Abstract In 2024, the Amazon region experienced severe wildfires driven by exceptional drought conditions. Advanced fire emission models estimated Amazon carbon monoxide (CO) emissions between 28 and 62 Tg during the main August‐September Amazon fire season. The majority of the 2024 CO emissions came from (understorey) forest fires, unlike the previous 6 years for which deforestation fires were the dominant CO source. The strong sensitivity of forest fires to climate makes them the dominant source of inter‐annual variability, and 2024 Amazon forest fire emissions were approximately four times the 2018–2023 average. Comparison with Sentinel‐5p observed atmospheric CO columns indicates nevertheless that even advanced fire emission models underestimate CO emissions by a factor of 1.5–3. This is likely due to prolonged smouldering of fires that enhance fuel consumption during droughts. These results provide strong indications that Amazon 2024 wildfire carbon emissions are also underestimated in advanced fire emission models.

The Brazilian Amazon contains approximately 40% of the world’s tropical rainforest and plays a critical role in preserving biodiversity and regulating water, energy and carbon cycles. However, deforestation and increasingly frequent droughts, heatwaves and wildfires threaten these rainforests. Amazonian fires are generally assumed to be entirely anthropogenic, which has led to lightning-ignited fires being underexplored. Here, we present the first detailed assessment of the spatiotemporal patterns of lightning-ignited fires in the Amazon rainforest to elucidate the role of lightning and human ignitions in shaping Amazon fire dynamics. To do this, we matched cloud-to-ground lightning strokes from the Global Lightning Dataset (GLD360) with individual fire events between 2019 and 2024 to obtain a probability of lightning ignition for each fire. We also calculated a human-ignition probability index using proximity to roads, waterways, and human land cover as proxies for human activity. By combining both probabilistic indices with ground-observed lightning ignitions from eight protected areas, we could optimize the threshold that determines if an ignition is more likely to be caused by lightning or human activities. We estimate that in the Brazilian Amazon, lightning caused on average 0.2%–0.4% of all fires each year (234–407 ignitions per year) and 1.1%–1.2% of the annually burned area (1226–1358 km ^2 per year) between 2019 and 2024. More than 89% of these fires occurred in the late dry season between August and November, peaking in September and October. Despite lightning-ignited fires contributing a small proportion of all Amazonian fires, they constitute over 25% of the fires in identified grid clusters in parts of the states of Pará (particularly in the Breves region), Amazonas, and Rondônia. This study provides the first estimation of the role of natural ignitions in Amazon fire dynamics and a scientific basis for understanding their contribution within the region.

Background Tropical peatland fires contribute to global carbon emissions and air pollution. Aims Enhance the globally used Canadian Fire Weather Index (FWI) system specifically over drained and undrained tropical peatlands in southeast Asia. Methodology We included simulated tropical peatland hydrology in the FWI, creating a new peatland-specific version of the FWI (FWIpeat). FWIpeat, the original FWI (FWIref) and the drought code (DC) were evaluated against satellite-based active fire occurrence from 2002 to 2018. Key results The DC shows superior performance in explaining fire occurrence over undrained tropical peatlands. Over drained peatlands, DC and FWIpeat show similar results, both outperforming FWIref. A comparison with an earlier study over boreal peatlands indicates much smaller improvements from FWIpeat for tropical peatlands, possibly due to a lower accuracy of the hydrological input data. Conclusions Our results highlight the importance of including information on deeper soil layers, i.e. the DC or groundwater table, when assessing fire danger. Implications Although this study offers a promising approach for operational fire management over tropical peatlands, we emphasise the need for further research to refine the hydrological input data and explore additional constraints from Earth observation data.

