Remote Sens. Environ. 217, 72–85 (2018).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

59. de Almeida, C. A. et al. Methodology for Forest Monitoring Used in PRODES and DETER Projects. INPE, version: 2021-01-26 http://urlib.net/rep/8JMKD3MGP3W34R/443H3RE (INPE, 2021).

60. Maurano, L. E. P., Escada, M. I. S. & Renno, C. D. Padrões espaciais de desmatamento e a estimativa da exatidão dos mapas do PRODES para Amazônia Legal Brasileira. Ciênc. Florest. 29, 1763–1775 (2019).

    Article  Google Scholar 

61. DETER. Real-Time Deforestation Detection System, Terrabrasilis, INPE https://terrabrasilis.dpi.inpe.br/app/dashboard/alerts/biomes/amazonia-nb/aggregated/# (INPE, 2025).

62. Alencar, A. et al. MapBiomas Degradation Module BETA - Algorithm Theoretical Basis Document (ATBD) https://doi.org/10.58053/MapBiomas/XPA9ZB (MapBiomas, 2024).

63. BDQueimadas. INPE Biomass burning Program https://terrabrasilis.dpi.inpe.br/queimadas/bdqueimadas (2025).

64. Enfield, D. B., Mestas-Nuñez, A. M., Mayer, D. A. & Cid-Serrano, L. How ubiquitous is the dipole relationship in tropical Atlantic sea surface temperatures? J. Geophys. Res. Oceans 104, 7841–7848 (1999).

    Article  ADS  Google Scholar 

65. DiGangi, E., Lapierre, J., Stock, M., Hoekzema, M. & Cunha, B. Analyzing lightning characteristics in central and southern South America. Electr. Power Syst. Res. 213, 108704 (2022).

    Article  Google Scholar 

66. Zhu, Y., Stock, M. & Lapierre, J. Upgrades of the Earth Networks Total Lightning Network in 2021. Remote Sens. 14, 2209 (2022).

    Article  ADS  Google Scholar 

67. Virts, K. S., Lang, T. J., Buechler, D. E. & Bitzer, P. M. Bayesian analysis of the detection performance of the lightning imaging sensors. J. Atmos. Ocean. Technol. 41, 441–455 (2024).

    Article  ADS  Google Scholar 

68. Schumacher, V. et al. Characteristics of lightning-caused wildfires in central Brazil in relation to cloud-ground and dry lightning. Agric. For. Meteorol. 312, 108723 (2022).

    Article  Google Scholar 

Download references

Authors and Affiliations

1. General Coordination of Earth Science (CGCT), National Institute for Space Research (INPE), São Paulo, Brazil

   Luciana V. Gatti, Camilla L. Cunha, Juan C. G. Blas, Mariane S. Reis, Luciano Marani, Egidio Arai, Cassiano Gustavo Messias, Vanúcia Schumacher, Kleber P. Naccarato, Luciana Soler, Claudio Almeida, Caio S. C. Correia, Raiane A. L. Neves, Alber H. Sanchez, Francine M. da Silva, Vagner L. Camilotti, Christiane Silvestrini, Filipe Loyola Lopes, Marcio Malacarne & Guilherme B. M. Machado

2. Nuclear and Energy Research Institute (IPEN), São Paulo, Brazil

   Luciana V. Gatti & Lucas Gatti Domingues

3. Yale University, New Haven, CT, USA

   A. Scott Denning

4. Amazon Environmental Research Institute (IPAM), Brasilia, Brazil

   Dhemerson E. Conciani & Ane A. C. Alencar

5. National Oceanic and Atmospheric Administration (NOAA), Boulder, CO, USA

   John B. Miller, Xin Lan & Gabrielle Pétron

6. CEMADEN, São Paulo, Brazil

   Jose Marengo

7. Institute of Advanced Studies, University of São Paulo, São Paulo, Brazil

   Carlos Nobre

8. Rio de Janeiro State University (UERJ), Rio de Janeiro, Brazil

   Sergio M. Corrêa

9. Environmental Studies Department, University of California, Santa Cruz, Santa Cruz, CA, USA

   Elliott Campbell

10. University of Kansas, Lawrence, KS, USA

    Jennifer E. Johnson

11. Department of Global Ecology, Carnegie Institution for Science, Stanford, CA, USA

    Joseph A. Berry

Contributions

L.V.G. and J.B.M. conceived of the basin-wide measurement programme and approach; L.V.G., C.L.C., J.C.G.B., A.S.D. and J.B.M. wrote the paper; C.N., J.A.B., E.C. and J.E.J. contributed with ideas, commented and reviewed the manuscript; L.V.G. and C.L.C. performed the results analysis; L.G.D., A.H.S., L.M. and L.V.G. contributed with region of influence study; M.S.R., E.A., C.L.C., L.M., J.C.G.B. and L.V.G. contributed with climate-data-weighted studies; J.M. contributed with climate analysis; M.S.R., C.G.M., V.L.C., L.S. and C.A. contributed with deforestation and degradation analysis; L.G.D., C.S.C.C., L.M., R.A.L.N., F.M.d.S., G.B.M.M., F.L.L. and C.S. contributed with CO₂ and CO concentration analysis; J.C.G.B. and M.M. contributed with maps and processing data analysis; L.V.G. contributed with biogenic CO estimation; C.L.C. produced the missForest; C.L.C., V.L.C. and S.M.C. produced statistical analysis; J.C.G.B. produced data on human economic activities in the Amazon region; D.E.C. and A.A.C.A. provided burned area quantification; X.L. and G.P. provided RPB, ASC and CPT CO₂ and CO data, respectively; V.S. and K.P.N. contributed with fire and lightning study.

Corresponding author

Correspondence to Luciana V. Gatti.

Competing interests

The authors declare no competing interests.

Peer review information

Nature thanks Albertus J. Dolman and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions