Vegetation greening enhances global fire activity

Vegetation greening enhances global fire activity | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Vegetation greening enhances global fire activity Yongguang Zhang, Gengke Lai, Chaoyang Wu, Alessandro Cescatti, and 8 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5467904/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Global terrestrial ecosystems have witnessed increased vegetation greenness 1 – 3 and intensified fire regimes 4 – 7 in many ecosystems worldwide, but the potential connections between them remain elusive. We quantify the impact of vegetation greening on global fire activity by examining changes in live and dead fine fuels based on multiple long-term satellite-based datasets. We show that, despite the recently observed human-driven decline in global burned area 8 , vegetation greening has led to an increase in global burned fraction at a rate of 0.014 ± 0.004% per year over 2001–2020. This amplifying effect is primarily driven by the increase in dead fine fuel (0.047 ± 0.009% per year), partially offset by the dampening effect of increased canopy live foliage (-0.018 ± 0.007% per year). Notably, current fire-vegetation models inaccurately represent the interactions between fire and greening, resulting in underestimations of fire responses to vegetation greening, particularly in arid and cold regions. Our findings highlight the widespread amplification of global fire activity caused by the ongoing trend of vegetation greening. They underscore the importance of considering this biogeochemical positive feedback in the land-climate system and support the efforts to mitigate its impact on ecosystems and societies. Earth and environmental sciences/Climate sciences/Climate change/Climate and Earth system modelling Earth and environmental sciences/Climate sciences/Climate change/Climate-change impacts Biological sciences/Ecology/Fire ecology Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Main Text In 2023, Canada experienced a significant rise in mega-fire incidents 9 , 10 , closely following the unprecedented wildfire outbreaks in Australia from 2019 to 2020 and Siberia in 2021. Global terrestrial ecosystems have recently experienced a substantial amplification of extreme wildfire events 6 , 11 , posing considerable threats to biodiversity 12 , 13 , terrestrial carbon storage 14 – 16 , air quality 17 – 19 , and thus regional-to-global climate 20 . Wildfires occur when certain conditions are met, including fuel availability and dryness, ignition source and favorable weather conditions 21 . Both local and regional evidences reveal that climate change has prolonged fire season 22 , intensified extreme fire weather conditions 23 , and exacerbated the extent 24 , 25 , intensity 26 , severity 27 , and emission 14 of wildfires in many ecosystems worldwide. Moreover, fuel availability also plays a crucial role in shaping fire regimes across many regions, as shown by natural fuel (aridity) gradients 28 and human activity 21 , 29 . There are substantial local-to-regional evidences of fire regime changes due to human modifications of fuels by land management 30 , 31 or forestry 32 , 33 , such as increased landscape fragmentation and reduced fuel continuity due to agricultural expansions in tropical areas 8 , enhanced fuel build-up driven by rural abandonment in southern Europe 31 , by active fire exclusion in western US 34 , 35 , and by forestry plantations in Chile and Portugal 36 . However, evidence of fuel role based on how climate change and CO 2 fertilization impact vegetation growth at the global scale remains limited 21 . Vegetation dynamics can regulate fire-climate interactions, either exacerbating or mitigating fire risk 37 . Therefore, understanding the mechanisms that control the response of fire to vegetation dynamics is key to gaining insights into how fire regimes will change under climate change. In response to elevated CO 2 concentration, climate change, nitrogen deposition, and land-use change, our planet is greening on large part of vegetated areas, as documented by satellite-based observations of a persistent increase in leaf area index (LAI) over the past four decades 1 , 2 . Vegetation greening can enhance green leaf cover, productivity and connectivity by accelerating photosynthesis and increasing carbon sequestration during the growing season 38 , 39 . This greening-induced fuel build-up can increase fire activity in some low-productivity arid regions in which fire regimes are regarded as fuel-limited 40 – 42 . However, larger green biomass may also create shadier environments (e.g. in forest), making fire more difficult to penetrate and thereby reducing fire activity, especially in moist environments. Additionally, vegetation greening is associated with increased production of dead fuels 21 , 43 , leading to an accumulation of dead fine fuel on the ground. This occurs through processes such as leaf senescence and shedding in forest ecosystems and higher grass productivity generating more dead grasses through seasonal drying in savannas. How these biogeochemical processes induced by vegetation greening combine to influence global fire activity, however, remains elusive. In this study, we tested whether vegetation greening, including increase in canopy live foliage and dead fine fuel (litter and dead grass) due to senescence and shedding, can exacerbate or mitigate global fire regimes, and which process dominants across global terrestrial ecosystems. We expected that: (ⅰ) at a global scale, most fires are initially fueled by dead fine fuel on the ground before spreading to other components (e.g. live fine fuel in the crown). Therefore, dead fine fuel dominants the greening effects on fire and can amplify global fire activity due to its higher flammability and lower moisture content. (ⅱ) For canopy greening, there are regional contrasting effects on fires, with an amplifying effect in fuel-limited regions due to increased availability and connectivity of live fine fuel (e.g. some arid and non-forest ecosystems), and a dampening effect in moisture-limited regions due to wetter and shadier environments (e.g. some forest ecosystems). This leads to two greening-induced fundamental processes that were tested in this study: increasing crown live fine fuel and ground dead fine fuel. To test these hypotheses, we proposed a framework to calculate two indices: LAI live and LAI dead . LAI live represents the interannual variation of live fine fuel and is defined as the maximum three-consecutive-month averaged LAI. LAI dead , a relative index for the interannual variation of dead fine fuel, is defined as the reduction of LAI during the senescence period, assuming that negative seasonal variations in LAI reflect the transfer of living leaf biomass to the litter pool and dead grass (Methods; Extended Data Figs. 1 and 2 and Fig. S1 ). We combined an ensemble of satellite-based global burned area [aggregated to grid cell-level burned fraction (BF) with unit of %; Fig. S2], fire radiative power (FRP) and LAI products, as well as long-term climatic variables (Table S1 ). Next, we explored the control effects of water and productivity conditions on determining how fire responds to vegetation greening. Moreover, we examined the fire response to vegetation greening in state-of-the-art fire-vegetation models participating in the Fire Model Intercomparison Project (FireMIP; Table S2) to compare simulations with observation-driven assessments. Based on the estimated fire sensitivities to vegetation greening from satellite observations, we further projected the changes in fire regime induced by long-term vegetation greening until the end of the 21st century under four Shared Socioeconomic Pathway (SSP) scenarios (Table S3). Global fire sensitivity to vegetation greening Our estimation of fire sensitivity to vegetation greening at each grid cell across the globe (Method) suggests that an increase in LAI dead magnified BF in ~ 80% of global vegetated land. Areas with positive fire sensitivity to LAI dead (∂BF/∂LAI dead ) spanned over most of the globe, with an overall sensitivity of 3.94 ± 0.53%/m 2 m − 2 (Fig. 1 a), indicating the widespread amplifying effect of increased LAI dead on fire activity. Hotspots of positive sensitivity are located in North American boreal area, Eurasian steppe, Siberia, and large arid areas in the southern hemisphere (Fig. 1 a). This positive ∂BF/∂LAI dead shows a bimodal distribution along the latitudinal gradients, with the strongest effect in the southern hemisphere and the high-latitude boreal area (Fig. 1 b). This global pattern of fire sensitivity to LAI dead was also supported by using fire radiative power (FRP), indicating that an increase in dead fine fuel also tends to magnify global fire intensity (Extended Data Fig. 3 a and c). On the other hand, our analysis revealed contrasting fire sensitivities to LAI live (∂BF/∂LAI live ) across different regions (Fig. 1 d). Approximately 47% of global vegetated areas showed positive fire responses to canopy greening, mainly distributed in Alaska, west coast North America, African savanna, and midwestern Australia. In contrast, the sensitivities were opposite in the remaining 53% of vegetated areas, including the high-latitude northern ecosystems and the tropics. This contrasting fire response to increased LAI live is also supported by the latitudinal distributions of ∂BF/∂LAI live (Fig. 1 e) and by using FRP to indicate variations in fire intensity (Extended Data Fig. 3 b and c). These spatial patterns of the