Fires in the Amazon are of great concern because they threaten the integrity of the tropical forest biome, the carbon cycle, and air quality. Fire emissions depend on the burning behaviour of vegetation biomass, woody debris, and litter. However, the effects of fuels on the combustion process and on the composition of fire emissions are simplified in current fire emission inventories and models. Several new fire emission approaches have recently been developed to better quantify fire emissions by either making use the improved spatial resolution of modern satellite observations or by developing new modelling approaches. Here we compare several current and novel approaches to quantify fuel consumption and fire emissions for the Amazon and Cerrado for the fire season in 2020. The approaches include the widely used GFAS, a top-down approach based on Sentinel-5p observations (KNMI.S5p), a bottom-up approach based on active fire observations from VIIRS (GFA.S4F), two bottom-up approaches based on MODIS burned area data (500-m version of GFED, REFIT.AC), a data-model fusion approach with dynamic emission factors that integrates several Earth observation products (TUD.S4F), and three dynamic global vegetation models in diagnostic mode with prescribed burned area. The different approaches to estimate fire emission show that forest and deforestation fires dominate the regional total fire emissions. However, large differences exist in the very high emissions of individual fires that mainly contribute to the regional total fire emissions. We found a higher agreement in estimated CO and NOx emissions between approaches for savannah fires (normalised RMSE < 20%) than for forest and deforestation fires (nRMSE 30%). We estimate that only 10% of all fire events contribute between 85% and 97% of the regional total fire emissions. By using the TUD.S4F data-model fusion approach with dynamic emission factors, we show that most fire CO emissions originate from the burning of woody debris, which burns with low combustion efficiency and hence has higher emission factors for CO. Comparisons with regional field-based investigations show, however, large differences in estimates of surface fuel loads and fuel consumption. Our results demonstrate the advantage of exploring several complementary fire emission approaches to better understand the underlying processes and to account for regional to global fire emissions and their uncertainties.

Global CO2 emissions from land use and land cover changes contribute about 10% of anthropogenic carbon emissions annually, while terrestrial ecosystems also remove about 29% of annual emissions through enhanced growth and ecosystem recovery. The conservation and restoration of soils, woody biomass and other carbon pools have therefore emerged as key components of national and corporate net zero strategies. Both compliance and voluntary carbon markets, including bilateral and global carbon trading agreements, have the potential to accelerate emissions reductions and removals at local, national, and global scales. The success of these markets relies on effective policy frameworks, transparency and independent assessments of project integrity. Trust in the voluntary markets was shaken during 2023 following media reports of over crediting, while at COP28 nations failed to agree on standards and frameworks for Article 6 of the Paris agreement (on bilateral and UN-supervised carbon markets). Carbon ratings agencies provide independent analysis on the effectiveness and claims of carbon projects operating within the voluntary market. These assessments strengthen trust through greater transparency and price discovery, so that funding can be unlocked for the most effective projects. Here we will discuss challenges and opportunities for scaling the carbon markets, based on our assessment of over 130 nature-based carbon projects globally, assessing key risk factors such as project additionality (what would have happened in the absence of carbon finance), carbon accounting, and non-permanence (risk of future reversal). Our methodologies combine novel monitoring tools, including satellite observations and machine learning models, with information about project finances and policy context to assess the efficacy of every credit issued. Lessons learned can help emerging markets scale, and inform future projections and modelling frameworks of terrestrial carbon fluxes.

The Amazon forest is fire sensitive, but, where fires were uncommon as a natural disturbance, deforestation and drought are accelerating fire occurrences, which threaten the integrity of the tropical forest, the carbon cycle and air quality. Fire emissions depend on fuel amount and type, moisture conditions and burning behaviour. Higher-resolution satellite data have helped more accurately map global burnt areas; however, the effects of fuels on the combustion process and on the composition of fire emissions remain uncertain in current fire emissions inventories. By using multiple Earth observation-based approaches, here we show that total fire emissions in the Amazon and Cerrado biomes are dominated by smouldering combustion of woody debris. The representation of woody debris and surface litter presents a critical uncertainty in fire emissions inventories and global vegetation models. For the fire season 1 August to 31 October 2020, for which all approaches are available, we found 372^605_277 Tg (median and range across approaches) of dry matter burnt, corresponding to carbon monoxide emissions of 39.1^59_27 Tg . Our results emphasize how Earth observation approaches for fuel and fire dynamics and of atmospheric trace gases reduce uncertainties of fire emission estimates. The findings enable diagnosing the representation of fuels, wildfire combustion and its effects on atmospheric composition and the carbon cycle in global vegetation–fire models. Fire emissions in the Amazon and Cerrado biomes are mainly produced from smouldering combustion of woody debris, according to observationally constrained fire emissions inventories.

This study presents FCI-FireDyn, a new algorithm developed to monitor wildfire dynamics using the Flexible Combined Imager (FCI) onboard the Meteosat Third Generation satellite. Leveraging the high temporal resolution of FCI (10-minute full-disk observations), the algorithm derives fire arrival time maps, rate of spread (ROS), and Burn Area (BA) evolution at sub-kilometer spatial resolution and 2-minute temporal intervals. The method combines threshold-based MWIR detection, spatio-temporal interpolation to reconstruct fire front progression and ROS fields at 175 m resolution. FCI-FireDyn was tested on three major fire events in Southern Europe (Portugal, Greece, and France) from the 2024 2025 seasons. The retrieved BA and Fire Growth Rate show good agreement with reference datasets from EFFIS, Copernicus EMS, and PT-FireSprd, with total final BA deviations below 20

Natural ecosystems are fundamental to biodiversity and reaching net-zero, but are at increasing risk from disturbance events like drought and fire. Across many landscapes, fire responds non-linearly to drought and temperature changes, obscuring evolving fire risk until critical thresholds are breached. In particular, in the absence of natural or human-made barriers, growing fire perimeters result in non-linear increases of daily burned area over the lifetime of any individual fire. Fuel conditions and structure further determine the velocity at which fires can spread across the landscape. Predicting fire extremes remains notoriously difficult due to these non-linear responses and complex interactions of natural and managed landscapes, short observational time-series from satellites, and rapid regional trends in climate and human activity. One potential new avenue of exploring fire extremes is through the use of novel object-based fire inventories, like the Global Fire Atlas or Amazon Dashboard. Here we use these novel approaches to assess several recently unfolding fire extremes, with special attention to South America. We find that fire extremes can both unfold within a single season or drought year, as well as over the course of multiple years with continued heightened fire activity across a particular landscape. Further characterization of fire types, based on unique characteristics of each fire object, helps better separate climate and land-use driven variability and change in fire extremes. Our results provide novel insights in the underlying mechanisms driving exceptional fire activity, which can inform estimates of future change and land management strategies.

Amazon forest conservation is critical for reaching net-zero carbon emissions and protecting regional biodiversity but these efforts are at risk from deforestation, fire and drought. In particular, accurate quantification of carbon losses from forest and deforestation fires are required to understand long-term impacts of fire on the carbon cycle and inform management strategies. Recent developments in the detection of burned area, near-real time tracking of fire patch metrics, and higher-resolution fire emissions models allow for improved estimates of carbon losses from fire. Nevertheless, independent validation of these novel approaches often remains elusive, leading to large disagreement between different emissions inventories. Here, we compare carbon emissions estimates from several state-of-the-art fire emissions models, including a 500-m resolution GFED version, GFAS, and the Sense4Fire project, in a case-study for the Amazon region. Where necessary, we have updated the models to extend to 2022 and to include the most recent version of model input data from MODIS (Collection 6.1). We analysed the added years of data to elucidate recent trends in fire-related carbon emissions across the Amazon and adjacent biomes. For validation, we ingested the CO emissions from the considered fire emissions models into an atmospheric transfer simulation (IFS-COMPO) and compared those to column CO observations from Sentinel-5P TROPOMI. Finally, we propose an optimization methodology for matching modelled CO concentrations to observations with the objective of constraining regional carbon losses from fire. Results provide novel insights into carbon losses from fire across different fire types and land use practices, and can be extended to global scale for improved estimates of global fire emissions.

Climate change increases fire-favorable weather in forests, but fire trends are also affected by multiple other controlling factors that are difficult to untangle. We use machine learning to systematically group forest ecoregions into 12 global forest pyromes, with each showing distinct sensitivities to climatic, human, and vegetation controls. This delineation revealed that rapidly increasing forest fire emissions in extratropical pyromes, linked to climate change, offset declining emissions in tropical pyromes during 2001 to 2023. Annual emissions tripled in one extratropical pyrome due to increases in fire-favorable weather, compounded by increased forest cover and productivity. This contributed to a 60% increase in forest fire carbon emissions from forest ecoregions globally. Our results highlight the increasing vulnerability of forests and their carbon stocks to fire disturbance under climate change.