interactions between fire activity and LAI dead and LAI live were further confirmed by the climatological gradients of fire sensitivities (Fig. 1 c and f). The consistently positive ∂BF/∂LAI dead were observed across climatological gradients, with higher sensitivities at moderate and low precipitation levels [mean annual total precipitation (PRE) 10 ℃ and PRE < 1000 mm], and a negative one in cold or moist areas (T 1000 mm) (Fig. 1 f). This finding is supported by a previous study across tropical continents, quantifying the precipitation threshold cross which fire regimes switch from fuel- to moisture-limited 44 . This pattern is also consistent with the intermediate fire-productivity hypothesis 40 suggesting that in warm and dry regions fire is mostly limited by fuel, whose production is positively related to LAI live . Conversely, in more humid regions, fire is limited by the occurrence of favorable dry climatic conditions. Higher LAI live , which is generally associated with wetter and shadier environments, is typically less favorable for fires. Comparatively, we found stronger fire sensitivities to LAI dead than LAI live in all climate zones (Fig. 1 a and d). Higher positive fire sensitivity to LAI dead were observed than LAI live in arid regions (7.92 ± 0.43 vs. 3.25 ± 0.64%/m 2 m − 2 ). Additionally, the magnitudes of positive effects of LAI dead were larger in tropical, temperate, and cold regions (3.01 ± 0.52, 2.30 ± 0.28, and 4.75 ± 0.88%/m 2 m − 2 , respectively), compared to the dampening effects of canopy greening (− 2.04 ± 0.32, − 1.30 ± 0.16, and − 1.18 ± 1.04%/m 2 m − 2 , respectively). Overall, the magnitudes of fire responses to the increase in LAI dead were stronger and more spatially consistent than those to increased LAI live . The stronger sensitivity of fire to LAI dead is likely due to the unidirectional positive effect of this parameter on fire. In fact, the LAI dead is both a relative index of available fuel amount and of conditions favorable to fire, because the moisture of dead fine fuel is lower and more susceptible to dry weather conditions 45 . On the contrary, LAI live is related to two antagonistic processes. On one side it positively correlates with biomass and therefore fuel amount and continuity (positive effect for fire), while on the other side, it relates to shady environments and conditions of higher wetness that are negative for fire. Which of these two antagonistic effects prevails is likely to depend on the local background climate and vegetation, and on the limiting factors for fire occurrence. We further compared the ∂BF/∂LAI dead and ∂BF/∂LAI live among different plant functional types (PFTs) and biomes. For all PFTs and biomes, fire activity exhibited stronger responses to LAI dead than LAI live . We also found the strongest positive sensitivities in shrubland. The response in needleleaf forests was stronger than that in broadleaf and mixed forests, and stronger for deciduous than evergreen forests (Extended Data Fig. 4 a). Moreover, the response was stronger in dry forests than in moist ones, and more pronounced in semi-arid, arid, boreal forest and tundra ecosystems (Extended Data Fig. 4 b). Additionally, significantly different ∂BF/∂LAI dead and ∂BF/∂LAI live were observed between forest and non-forest ecosystems ( p -value < 0.01). For LAI dead , sensitivity in non-forest was approximately twice that in forest (6.89 ± 0.39 vs. 3.90 ± 0.73%/m 2 m − 2 ) (Extended Data Fig. 4 c). BF in forest and non-forest responded to LAI live change in opposite ways, with sensitivities of − 2.23 ± 0.73 and 1.04 ± 0.33%/m 2 m − 2 , respectively (Extended Data Fig. 4 d). That is in forest, greening generates shadier conditions that inhibit fires, while in open-canopy (non-forest) ecosystems, it contributes to fuel build-up (live and dead) and increases fire activity. Vegetation greening as a key role in global fire activity We further quantified the actual global trend in fire activity induced by the long-term trends in LAI over the last two decades (Methods). From an ensemble of satellite-based LAI products, we found a widespread greening trend for LAI live with a global mean of 0.013 m 2 m − 2 y − 1 , and for LAI dead of 0.011 m 2 m − 2 y − 1 (Extended Data Fig. 2 and Fig. S1 ). Correspondingly, the global mean greening-induced trend in burned fraction (δBF LAI−total , total effects of LAI live and LAI dead ) was 0.014 ± 0.004% y − 1 (Fig. 2 c and d), although burned area declined during the last two decades (Fig. S2). This strengthening effect was attributed to the imbalance between a larger amplifying effect of LAI dead (δBF LAI−dead , 0.047 ± 0.009% y − 1 ) and a slightly dampening effect of LAI live (δBF LAI−live , − 0.018 ± 0.007% y − 1 ) (Fig. 2 a-d). However, spatial variabilities emerge across climate zones (Fig. 2 d). The BF trend induced by greening was largest in arid regions (0.034 ± 0.011% y − 1 ) due to the amplifying effect of both LAI dead and LAI live , followed by cold regions (0.023 ± 0.007% y − 1 ), especially in Siberia and Alaska regions (Fig. 2 c). The effect of LAI dead dominates the amplifying effect across cold areas (0.070 ± 0.014% y − 1 ), partly offset by the dampening effect of LAI live change (− 0.020 ± 0.016% y − 1 ). The amplifying effect of LAI dead was larger than the dampening effect of LAI live in the tropics, resulting in an overall increase in fire activity. The effect of greening on fire was minimum in temperate regions. The methodology used to quantify the effects of long-term trends in LAI on BF trend was also applied to each climate factor (T and PRE) to assess the relative contributions of vegetation greening and climate change (Methods). We found that although warming dominated global fire dynamics, increasing global BF at a rate of 0.035 ± 0.0005% y − 1 , more than twice the total greening effect (0.014 ± 0.004% y − 1 ), changes in LAI dead still played a key role in controlling trends of BF, as evidenced by the dominant contribution of this parameter at global scale (Fig. 2 e). The contribution of LAI live was relatively lower at global level (Fig. 2 e), resulting from the large spatial complementarity across different climate zones (Fig. 2 b). In cold regions, LAI dead had the greatest effect on the variations of BF, resulting from a combination of high fire sensitivity to LAI dead (Fig. 1 a) and large increased trend in LAI dead (Extend Data Fig. 2 d). This may be indirectly supported by the fact that in boreal forest and tundra, litter and soil organic materials, whose accumulations are related to annual transfer from live biomass to dead (LAI dead in this study), account for more than 80% of fuel consumption during fires 46 . Drivers of spatial variabilities of ∂BF/∂LAI and ∂BF/∂LAI In order to interpret the spatial variability of the fire sensitivity to LAI live and LAI dead , we explored the underlying mechanisms that explain the divergent responses of fire activity (sign and magnitude) to vegetation change (LAI live and LAI dead ) across different regions. The possible drivers include water availability, quantified by aridity index and relative humidity (RH), and vegetation growth, quantified by LAI live and tree cover. We found significant negative correlations between ∂BF/∂LAI dead and aridity index ( r = − 0.88, p -value < 0.01) and LAI live ( r = − 0.94, p -value < 0.01) (Fig. 3 a and b). Similar relationships were observed for RH and tree cover (Extended Data Fig. 5 ). These results reveal a stronger positive fire response to LAI dead in arid and low-productivity areas than that in moist and high-productivity areas. We also found that water availability and vegetation growth had a strong capability to regulate the spatial variabilities of ∂BF/∂LAI live , revealed by the significant negative correlations between ∂BF/∂LAI live and aridity index and LAI live ( r = − 0.70 and − 0.78, respectively, both p -value < 0.01) (Fig. 3 c and d). These controlling effects were further confirmed by the thresholds of aridity index (0.364 ± 0.002) and LAI live (2.25 ± 0.035 m 2 m − 2 ) that can partition positive and negative ∂BF/∂LAI live (Fig. 3 e and f). These findings were further supported by RH and tree cover (Extended Data Fig. 6). These results revealed that an increased LAI live could increase fire activity in arid and low-productivity areas (fuel build-up limited), while the contrary happens in moist and high-productivity areas (fuel-moisture limited). This finding is supported by the intermediate fire-productivity hypothesis, which describes changes in fire activity along with productivity and moisture gradients regulated by critical thresholds 28 , 40 , 44 , 47 . The observed thresholds of water availability and productivity are relevant for projecting the changes in fire regimes from fuel build-up limited to fuel-moisture limited or vice versa under greening and warming scenarios. Diagnosis of global fire sensitivity to vegetation greening in FireMIP models Fire sensitivities to vegetation greening derived from satellite observations are a valuable benchmark to diagnose the ability of models to simulate the fire-vegetation interactions. We thus investigated whether state-of-the-art fire-vegetation models participating in the FireMIP can capture the observed fire sensitivities to vegetation greening (including increases in canopy live foliage and dead fine fuel). As not all models provide monthly LAI outputs, we used alternative net primary productivity (NPP) outputs from models to depict vegetation greening. The sensitivities from satellite were derived from satellite LAI driven