Abstract. Fires are a key component of the global carbon cycle and humans are changing their characteristics. Fire emission monitoring is important to keep track of those changes and TROPOMI satellite observations of tropospheric nitrogen dioxide, carbon monoxide and the absorbing aerosol index can be used to quantify and verify the accuracy and precision of global wildfire emission estimates on a daily basis. Here we use TROPOMI observations to evaluate a new fire emission database based on Global Fire Atlas input for the Sense4Fire project (GFA-S4F) and from the Copernicus Atmosphere Monitoring (CAMS) Global Fire Assimilation System (GFAS) for a number of test regions worldwide representative of the most important wildfire type environments. The main focus is on Amazon and Cerrado biomes (tropical rain forests and deforestation) during August–September 2020, but analyses are also made for a region in sub-Saharan Africa (savannah) as well as two regions in Siberia (steppe and boreal forests/tundra). GFA-S4F and GFAS fire emissions are used as input for global atmospheric composition model simulations based on IFS-COMPO, i.e. an extension of ECMWF’s Integrated Forecasting System (IFS) for simulating atmospheric composition. Comparing the model output with the TROPOMI observations then provides an indirect check on the realism of these emission estimates. Furthermore, for tropospheric nitrogen dioxide the IFS-COMPO model simulations are also used to estimate the model sensitivity of tropospheric nitrogen dioxide columns with respect to fire emission changes. This local relationship is used to optimize the fire NOx emissions directly using the Sentinel-5p nitrogen dioxide observations. The results reveal that for small fires emission nitrogen dioxide estimates are realistic on average albeit with a large spread, i.e. for individual fires emissions can be significantly under or overestimated regardless of emission database. However, for large fires nitrogen dioxide emissions are systematically and largely overestimated in all four regions. The overestimation can be an order of magnitude or even more. For area total nitrogen dioxide emissions this “large fire bias” is of minor importance, i.e. total nitrogen dioxide emissions are dominated by small fires. The GFA-S4F emission estimates were characterized by a larger positive bias for large fire NO2 emission cases compared to GFAS. The source of this bias is not well understood. With optimized NO2 emissions by direct adjustment of emission using TROPOMI nitrogen dioxide observations the large positive bias can efficiently be resolved. Combined with an update of soil NOx emissions – causing too low background NOx levels – a fairly good agreement between IFS-COMPO and TROPOMI was reached. Carbon monoxide was generally underestimated using GFAS emission (~50 % on average for the selected regions). Updating carbon monoxide emissions over the Amazon region by incorporating more Sentinel satellite data (GFA-S4F) did reduce this fire CO bias significantly (to ~25 % on average). Overall, the results show that TROPOMI data allows for systematically identifying uncertainties and errors in satellite-data based fire emissions. The results also suggest that the use of dynamic emission factors may further improve satellite based global emissions inventories. In addition, the results also highlight that the use of TROPOMI data could be much more detailed and refined towards assessing individual fires on a daily basis for better understanding fire dynamics and to improve and diversify fire emission factors.

The Canadian Fire Weather Index (FWI) system, even though originally developed and calibrated for an upland Jack pine forest, is used globally to estimate fire danger for any fire environment. However, for some environments, such as peatlands, the applicability of the FWI in its current form, is often questioned. In this study, we replaced the original moisture codes of the FWI with hydrological estimates resulting from the assimilation of satellite-based L-band passive microwave observations into a peatland-specific land surface model. In a conservative approach that maintains the integrity of the original FWI structure, the distributions of the hydrological estimates were first matched to those of the corresponding original moisture codes before replacement. The resulting adapted FWI, hereafter called FWIpeat, was evaluated using satellite-based information on fire presence over boreal peatlands from 2010 through 2018. Adapting the FWI with model- and satellite-based hydrological information was found to be beneficial in estimating fire danger, especially when replacing the deeper moisture codes of the FWI. For late-season fires, further adaptations of the fine fuel moisture code show even more improvement due to the fact that late-season fires are more hydrologically driven. The proposed FWIpeat should enable improved monitoring of fire risk in boreal peatlands.

Abstract. The Canadian Fire Weather Index (FWI) system, even though originally developed and calibrated for an upland jack pine forest, is used globally to estimate fire danger for any fire environment. However, for some environments, such as peatlands, the applicability of the FWI in its current form, is often questioned. In this study, we replaced the original moisture codes of the FWI with hydrological estimates resulting from the assimilation of satellite-based L-band passive microwave observations into a peatland-specific land surface model. In a conservative approach that maintains the integrity of the original FWI structure, the distributions of the hydrological estimates were first matched to those of the corresponding original moisture codes before replacement. The resulting adapted FWI, hereafter called PEAT-FWI, was evaluated using fire occurrences over boreal peatlands from 2010 through 2018. Adapting the FWI with model- and satellite-based hydrological information was found to be beneficial to estimate fire danger, especially when replacing the deeper moisture codes of the FWI. For late-season fires, further adaptations of the fine fuel moisture code show even more improvement due to the fact that late-season fires are more hydrologically driven. The proposed PEAT-FWI should enable improved monitoring of fire risk in boreal peatlands.