BEPS NPP and MODIS burned area products. Given that there presented strong positive relationships between the sensitivities from NPP and those from LAI (Fig. S4), the satellite results from NPP can be used to diagnose the capability of models. We found that fire-vegetation models captured the direction of fire response to change in dead fine fuel with positive sensitivities over four climate zones, but underestimated their magnitudes on average (multi-model mean vs. satellite) (Fig. 4 ), particularly in arid and cold regions (Fig. 4 b and d). The models misrepresented the fire sensitivity to changes in canopy live foliage in arid regions with four out of seven models showing negative responses (Fig. 4 b), which may result from the misrepresentation of drylands LAI in the models 48 . Moreover, all models underestimated the magnitude of fire responses to both canopy live foliage and dead fine fuel in high-latitude cold regions (Fig. 4 d). These discrepancies between model simulations and observations may result from the failure of models in reproducing the magnitude and spatial pattern of interannual variability of burned area 8 , 48 . Models also have difficulty in replicating the length of the growing season 48 , which may bias the representations of accumulated live and dead fuels. Amongst seven FireMIP models, the CLM overestimated the fire responses to vegetation greening in tropical and temperate regions (Fig. 4 a and c). A recent study also found that CLM exhibited the largest fire impacts on the carbon cycle 49 . These evidences suggest that CLM may be too sensitive to changes in terrestrial carbon storage. Overall, the FireMIP models underestimated the interactions between fire activity and vegetation greening, particularly in arid and cold regions. This finding is supported by an evaluation showing that FireMIP models generally underestimate the sensitivities to pre-season vegetation productivity in semi-arid ecosystems 50 . Projection of future greening-induced fire changes Given the estimated fire sensitivities to LAI live and LAI dead from satellite observations (Fig. 1 a and d) and the ongoing persistent greening projected for the future (Fig. S5) 1 , we further projected the variations of fire activity (indicated by BF) induced by vegetation greening until the end of the 21st century using four SSP scenarios (SSP126, 245, 370, and 585) of LAI, considering two processes of greening effect (live and dead fine fuels) (Methods). Assuming constant sensitivities for the next decades, we found a projected increase of greening-induced fire activity in 65% of global vegetated areas over 2081–2100 compared with 2001–2020, and increasingly severe from low to high emission scenarios (Fig. 5 a–d). The arid and cold regions were projected to experience larger amplifications with BF increase of 0.67 ± 0.18% and 0.96 ± 0.17% at SSP126, respectively, and correspondingly of 2.28 ± 0.89% and 2.17 ± 0.36% at SSP585 (Fig. 5 e). Moreover, vegetation greening could lead to increased BF at a rate of 0.010–0.021% y − 1 during 2015–2100 under the four SSP scenarios (Fig. 5 f). Towards understanding the fire-greening feedback Our study offers insights into how global fire activity responds to vegetation dynamics driven by global change factors (CO 2 fertilization, climate change, and land use change, etc.). We provide evidence of a positive feedback between vegetation greening and fire, which is contributing to the intensification of fire regimes, increasing both the area burned (Fig. 1 ) and the intensity of fires (Extended Data Fig. 3 ). Elevated atmospheric CO 2 can enhance plant growth and litter production 21 . Therefore, the greening effects on fire are achieved through two processes, i.e. increasing live and dead fine fuels (represented by LAI live and LAI dead , respectively). More greening during the growing season results in more dead litters and dry grasses due to senescence and shedding during the senescence period (Extended Data Fig. 2 b and d). Our results highlight the importance of CO 2 fertilization and climate change-induced modification of vegetation growth, and consequently fuels, for predicting changes in fire regimes. We argue that LAI live and LAI dead represent live and dead fine fuels, and do not include coarse woody components such as stem, branch, and bark. Fine fuels, either alive (foliage) or dead (litter and dry grass), are the most flammable components, driving fire activity and emission across many ecosystems worldwide 46 . Moreover, many factors may induce uncertainties in the LAI dead calculation. The seasonal reduction of LAI can also be caused by disturbances or human activities, such as insect hazards, grazing, and timber harvest, which may not result in the formation of dead fuels. Despite these limitations, we found a strong effect of LAI dead changes in fire activity. In addition, LAI dead is a relative index indicating the interannual variation of dead fine fuel rather than its absolute quantity, which is demonstrated by comparing with global litterfall production dataset (Fig. S6) and site-level ground litterfall measurements (Fig. S7 and Table S4). The natural formation of dead fine fuel is related to leaf area and longevity, turnover time, the rate of decomposition, and other biogeochemical processes 43 , 51 . Improving the representation of these processes may benefit the modeling of dead fuel accumulation in fire-enable dynamics global vegetation models. Although temperature and precipitation can partly explain the variations of fire weather and foliar and dead fuel moisture 23 , 52 , this oversimplification in the attribution model (see Methods) may overlook the direct effects of fire weather and water deficits in the soil and atmosphere. Additionally, the annual aggregated climate variables may obscure the influence of seasonal climate factors on fires 53 , such as high temperature and low precipitation during the fire season, which are more directly related to the hot and dry conditions that drives fires. Therefore, we also used fire season maximum temperature and precipitation as the climatic drivers and obtained robust consistent findings (Fig. S8a and b). Additionally, we directly employed the fire weather index (FWI) and climatic water deficit (CWD) during the fire season, with CWD calculated as the difference between potential and actual evapotranspiration and more closely related to the water deficit that determines fuel flammability 53 . The results remained unchanged (Fig. S8c and d). Vapor pressure deficit (VPD) can largely explain the changes in extreme FWI 23 and serve as a reliable predictor of dead fuel moisture content 54 (Fig. S9). We also induced surface soil moisture (indicating soil water supply) and relative humidity (indicating atmospheric water demand) into the regression models, and obtained robust results (Fig. S10). Our findings were also robust when using nonlinear multiple regression model (Fig. S11), different sources of satellite LAI data (Fig. S12), using different spatial moving windows to increase sample size of model training (Fig. S13), as well as different size of temporal window to calculate LAI live and LAI dead (Fig. S14). We also constructed the bioclimatic spaces to qualitatively analyze the fire responses to vegetation greening and the control effects of moisture and productivity on their spatial variabilities (Fig. S15 and S16). Additionally, we also estimated the sensitivities using random forest and explainable machine learning (SHapley Additive exPlanations, SHAP), which further strengthens the robustness of the results obtained with traditional statistical approaches (Fig. S17). As the fire-enabled dynamics global vegetation models in the FireMIP are designed for future burned area projections under climate change, their abilities are crucial for future projections of fire dynamics and impacts and terrestrial carbon budget, depending on how to represent relationships between climate, vegetation, socio-economics and burned area in the models. Our findings provide a benchmark to examine whether the models can replicate the fire-greening interactions from satellite observations. Previous analysis using the sensitivity runs of FireMIP models with CO 2 concentration fixed indicated that most models did not show a clear fire response to CO 2 fertilization, except for LPJ-GUESS-SPITFIRE and JSBACH-SPITFIRE showing that CO 2 fertilization considerably contributed to an increase in burned area 55 . As CO 2 fertilization impacts fire activity mainly through altering vegetation dynamics, the sensitivity runs of models can indirectly support our findings that current models underestimate the fire-greening interactions. Altogether, the biases shown by models go in the direction of underestimating the positive feedback between greening and fire, potentially leading to an underestimation of future fire regimes and an overoptimistic projection of the future terrestrial carbon budget. Vegetation greening is one of the highly credible evidence of anthropogenic climate change 1 . Our study provides a new perspective to understanding the impact of vegetation greening, i.e. potential intensification of fire regimes (area burnt and intensity), which has received relatively less attention in previous studies. Global burned area declined on average over the past two decades, primarily driven by human-induced declines in savannas 8 . However, in many other ecosystems, burned area is increasing such as the western America and boreal forests 37 , in which vegetation greening plays an important role driving the amplification of fire activity. Our results call on the urgent need to monitor ecosystems with a potential strong coupling between greening and fire, such as high-latitude northern ecosystems 7 . The Arctic tundra is experiencing greening that is associated with the shift of vegetation composition, such as tundra shrub expansion 56 . This expansion may amplify fire activity because of the largest positive sensitivities to LAI live and LAI dead of shrubland compared with other PFTs (Extended Data Fig. 4 a). Altogether, greening-induced intensification of fire regimes may threaten the historic carbon stored in vegetation, soil and peatland in boreal forest and Arctic tundra ecosystems, and release large amounts of carbon into the atmosphere 24 , 57 – 59 . This mechanism may indeed be considered as a positive feedback in the climate system that is partially offsetting the negative feedback driven by the fertilization effect of CO 2 on greening and primary productivity. Relatedly, the massive afforestation actions committing to mitigate climate change (e.g. Bonn Challenge, Great Green Wall, and the Three Billion Tree Pledge, etc.) have contributed to the vegetation greening, especially in dry areas and savannas 3 , 60 , 61 . Our results suggest that these actions could also produce counterproductive side effects, that is, potential amplification of fire regimes and a consequent increase of the vulnerability of the land carbon stock 33 , 60 , 61 . Our study reinforces that these afforestation actions would require extensive forest management (e.g. fuel management, especially dead fine fuel) to provide any beneficial effects, which is not guaranteed in low-income regions where most afforestation are foreseen. The strategies of fuel management (e.g. prescribed burning) should be environmentally sustainable and optimized to maximize the benefits of fuel reduction to mitigate fire hazard, while minimizing the adverse effects on ecosystem services (e.g. carbon sequestration and soil functioning) and biodiversity 7 , 18 . In summary, we conclude that while global burned area has been declining primarily driven by agricultural expansion and intensification over the past two decades, vegetation greening is contributing to an intensification of global fire regimes in addition to climate warming. This amplification is primarily attributed to the increased dead fine fuel, which can counteract the potential dampening effect of increased canopy live foliage. Vegetation greening has the potential to enhance leaf cover and connectivity, and it also leads to an increase in dead fine fuel due to leaf and grass senescence. These biogeochemical feedbacks from vegetation greening could further exacerbate the fire regime changes driven by climate change. As vegetation effect on fire is identified as a main deficiency of current fire-vegetation models 50 , future efforts should focus on better characterizing the processes linking fire activity and vegetation dynamics in data-driven and process-based fire models to improve predictions on future trajectories of the terrestrial carbon budget. In addition, this improvement will enhance our ability to predict fire dynamics, provide early fire warnings, and implement effective mitigation strategies to reduce their impacts on climate, ecosystems and human societies. Declarations Acknowledgments: We are deeply indebted to the data providers and the managers of the FireMIP models output data. We also give our sincere thanks to all data providers listed in Table S1 for continuous efforts and for sharing their data. We thank Xishuangbanna Station for Tropical Rainforest Ecosystem Studies for providing litterfall ground measurements at Bubeng and Menglun sites, Dr. M. Mund for providing litterfall data at Hainich site, and Dr. M. Detto for providing litterfall data at BCI site. Other litterfall field data can be acquired online. Funding: This work was funded by the National Natural Science Foundation of China (42125105), and the National Key Research and Development Program of China (2019YFA0606601 and 2019YFA0606603). J.P. was funded by the TED2021-132627B-I00 grant funded by the Spanish MCIN, AEI/10.13039/501100011033, the European Union NextGenerationEU/PRTR, the Fundación Ramón Areces project CIVP20A6621 and the Catalan government grant SGR221-1333. Author contributions: Y.G.Z. designed the research. G.K.L., Y.G.Z. and C.Y.W. wrote the first draft of the manuscript. G.K.L. performed the analyses and visualization. All authors assessed the research analyses and contributed to the writing of the manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: All data used in this study are available online. The specific links for each observation dataset, FireMIP models' outputs, and CMIP6 outputs are presented in Table S1, Table S2, and Table S3, respectively. References Piao, S. et al. Characteristics, drivers and feedbacks of global greening. Nature Reviews Earth & Environment 1, 14–27 (2020). Zhu, Z. et al. Greening of the Earth and its drivers. Nature Climate Change 6, 791–795 (2016). Chen, C. et al. China and India lead in greening of the world through land-use management. Nat Sustain 2, 122–129 (2019). Canadell, J. G. et al. Multi-decadal increase of forest burned area in Australia is linked to climate change. Nat Commun 12, 6921 (2021). Senande-Rivera, M., Insua-Costa, D. & Miguez-Macho, G. Spatial and temporal expansion of global wildland fire activity in response to climate change. Nature Communications 13 (2022). Cunningham, C. X., Williamson, G. J. & Bowman, D. M. J. S. Increasing frequency and intensity of the most extreme wildfires on Earth. Nature Ecology & Evolution 8, 1420–1425 (2024). Jones, M. W. et al. Global rise in forest fire emissions linked to climate change in the extratropics. Science 386, eadl5889 (2024). Andela, N. et al. A human-driven decline in global burned area. Science 356, 1356–1362 (2017). Jain, P. et al. Drivers and Impacts of the Record-Breaking 2023 Wildfire Season in Canada. Nature Communications 15, 6764 (2024). Byrne, B. et al. Carbon emissions from the 2023 Canadian wildfires. Nature (2024). Jones, M. W. et al. State of Wildfires 2023–2024. Earth Syst. Sci. Data 16, 3601–3685 (2024). Kelly, L. T. et al. Fire and biodiversity in the Anthropocene. Science 370 (2020). Grau-Andrés, R., Moreira, B. & Pausas, J. G. Global plant responses to intensified fire regimes. Global Ecology and Biogeography , e13858 (2024). Zheng, B. et al. Record-high CO2 emissions from boreal fires in 2021. Science 379, 912–917 (2023). van der Velde, I. R. et al. Vast CO2 release from Australian fires in 2019–2020 constrained by satellite. Nature 597, 366–369 (2021). Byrne, B. et al. Unprecedented Canadian forest fire carbon emissions during 2023. PREPRINT (Version 1) available at Research Square (30 November 2023). Bowman, D. et al. Wildfires: Australia needs national monitoring agency. Nature 584, 188–191 (2020). Bowman, D. M. J. S. & Sharples, J. J. Taming the flame, from local to global extreme wildfires. Science 381, 616–619 (2023). Solomon, S. et al. Chlorine activation and enhanced ozone depletion induced by wildfire aerosol. Nature 615, 259–264 (2023). Zhao, J. et al. Forest fire size amplifies postfire land surface warming. Nature 633, 828–834 (2024). Pausas, J. G. & Keeley, J. E. Wildfires and global change. Frontiers in Ecology and the Environment 19, 387–395 (2021). Jolly, W. M. et al. Climate-induced variations in global wildfire danger from 1979 to 2013. Nat Commun 6, 7537 (2015). Jain, P., Castellanos-Acuna, D., Coogan, S. C. P., Abatzoglou, J. T. & Flannigan, M. D. Observed increases in extreme fire weather driven by atmospheric humidity and temperature. Nature Climate Change 12, 63–70 (2022). Descals, A. et al. Unprecedented fire activity above the Arctic Circle linked to rising temperatures. Science 378, 532–537 (2022). Burton, C. et al. Global burned area increasingly explained by climate change. Nature Climate Change (2024). Balch, J. K. et al. Warming weakens the night-time barrier to global fire. Nature 602, 442–448 (2022). Flannigan, M. et al. Global wildland fire season severity in the 21st century. Forest Ecology and Management 294, 54–61 (2013). Pausas, J. G. & Paula, S. Fuel shapes the fire-climate relationship: evidence from Mediterranean ecosystems. Global Ecology and Biogeography 21, 1074–1082 (2012). Bowman, D. M. J. S. et al. Vegetation fires in the Anthropocene. Nature Reviews Earth & Environment 1, 500–515 (2020). Dubinin, M., Luschekina, A. & Radeloff, V. C. Climate, Livestock, and Vegetation: What Drives Fire Increase in the Arid Ecosystems of Southern Russia? Ecosystems 14, 547–562 (2011). Pausas, J. G. & Fernández-Muñoz, S. Fire regime changes in the Western Mediterranean Basin: from fuel-limited to drought-driven fire regime. Climatic Change 110, 215–226 (2011). Covington, W. W. & Moore, M. M. Southwestern Ponderosa Forest Structure: Changes Since Euro-American Settlement. Journal of Forestry 92, 39–47 (1994). Leverkus, A. B., Thorn, S., Lindenmayer, D. B. & Pausas, J. G. Tree planting goals must account for wildfires. Science 376, 588–589 (2022). Hagmann, R. K. et al. Evidence for widespread changes in the structure, composition, and fire regimes of western North American forests. Ecological Applications 31, e02431 (2021). Dodge, M. Forest Fuel Accumulation—A Growing Problem. Science 177, 139–142 (1972). Gómez-González, S., Ojeda, F. & Fernandes, P. M. Portugal and Chile: Longing for sustainable forestry while rising from the ashes. Environmental Science & Policy 81, 104–107 (2018). Jones, M. W. et