The Fire Weather Index (FWI) is used worldwide to estimate the danger of wildfires. The FWI system integrates meteorological parameters and empirically combines them into several moisture codes, each representing a different fuel type. These moisture codes are then used in combination with wind speed to estimate a fire danger. Originally, the FWI system was developed for a standard jack pine forest, however, it is widely used by fire managers to assess the fire danger in different environments as well. Furthermore, it is often also used to assess the vulnerability of organic soils, such as peatlands, to ignition and depth of burn. The utility of which is often questioned. This research aims at improving the original FWI for northern peatlands by replacing parts of the original, purely weather-based FWI system with satellite-informed model estimates of peat moisture and water level. These come from a data assimilation output combining the NASA catchment model, including the peat modules PEATCLSM, and Soil Moisture and Ocean Salinity (SMOS) L-band brightness temperature observations. The predictive power of the new, peat-specific FWI (PEAT-FWI) is evaluated against the original FWI against fire data of the global fire atlas from 2010 through 2018 over the major northern peatlands areas. For the evaluation, the fires are split up in early and late season fires, as it is hypothesized that late fires are more hydrological driven, and the predictive power of the PEAT-FWI will thus differ between the two types of fires. Our results indeed indicate that the PEAT-FWI improves the predictive capability of estimating fire risk over northern peatlands in particular for late fires. By using a receiver operating characteristics (ROC) curve to evaluate the predictive power of the FWI against a random estimate, the area under the curve increases by up to 10% for the PEAT-FWI compared to the original FWI. The recent version 7 release of the operational Soil Moisture Active Passive (SMAP) Level-4 Soil Moisture Data Assimilation Product now includes PEATCLSM, thus, the proposed PEAT-FWI is straightforward to include in operational FWI products.

Abstract. Long-term records of burned area are needed to understand wildfire dynamics, assess fire impacts on ecosystems and air quality, and improve fire forecasts. Here we fuse multiple streams of remote sensing data to create a 24-year (1997–2020) dataset of monthly burned area as a component of the 5th version of the Global Fire Emissions Database (GFED5). During 2001–2020, we use the Moderate Resolution Imaging Spectroradiometer (MODIS) MCD64A1 burned area product and correct for the errors of commission and omission. Adjustment factors are estimated based on region, land cover, and tree cover fraction, using spatiotemporally aligned burned area from Landsat or Sentinel-2. Burned area in cropland, peatland, and deforestation regions is estimated from MODIS active fire detections. Along Track Scanning Radiometer (ATSR) and Visible and Infrared Scanner (VIRS) active fire data are used to extend the time series back to 1997. Global annual burned area is estimated to be 774 ± 63 Mha yr−1 during 2001–2020 or 5.9 ± 0.5 % of ice-free land. Burned area declined by 1.21 ± 0.66 % yr−1, a cumulative decrease of 24.2 ± 13.2 % over 20 years. The global reduction is primarily driven by decreases in fire within savannas, grasslands, and croplands. Forest, peat, and deforestation fires did not exhibit long-term trends. GFED5 global burned area is 93 % higher than MCD64A1, 61 % higher than GFED4s, and in closer agreement with burned area products from higher-resolution satellite sensors. These data may reduce discrepancies between fire emission estimates from top-down and bottom-up approaches, and improve understanding of global fire impacts on the carbon cycle and climate system.