al. Global and Regional Trends and Drivers of Fire Under Climate Change. Reviews of Geophysics 60, e2020RG000726 (2022). Keenan, T. F. et al. Net carbon uptake has increased through warming-induced changes in temperate forest phenology. Nature Climate Change 4, 598–604 (2014). Chen, J. M. et al. Vegetation structural change since 1981 significantly enhanced the terrestrial carbon sink. Nat Commun 10, 4259 (2019). Pausas, J. G. & Ribeiro, E. The global fire-productivity relationship. Global Ecology and Biogeography 22, 728–736 (2013). Forkel, M. et al. Recent global and regional trends in burned area and their compensating environmental controls. Environmental Research Communications 1 (2019). Pausas, J. G. & Bradstock, R. A. Fire persistence traits of plants along a productivity and disturbance gradient in mediterranean shrublands of south-east Australia. Global Ecology and Biogeography 16, 330–340 (2007). Li, S. et al. Benchmark estimates for aboveground litterfall data derived from ecosystem models. Environmental Research Letters 14 (2019). Alvarado, S. T., Andela, N., Silva, T. S. F., Archibald, S. & Poulter, B. Thresholds of fire response to moisture and fuel load differ between tropical savannas and grasslands across continents. Global Ecology and Biogeography 29, 331–344 (2019). Matthews, S. Dead fuel moisture research: 1991–2012. International Journal of Wildland Fire 23 (2014). van Wees, D. et al. Global biomass burning fuel consumption and emissions at 500 m spatial resolution based on the Global Fire Emissions Database (GFED). Geoscientific Model Development 15, 8411–8437 (2022). Abatzoglou, J. T., Williams, A. P., Boschetti, L., Zubkova, M. & Kolden, C. A. Global patterns of interannual climate-fire relationships. Glob Chang Biol 24, 5164–5175 (2018). Hantson, S. et al. Quantitative assessment of fire and vegetation properties in simulations with fire-enabled vegetation models from the Fire Model Intercomparison Project. Geoscientific Model Development 13, 3299–3318 (2020). Lasslop, G. et al. Global ecosystems and fire: Multi-model assessment of fire‐induced tree‐cover and carbon storage reduction. Global Change Biology 26, 5027–5041 (2020). Forkel, M. et al. Emergent relationships with respect to burned area in global satellite observations and fire-enabled vegetation models. Biogeosciences 16, 57–76 (2019). van der Werf, G. R. et al. Global fire emissions estimates during 1997–2016. Earth System Science Data 9, 697–720 (2017). Resco de Dios, V. et al. A semi-mechanistic model for predicting the moisture content of fine litter. Agricultural and Forest Meteorology 203, 64–73 (2015). Littell, J. S. Drought and Fire in the Western USA: Is Climate Attribution Enough? Current Climate Change Reports 4, 396–406 (2018). Clarke, H. et al. Forest fire threatens global carbon sinks and population centres under rising atmospheric water demand. Nat Commun 13, 7161 (2022). Teckentrup, L. et al. Response of simulated burned area to historical changes in environmental and anthropogenic factors: a comparison of seven fire models. Biogeosciences 16, 3883–3910 (2019). Myers-Smith, I. H. et al. Complexity revealed in the greening of the Arctic. Nature Climate Change 10, 106–117 (2020). Walker, X. J. et al. Increasing wildfires threaten historic carbon sink of boreal forest soils. Nature 572, 520–523 (2019). Fan, L. et al. Siberian carbon sink reduced by forest disturbances. Nature Geoscience 16, 56–62 (2022). Pellegrini, A. F. A. et al. Fire effects on the persistence of soil organic matter and long-term carbon storage. Nature Geoscience 15, 5–13 (2021). Hermoso, V., Regos, A., Moran-Ordonez, A., Duane, A. & Brotons, L. Tree planting: A double-edged sword to fight climate change in an era of megafires. Glob Chang Biol 27, 3001–3003 (2021). Gómez-González, S. et al. Afforestation and climate mitigation: lessons from Chile. Trends in Ecology & Evolution 39, 5–8 (2024). Methods Observation-based global vegetation and fire dynamics The leaf area index (LAI) is an essential structural parameter for the description of plant canopies, often used as a proxy of vegetation greenness 1 ,62 . We used observations from seven satellite-based LAI products over the last two decades to investigate how fire responds to widespread greening. These products are derived from different optical sensors with diverse observational capabilities and methodologies of processing and retrieval. This allowed us to explore the robust relationships between fire activity and vegetation greening that are independent of sensors and retrieval methods. Specifically, we utilized the latest version of Global Land Surface Satellite LAI (GLASS V6, 2001–2020) 63 , the MODIS LAI (MOD15A2H V6, 2001–2020) 64 , the long-term Global Mapping LAI (GLOBMAP V3, 2001–2020) 65 , the latest Global Inventory Modeling and Mapping Studies LAI (GIMMS LAI4g, 2001–2020) 66 , the NOAA Climate Data Record LAI (TCDR V4, 2001–2020) 67 , and the European Geoland2 version 2 LAI derived from SPOT/VEGETATION & PROBA-V (GEOV2-VGT, 2001–2019) 68 and AVHRR (GEOV2-AVHRR, 2001–2019) 69 . These LAI products were averaged to monthly and resampled to 0.25° by the method of bilinear interpolation. Snow contamination significantly degrades the accuracy of satellite LAI products, particularly in the mid-high latitudes of Northern Hemisphere 70 , which will further influence the calculation of LAI dead . Amongst these LAI products, only GLASS and GLOBMAP have processed the snow-contaminated pixels and filled the data gaps 63,71 . Thus, we reconstructed a monthly LAI basis from 2001 to 2020 for the month with snow cover by averaging these two products. The month with snow cover was identified as the month with average air temperature below 0 ℃ 72 based on the ERA5-Land monthly 2 m air temperature dataset. Then, the snow-contaminated LAI values in other five LAI products (MOD15A2H, GIMMS, TCDR, GEOV2-VGT, and GEOV2-AVHRR) were replaced by the reconstructed LAI basis. After that, these monthly LAI products were partitioned to yearly LAI live and LAI dead to represent live and dead fine fuels, respectively (see Derivation of annual LAI live and LAI dead ). We used burned area (BA) to represent global fire activity over the period 2001–2020. The two-decade BA detection was derived from two MODIS-based products, NASA’s standard BA product (MCD64A1 V6) 73 , and European Fire_cci version 5.1 (FireCCI51) 74 . The MCD64A1 V6 provides monthly 500-m global BA observations. We aggregated original monthly 500-m BA records into yearly 0.25° resolution by calculating burned fraction (BF) as the percentage of burned pixels over a whole year within a 0.25° × 0.25° grid cell. We focused on fires that occurred in natural vegetated areas, thus any burned areas detected in croplands and non-vegetated areas were masked. The FireCCI51 provides monthly 250-m global BA observations, which comprise three data layers: date of the first detection (JD), confidence level (CL), and land cover of burned pixels (LC) based on the Land Cover CCI maps. We selected high-confidence (CL > 50%) burned pixels (JD > 0) with natural vegetation covering (excluding croplands and non-vegetated pixels by LC layer) to calculate yearly BF within 0.25° × 0.25° grid cells. We also used another proxy to represent fire intensity, fire radiative power (FRP), which indicates the rate of emission of fire radiative energy and biomass consumption 75 . The FRP was derived from MCD14ML V6 active fire product over the period 2001–2020 76 . This product provides information on hotspot detections, including coordinates, FRP, acquisition time, detection confidence, and fire type. For each year, we allocated these fire detections into global 0.25° × 0.25° grid cells with the WGS84 coordinate system, and averaged FRP to represent FRP per detection. We only retained the fire detections labelled as vegetation fires and with confidence larger than 50%. Thus, we obtained the yearly global grid cell product of FRP per detection with 0.25° resolution for 2001–2020. Climate data The climate conditions during the period 2001–2020 were characterized using the ERA5-Land monthly averaged products with 0.1° resolution provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) 77 . The climatic variables include air temperature (T) and dewpoint temperature (Td) at 2 m above the surface, total precipitation (PRE), and surface soil moisture (SM) at a layer of 0–7 cm. The T and PRE represent background climate conditions. The T and Td were used to calculate vapor pressure deficit (VPD) and relative humidity (RH) following ref. 78 to represent atmospheric water demand and drought. These monthly climatic variables were annually averaged to yearly for T, VPD, RH and SM, and annually accumulated for PRE. We also used monthly maximum temperature (T max ) and climatic water deficit (CWD) derived from TerraClimate dataset over 2001–2020, with a 1/24° spatial resolution 79 . Fire season T max can serve to indicate the extreme hot condition driving the fire occurrence and spread. CWD, calculated as the difference between potential and actual evapotranspiration, is closely related to the water demand and availability that drive fuel flammability 53 . Fire weather index (FWI), obtained from ERA5-based global meteorological wildfire danger dataset 80 , was also used to directly represent fire weather conditions. These climate variables were upscaled to 0.25° using bilinear interpolation. Auxiliary data Results of fire sensitivities to vegetation greening were explored for different climate zones (tropical, arid, temperate, and cold), derived from a Köppen-Geiger climate classification map (Fig. S3) 81 , for different plant functional types (PFTs) (Fig. S18) based on the IGBP classification scheme from the MCD12Q1 V6 land cover type product 82 , and different biomes (Fig. S19) from the Terrestrial Ecoregions of the World 83 . The MCD12Q1 V6 product over the period 2001–2020 was also used to mask the MCD64A1 V6 burned pixels located in croplands and non-vegetated areas. The 20-year (2001–2020) averaged RH derived from ERA5-Land monthly data, and the 30-year (1970–2000) climatological averaged aridity index derived from Global Aridity Index and Potential Evapotranspiration (ET0) Climate Database V2 84 were regarded as the proxies of water availability. The 20-year averaged tree cover component of vegetation continuous field from MOD44B V6.1 product 85 represented vegetation growth. The proxies of water availability and vegetation growth were used to explore the underlying mechanism driving the spatial variability of fire response to vegetation greening. All above data were upscaled to 0.25° using bilinear interpolation, except for Köppen-Geiger climate classification map and MCD12Q1 land cover map, which used the nearest neighbor method. To project global variations of fire activity induced by vegetation greening until the end of the 21st century under future scenarios, we also used monthly LAI (historical: 2000–2014; future: 2015–2100) from three CMIP6 models under SSP126, 245, 370 and 585 scenarios (Fig. S5 and Table S3). Future monthly LAI was aggregated to yearly LAI live and LAI dead . The future projections were conducted under a common 1×1 degree spatial resolution. Derivation of annual LAI live and LAI dead We chose LAI over other indices of biomass (including foliage and woody components) for the following reasons: (ⅰ) LAI is a widely used proxy for vegetation growth, and much evidence for global greening is provided by the widespread, persistent increase in LAI 1 , 2 . (ⅱ) Long-term LAI products allow the investigation of how fire activity responds to vegetation change at spatiotemporal scales that cannot be explored with biomass data. (ⅲ) Different LAI products derived from different sensors and retrieval methods allow for robust estimation of the effects of vegetation greening on fires. (ⅳ) We only consider the leaf parts of live and dead fuels, which can be well estimated from seasonal changes in LAI 1 . Fine fuel is also the most available to fire and the component that drives fires in many ecosystems. (ⅴ) At the inter-annual scale, changes in woody biomass are rather limited and therefore may not be relevant for assessing greening-induced fire changes. (ⅵ) LAI provides information on the short-term dynamics of any type of ecosystems (i.e. changes in leaf biomass), but this is not the case for woody biomass. (ⅶ) Vegetation greenness variables have previously been used as proxies for fuel loads in fire modelling 41 ,86 . From the perspective of pyrogeography, vegetation greening can potentially increase fuel availability and connectivity (both vertical and horizontal). Considering the different fuel components, we proposed a framework to clarify the effects of greening on fire activity: (ⅰ) greening leads to flourishing canopy during the growing season, which increases live foliage biomass and continuity (live fine fuel); (ⅱ) more flourishing leaves lead to more litter and dead grass (dead fine fuel) due to leaf senescence and shedding during the senescence period. Both the enhancements of canopy live foliage and dead fine fuel can influence fire activity. To represent these two components, we calculated LAI live and LAI dead (Extended Data Fig. 1). We considered consecutive two-year LAI seasonal variations. The LAI live was calculated as the maximum three-consecutive-month averaged LAI, which approximately represents the available amount of canopy foliage biomass for burning (live fine fuel). The LAI dead was the difference between maximum and the subsequent minimum three-consecutive-month averaged LAI (Extended Data Fig. 1). This negative seasonal variation of LAI is related to a transfer of living leaf biomass to the litter pool, including the reductions due to natural leaf shedding (e.g. in forests) and annual drying (e.g. grasslands in temperate steppes and tropical savanna ecosystems). That is, the method should capture fuel dynamics in both crown-fire regimes and grass-fueled surface-fire regimes. According to this method, we calculated annual LAI live and LAI dead over the period 2002–2020 (Extended Data Fig. 2 and Fig. S1 ). We used a long-term global litterfall production dataset from 2002–2013 to examine the relationship between LAI dead and litterfall production. We found that LAI dead has a good spatiotemporal consistency with litterfall production over a large part of vegetated area, particularly in high-latitude boreal forest and non-forest areas. But for some tropical, subtropical and temperate forest areas, their relationships were undesired (Fig. S6). Thus, we further collected 6 site-level litterfall ground measurements at forest ecosystems, and found significantly positive relationship between the detrended LAI dead and detrended litterfall measurements across different forest types (see Supplementary Text S1, Fig. S7 and Table S4). These validations demonstrated that LAI dead is a robust relative index that can indicate the inter-annual variations of litterfall rather than its absolute quantity. The detailed datasets and methods for validations were presented in Supplementary Text S1. Sensitivity of fire activity to vegetation greening Fire activity, measured as yearly BF in our study, is not a phenomenon caused by a single factor but rather triggered by many concurrent conditions, including high fuel availability, high fuel aridity, and fire weather. To disentangle the LAI contribution to fire activity from multiple drivers, we used ridge regression at the interannual scale to calculate fire sensitivity to changes in LAI live and LAI dead . Ridge regression is designed to solve the collinearity problem between predictors of ordinary multiple linear regression 87 . Following the method of ref. 62 , we related year-to-year variations of BF (ΔBF) with year-to-year variations of two components of LAI (ΔLAI live and ΔLAI dead ) and background climate (ΔT and ΔPRE) in the attribution model (Eq. 1): $$\:\varDelta\:{BF}_{y}=\:{\beta\:}_{0}+\:\frac{\partial\:BF}{\partial\:{LAI}_{live}}\varDelta\:{LAI}_{live,y}+\frac{\partial\:BF}{\partial\:{LAI}_{dead}}\varDelta\:{LAI}_{dead,y}+\frac{\partial\:BF}{\partial\:T}\varDelta\:{T}_{y}+\frac{\partial\:BF}{\partial\:PRE}\varDelta\:{PRE}_{y}$$ 1 where the Δ operator indicates the difference between two consecutive years (year-to-year variation). This treatment can disentangle the resulting signal from possible long-term dependencies on covariates 88 . We only considered the pixels in which fire changes in two consecutive years when training the attribution model based on long-term series of observation and reanalysis data, thus Δ operator was conducted in three situations: fire◊fire; no fire◊fire; and fire◊no fire. The sensitivities of BF to LAI live (∂BF/∂LAI live ) and LAI dead (∂BF/∂LAI dead ) were calculated as the partial derivatives of the attribution model. The positive (negative) sensitivity means that an increase in canopy live foliage or dead fine fuel amplifies (dampens) fire activity. To obtain a more reliable and spatially consistent estimates by increasing sample size, we used a 9 × 9 spatial moving window (2.25° × 2.25°) to estimate the sensitivities of centered grid cells. In addition to the burned area, we also used FRP in the model to quantify the effect of greening on fire intensity (Extended Data Fig. 3). Long-term fire variations related to long-term trends in LAI (δBF LAI ) were calculated as Eq. 2: $$\:\delta\:{BF}^{LAI}=\frac{\partial\:BF}{\partial\:LAI}\times\:\delta\:LAI$$ 2 where ∂BF/∂LAI is the fire sensitivity to LAI live or LAI dead estimated in Eq. 1, and δLAI is the long-term trend in LAI live or LAI dead , calculated by the Mann–Kendall test and Theil–Sen slope estimator (MK-TS). Similarly, to obtain spatially consistent trends in LAI, we calculated the trend of centered grid cells after averaging LAI within a 9 × 9 spatial moving window (2.25° × 2.25°). The total effects of vegetation greening on fire activity (δBF LAI−total ), including the effects of two processes, were the sum of δBF LAI−live and δBF LAI−dead . When δBF LAI > 0 (< 0), this indicates that long-term trends in LAI over the last two decades amplify (dampen) fire activity. This method was also applied for the climatic variables in the regression model to calculate the BF trends induced by long-term trends in T (δBF T ) and PRE (δBF PRE ) to compare the relative contributions of vegetation greening and climate change. Drivers of spatial variability of ∂BF/ ∂LAI live and ∂BF/ ∂LAI dead According to the intermediate fire–productivity hypothesis, fire activity exhibits a unimodal relationship with productivity 28,40 . This model is regulated by whether fire activity is dominated by fuel productivity (fuel-limited; in low-productivity environments) or fuel moisture (moisture-limited; in high-productivity environments). Inspired by this, we investigated the control effects of water availability (aridity index and RH) and vegetation growth (LAI live and tree cover) on the spatial variability of ∂BF/∂LAI live and ∂BF/∂LAI dead . There was an apparent opposite ∂BF/∂LAI live across the globe (Fig. 1d), thus, we identified the critical