&lt;p&gt;Emissions from vegetation fires in tropical forests have the potential to turn the global land carbon sink into a source, affect atmospheric chemistry, and hence air quality. While natural forest fires are a rare phenomenon in tropical forests of South America and are usually of rather low intensity, deforestation fires and small land clearings in systems with high fuel loads can cause intense fires and high emissions. However, the high moisture content in tropical forests causes incomplete combustion and higher emissions of carbon monoxide (CO) than of carbon dioxide. The interacting effects of land use change, fuel load and moisture on fire intensity and emissions is, however, difficult to quantify at large scales because not all of those components are readily available from Earth observations in a consistent way.&amp;#160;&lt;/p&gt; &lt;p&gt;Here, we make use of several satellite products on vegetation, fire activity and atmospheric composition to quantify the effects of land use, fuel loads, fuel moisture on fuel consumption, emission factors and hence on emissions and atmospheric trace gas concentration. First, we use observations of active fires and fire radiative power from the VIIRS and Sentinel-3 SLSTR sensors to map different fire types (forest fires, deforestation fires, small land clearing and agricultural fires, savannah fires). Second, we integrate satellite products of canopy height, above-ground biomass, leaf area index, land cover and soil moisture in a novel data-model fusion framework to estimate fuel loads and moisture in vegetation, surface litter and woody debris. We then combine in a bottom-up approach the fire types with fuel loads and moisture to estimate fuel consumption and fire emissions using default emission factors. Third, we use observations from Sentinel-5p TROPOMI and the Integrated Forecast Systems (IFS) of the Copernicus Atmosphere Monitoring Service to compare the bottom-up estimates with distributions of CO and NOx in the atmosphere, which allows optimising emissions and associated emission factors.&lt;/p&gt; &lt;p&gt;Our reconciled estimates of fire emissions outperform previous CO estimates e.g. from the Global Fire Assimilation System, which demonstrates an improved estimation of fire carbon emissions. The results show that the high fire intensity and emissions in tropical deforestation fires originate from the burning of high loads of woody biomass and coarse woody debris. The high fuel moisture content causes higher emission factors of CO in tropical forests than in savannah fires and hence higher absolute emissions of CO. Our new model approaches and satellite products allow to provide an integrated assessments on the effects of fuel and fire behaviour on fire emissions.&lt;/p&gt;

Modeling the spread of wildland fires is essential for assessing and managing fire risks. However, this task remains challenging due to the partially stochastic nature of fire behavior and the limited availability of observational data with high spatial and temporal resolutions. Herein, we propose an attention-based deep learning modeling approach that can be used to learn the complex behaviors of wildfires across different fire-prone regions. We integrate optimized spatial and channel attention modules with a convolutional neural network (CNN) modeling architecture and train the attention-based fire spread models using a recently derived fire-tracking satellite observational dataset in conjunction with corresponding fuel, terrain, and weather conditions. The evaluation results and their comparison with benchmark models, such as a deeper and more complex autoencoder model and the semi-empirical FARSITE fire behavior model, demonstrate the effectiveness of the attention-based models. These new data-driven fire spread models exhibit promising modeling performances in both the next-step prediction (i.e., predicting fire progression from one timestep earlier) and recursive prediction (i.e., recursively predicting final fire perimeters from initial ignition points) of observed large wildfires in California, and they provide a foundation for further practical applications including short-term active fire spread prediction and long-term fire risk assessment.

Fires mediate grass and tree competition and alter vegetation structure in savanna ecosystems, with important implications for regional carbon, water, and energy fluxes. However, direct observations of how fire frequency influences vegetation structure and post‐fire recovery have been limited to small experimental field studies. Here, we combined lidar‐derived canopy height and canopy cover from NASA's Ice, Cloud, and land Elevation Satellite‐2 with over two decades of burned area data from the Moderate Resolution Imaging Spectroradiometer sensors to provide the first biome‐wide estimates of post‐fire changes in canopy structure for major vegetation types in the Cerrado (Brazil). Mean canopy height decreased with increasing burn frequency for all natural cover types, with the greatest decline observed for forests and savannas. The ability to separate changes in fractional canopy cover from height growth using lidar data highlighted the long‐time scales of vegetation recovery in forests and savannas after fire. For forests in medium and high precipitation areas, canopy cover returned to unburned values within 5 years following fire, whereas mean canopy height remained below unburned values, even in the oldest fires (14–20 years). Recovery times increased with decreasing rainfall, with average values of both fractional cover and canopy height below unburned areas after 14–20 years for savannas. We observed gradual recovery of vegetation height and cover over decades, even in mesic or wet savanna regions like the Cerrado. Infrequent fire activity, particularly in areas with greater land management, influences ecosystem structure across the biome, with important consequences for biodiversity conservation.