thresholds of water availability and vegetation growth that can partition the positive and negative sensitivity. We searched for the optimal threshold based on the criterion that maximizes the product of the proportion of positive sensitivity below the threshold and the proportion of negative sensitivity above the threshold 89 . Multi-product ensemble To make the estimates of fire response to vegetation greening robust and reliable, we developed a multi-product ensemble with eight members that includes any combinations of four LAI (GLASS V6, MOD15A2H V6, GLOBMAP V3 and GIMMS LAI4g) and two BA products (MCD64A1 V6 and FireCCI51) over the period 2001–2020. Fire sensitivities to LAI, fire variations related to long-term LAI trends, and critical thresholds (Fig. S20 and S21) that regulate the spatial variability of ∂BF/∂LAI live were represented by the multi-product ensemble mean. The uncertainties were represented by the standard error across products. Robustness of fire sensitivities to vegetation greening We validated the robustness of our estimates of fire sensitivity to vegetation greening (including LAI live and LAI dead ) through the following aspects: (ⅰ) The simplification of only considering temperature and precipitation as climatic driver in the attribution model (Eq. 1) may overlook the direct effects of fire weather and water deficits on fires. Moreover, the annual aggregated climatic variables may obscure the influence of seasonal climatic factors on fires 53 , such as high temperature and low precipitation during the fire season, which are more directly related to the hot and dry conditions that drives fires. Therefore, we also used fire season T max and PRE (Fig. S8a and b) or FWI and CWD (Fig. S8c and d) as the climatic drivers. The fire season was defined as the three consecutive months centered around the month with maximum burned area (Fig. S8e). (ⅱ) We considered VPD as the climatic driver (Fig. S9). Because VPD can largely explain the variations of extreme fire weather 23 and serve as a reliable predictor of dead fuel moisture content 54 . (ⅲ) We also considered surface soil moisture (indicating soil water supply) and relative humidity (indicating atmospheric water demand) to replace precipitation in the model, as they are more related to water balance deficit that drives fuel flammability (Fig. S10). (ⅳ) We used the model (Eq. 3) that includes the interaction terms between vegetation dynamics and climate (e.g. LAI co-varies with temperature) to validate our estimates independent of the possible interactions among drivers (Fig. S11). Although Eq. 3 is not a strictly nonlinear model that completely describes the interactions in the climate–vegetation–fire system, the result suggests that the impacts of interaction terms are relatively small. (ⅴ) To validate the robustness of estimates from different satellite sensors and observational period, we used LAI products of AVHRR-based TCDR v4 for 2001–2020, and GEOV2-VGT and GEOV2-AVHRR for 2001–2019 to calculate sensitivities (Fig. S12). (ⅵ) To validate the robustness of estimates from different spatial window sizes, we tested 7 × 7 (1.75° × 1.75°) and 11 × 11 (2.75° × 2.75°) spatial moving windows (Fig. S13). (ⅶ) We also calculated LAI live as the maximum monthly LAI, and LAI dead as the difference between maximum and the subsequent minimum monthly LAI (Fig. S14). The results confirmed that our findings were independent of the size of temporal window to calculate LAI live and LAI dead . (ⅷ) We additionally used the bioclimatic spaces to qualitatively analyze fire responses to LAI changes (Supplementary Text S2 and Fig. S15 and S16). (ⅸ) We also used random forest and explainable machine learning (SHapley Additive exPlanations, SHAP) methods, and the consistent findings strengthen the robustness of the results than using traditional statistical method (Supplementary Text S3; Fig. S17). The above validations consistently demonstrated the robustness of our results that fire activity has globally consistent positive responses to increased LAI dead and regional contrasting responses to increased LAI live . FireMIP models To examine whether state-of-the-art fire-vegetation models enable to reproduce the observed sensitivities of global fire activity to vegetation greening, we used the available outputs of monthly net primary productivity (NPP) (not all models provide monthly LAI outputs) and burned fraction from seven models participating in the FireMIP 90 , including CLM, JSBACH-SPITFIRE, LPJ-GUESS-SPITFIRE, ORCHIDEE-SPITFIRE, CTEM, JULES-INFERNO, and LPJ-GUESS-SIMFIRE-BLAZE (Table S2). We used the model output of NPP given its availability and strong positive relationship with LAI, especially in drylands 91 . The FireMIP aims to improve our understanding of fire processes and their representation in global models, and further projections of global fire dynamics and impacts on ecosystems and human societies 48 ,92 . We used the similar method to represent the two processes of vegetation greening influencing fire activity, i.e. increasing live and dead fine fuels, and then used ridge regression to estimate fire sensitivities to greening after eliminating the effects of temperature and total precipitation. The comparisons were conducted during the overlapping period between satellite observations and FireMIP models of 2001–2012. To correspond with the models' outputs, we used the global dataset of satellite LAI data driven NPP during 2001-2012 39,93 and MODIS burned area product to derive satellite-based results. Daily NPP predictions are based on the process-based Boreal Ecosystem Productivity Simulator (BEPS) and driven by satellite-based LAI, clumping index, land cover, meteorological data, atmospheric CO 2 concentration. etc. 39 . The daily NPP was averaged to monthly and then partitioned into yearly live and dead components. Given that there showed strong positive relationships between the fire sensitivities to vegetation greening derived from NPP and those from LAI (Fig. S4), the satellite results from NPP can be used to diagnose the capability of FireMIP models.</p References 62 Forzieri, G., Alkama, R., Miralles, D. G. & Cescatti, A. Satellites reveal contrasting responses of regional climate to the widespread greening of Earth. Science 356 , 1180-1184 (2017). 63 Ma, H. & Liang, S. Development of the GLASS 250-m leaf area index product (version 6) from MODIS data using the bidirectional LSTM deep learning model. Remote Sensing of Environment 273 (2022). 64 Myneni, R., Knyazikhin, Y. & Park, T. MOD15A2H MODIS/Terra Leaf Area Index/FPAR 8-Day L4 Global 500m SIN Grid V006. https://doi.org/10.5067/MODIS/MOD15A2H.006 (2015). 65 Liu, Y., Liu, R. & Chen, J. M. Retrospective retrieval of long-term consistent global leaf area index (1981–2011) from combined AVHRR and MODIS data. Journal of Geophysical Research: Biogeosciences 117 , G04003 (2012). 66 Cao, S. et al. Spatiotemporally consistent global dataset of the GIMMS Leaf Area Index (GIMMS LAI4g) from 1982 to 2020. Earth Syst. Sci. Data Discuss. 2023 , 1-31 (2023). 67 Claverie, M., Matthews, J., Vermote, E. & Justice, C. A 30+ Year AVHRR LAI and FAPAR Climate Data Record: Algorithm Description and Validation. Remote Sensing 8 (2016). 68 Verger, A., Baret, F. & Weiss, M. Near Real-Time Vegetation Monitoring at Global Scale. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 7 , 3473-3481 (2014). 69 Verger, A., Baret, F. & Weiss, M. Algorithm Theoretical Basis Document - GEOV2/AVHRR: Leaf Area Index (LAI), Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) and Fraction of green Vegetation Cover (FCOVER) from LTDR AVHRR. (Available at https://www.theia-land.fr/wp-content/uploads/2022/03/THEIA-SP-44-0207-CREAF_I2.50-1.pdf). (2020). 70 Yan, K. et al. Evaluation of MODIS LAI/FPAR Product Collection 6. Part 2: Validation and Intercomparison. Remote Sensing 8 (2016). 71 Chen, J. M., Feng, D. & Mingzhen, C. Locally adjusted cubic-spline capping for reconstructing seasonal trajectories of a satellite-derived surface parameter. IEEE Transactions on Geoscience and Remote Sensing 44 , 2230-2238 (2006). 72 Wu, C. et al. Increased drought effects on the phenology of autumn leaf senescence. Nature Climate Change 12 , 943-949 (2022). 73 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). 74 Lizundia-Loiola, J., Otón, G., Ramo, R. & Chuvieco, E. A spatio-temporal active-fire clustering approach for global burned area mapping at 250 m from MODIS data. Remote Sensing of Environment 236 (2020). 75 Freeborn, P. H., Wooster, M. J., Roy, D. P. & Cochrane, M. A. Quantification of MODIS fire radiative power (FRP) measurement uncertainty for use in satellite-based active fire characterization and biomass burning estimation. Geophysical Research Letters 41 , 1988-1994 (2014). 76 Giglio, L., Schroeder, W. & Justice, C. O. The collection 6 MODIS active fire detection algorithm and fire products. Remote Sensing of Environment 178 , 31-41 (2016). 77 Muñoz-Sabater, J. et al. ERA5-Land: a state-of-the-art global reanalysis dataset for land applications. Earth System Science Data 13 , 4349-4383 (2021). 78 Yuan, W. et al. Increased atmospheric vapor pressure deficit reduces global vegetation growth. Science Advances 5 , eaax1396 (2019). 79 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). 80 Vitolo, C. et al. ERA5-based global meteorological wildfire danger maps. Scientific Data 7 (2020). 81 Beck, H. E. et al. Present and future Koppen-Geiger climate classification maps at 1-km resolution. Sci Data 5 , 180214 (2018). 82 Sulla-Menashe, D., Gray, J. M., Abercrombie, S. P. & Friedl, M. A. Hierarchical mapping of annual global land cover 2001 to present: The MODIS Collection 6 Land Cover product. Remote Sensing of Environment 222 , 183-194 (2019). 83 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 (2001). 84 Trabucco, A. & Zomer, R. Global Aridity Index and Potential Evapotranspiration (ET0) Climate Database v2. https://doi.org/10.6084/m9.figshare.7504448.v3 (2019). 85 DiMiceli, C. M. et al. Annual global automated MODIS vegetation continuous field (MOD44B) at 250 m spatial resolution for data years beginning day 65, 2000-2014, collection 5 percent tree cover, version 6. https://doi.org/10.5067/MODIS/MOD44B.061 (2017). 86 Knorr, W., Kaminski, T., Arneth, A. & Weber, U. Impact of human population density on fire frequency at the global scale. Biogeosciences 11 , 1085-1102 (2014). 87 Dormann, C. F. et al. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography 36 , 27-46 (2013). 88 Forzieri, G. et al. Increased control of vegetation on global terrestrial energy fluxes. Nature Climate Change 10 , 356-362 (2020). 89 Guan, K. et al. Photosynthetic seasonality of global tropical forests constrained by hydroclimate. Nature Geoscience 8 , 284-289 (2015). 90 Hantson, S. et al. Model outputs: Quantitative assessment of fire and vegetation properties in historical simulations with fire-enabled vegetation models from the Fire Model Intercomparison Project [Data set]. Zenodo https://doi.org/10.5281/zenodo.3555562 (2019). 91 Pan, N., Wang, S., Wei, F., Shen, M. & Fu, B. Inconsistent changes in NPP and LAI determined from the parabolic LAI versus NPP relationship. Ecological Indicators 131 (2021). 92 Rabin, S. S. et al. The Fire Modeling Intercomparison Project (FireMIP), phase 1: experimental and analytical protocols with detailed model descriptions. Geoscientific Model Development 10 , 1175-1197 (2017). 93 He, Q. et al. Drought Risk of Global Terrestrial Gross Primary Productivity Over the Last 40 Years Detected by a Remote Sensing‐Driven Process Model. Journal of Geophysical Research: Biogeosciences 126 (2021). Additional Declarations There is NO Competing Interest. Supplementary Files GreeningOnFireSI.docx Supplementary Information ExtendedDataFig.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5467904","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":391961253,"identity":"1a61c402-2bf9-4864-9756-82dc8f92ef8b","order_by":0,"name":"Yongguang Zhang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAArUlEQVRIiWNgGAWjYHACxgcMPCA6gXgtzAYka2GTgNDEajE4v/xZdYHMYQZ+9hwDhp87iNFy40Ha7Rk8hxkke94YMPaeIUrLgWO3eYBaDG7kGDAzthGl5WBbMUiLPfFazjezMYNtkSBWi+QNNmZpHp50HokzzwoO9hKjhe/88YefeXus5fjbkzc++EmMFoUbCcD474FE5gEiNDAwyPeD1P0gSu0oGAWjYBSMVAAAp3g0J7zbd4AAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0001-8286-300X","institution":"Nanjing University","correspondingAuthor":true,"prefix":"","firstName":"Yongguang","middleName":"","lastName":"Zhang","suffix":""},{"id":391961254,"identity":"a27e7af2-5b60-4bc3-9078-40d4d2ec3bcd","order_by":1,"name":"Gengke Lai","email":"","orcid":"","institution":"Nanjing University","correspondingAuthor":false,"prefix":"","firstName":"Gengke","middleName":"","lastName":"Lai","suffix":""},{"id":391961255,"identity":"63ded160-9943-4784-8db4-e1a7d8ff4cac","order_by":2,"name":"Chaoyang Wu","email":"","orcid":"https://orcid.org/0000-0001-6163-8209","institution":"Institute of Geographic Sciences and Natural Resources Research","correspondingAuthor":false,"prefix":"","firstName":"Chaoyang","middleName":"","lastName":"Wu","suffix":""},{"id":391961256,"identity":"44a7955b-a84a-4f8a-86bb-c031c5d9238e","order_by":3,"name":"Alessandro Cescatti","email":"","orcid":"","institution":"Directorate for Sustainable Resources","correspondingAuthor":false,"prefix":"","firstName":"Alessandro","middleName":"","lastName":"Cescatti","suffix":""},{"id":391961257,"identity":"26519c08-8728-418e-a583-828da1ff7de0","order_by":4,"name":"Gregory Duveiller","email":"","orcid":"https://orcid.org/0000-0002-6471-8404","institution":"Max Planck Institute for Biogeochemistry","correspondingAuthor":false,"prefix":"","firstName":"Gregory","middleName":"","lastName":"Duveiller","suffix":""},{"id":391961258,"identity":"41ebf8f5-4439-4dc4-a6fe-ed4164b159dd","order_by":5,"name":"Juli Pausas","email":"","orcid":"https://orcid.org/0000-0003-3533-5786","institution":"Centro de Investigaciones sobre Desertificatión (CIDE), Consejo Superior de Investigaciones Científicas, Universitat de València","correspondingAuthor":false,"prefix":"","firstName":"Juli","middleName":"","lastName":"Pausas","suffix":""},{"id":391961259,"identity":"2933f758-e460-42f6-8b1a-3ea142c0f8d7","order_by":6,"name":"Stijn Hantson","email":"","orcid":"https://orcid.org/0000-0003-4607-9204","institution":"Universidad del Rosario","correspondingAuthor":false,"prefix":"","firstName":"Stijn","middleName":"","lastName":"Hantson","suffix":""},{"id":391961260,"identity":"ba342f45-2a2e-4a4b-8d80-0ce4182fa804","order_by":7,"name":"Zhaoying Zhang","email":"","orcid":"","institution":"International Institute for Earth System Sciences, Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application, Nanjing University","correspondingAuthor":false,"prefix":"","firstName":"Zhaoying","middleName":"","lastName":"Zhang","suffix":""},{"id":391961261,"identity":"1dc93436-0fff-4870-bd92-556a4e1ee78a","order_by":8,"name":"Adrià Descals","email":"","orcid":"https://orcid.org/0000-0003-1644-3036","institution":"CREAF‐CSIC‐UAB","correspondingAuthor":false,"prefix":"","firstName":"Adrià","middleName":"","lastName":"Descals","suffix":""},{"id":391961262,"identity":"d1f4edf4-71bf-4387-989b-276df6c8dc3e","order_by":9,"name":"min Cao","email":"","orcid":"","institution":"Xishuangbanna Tropical Botanical Garden Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"min","middleName":"","lastName":"Cao","suffix":""},{"id":391961263,"identity":"4962cd5a-849a-40e9-bba7-bdb92eaed020","order_by":10,"name":"Huazheng Lu","email":"","orcid":"","institution":"Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Huazheng","middleName":"","lastName":"Lu","suffix":""},{"id":391961264,"identity":"7f88f350-12be-48b4-9671-490eb02a110c","order_by":11,"name":"Josep Peñuelas","email":"","orcid":"https://orcid.org/0000-0002-7215-0150","institution":"CREAF-CSIC","correspondingAuthor":false,"prefix":"","firstName":"Josep","middleName":"","lastName":"Peñuelas","suffix":""}],"badges":[],"createdAt":"2024-11-17 02:35:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5467904/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5467904/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":71945805,"identity":"cba038fa-639b-4e63-8411-27a694509016","added_by":"auto","created_at":"2024-12-20 03:15:18","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":2503168,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGlobal fire sensitivity to LAI\u003c/strong\u003e\u003csub\u003e\u003cstrong\u003edead\u003c/strong\u003e\u003c/sub\u003e\u003cstrong\u003e (∂BF/∂LAI\u003c/strong\u003e\u003csub\u003e\u003cstrong\u003edead\u003c/strong\u003e\u003c/sub\u003e\u003cstrong\u003e) and LAI\u003c/strong\u003e\u003csub\u003e\u003cstrong\u003elive\u003c/strong\u003e\u003c/sub\u003e\u003cstrong\u003e (∂BF/∂LAI\u003c/strong\u003e\u003csub\u003e\u003cstrong\u003elive\u003c/strong\u003e\u003c/sub\u003e\u003cstrong\u003e) over the period 2001–2020.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e, Global pattern of multi-product ensemble means of ∂BF/∂LAI\u003csub\u003edead\u003c/sub\u003e calculated by ridge regression. Black dots indicate regions with significant sensitivity (\u003cem\u003ep\u003c/em\u003e-value \u0026lt; 0.1). The inset at bottom left indicates ensemble means of median sensitivity for global (red), tropical (green), arid (orange), temperate (pink), and cold (blue) regions. The error bar indicates the standard error across products. \u003cstrong\u003eb\u003c/strong\u003e, Ensemble mean of latitudinal median of ∂BF/∂LAI\u003csub\u003edead\u003c/sub\u003e. The shaded area indicates the standard error across products. \u003cstrong\u003ec\u003c/strong\u003e, Ensemble mean of ∂BF/∂LAI\u003csub\u003edead\u003c/sub\u003e binned as a function of annual mean temperature and total precipitation. \u003cstrong\u003ed–f\u003c/strong\u003e, Same as \u003cstrong\u003ea–c\u003c/strong\u003e but for ∂BF/∂LAI\u003csub\u003elive\u003c/sub\u003e. Climate zones were derived from Köppen-Geiger climate classification map (Fig. S3). The sensitivity was calculated based on the year-to-year variations of burned fraction and driving factors (see Methods).\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-5467904/v1/c39c99059ba4e88c994b335f.png"},{"id":71946401,"identity":"281f8abb-fe11-497d-8c15-743cb2370d81","added_by":"auto","created_at":"2024-12-20 03:23:20","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":3455878,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBurned fraction trend induced by vegetation greening over the period 2001–2020.\u003c/strong\u003e Ensemble mean of trends in burned fraction (BF) associated with long-term changes in LAI\u003csub\u003edead\u003c/sub\u003e (δBF\u003csup\u003eLAI-dead\u003c/sup\u003e, \u003cstrong\u003ea\u003c/strong\u003e) and LAI\u003csub\u003elive\u003c/sub\u003e (δBF\u003csup\u003eLAI-live\u003c/sup\u003e, \u003cstrong\u003eb\u003c/strong\u003e). \u003cstrong\u003ec\u003c/strong\u003e, Total effects of greening on BF (δBF\u003csup\u003eLAI-total\u003c/sup\u003e) calculated by the sum of δBF\u003csup\u003eLAI-live\u003c/sup\u003e and δBF\u003csup\u003eLAI-dead\u003c/sup\u003e. \u003cstrong\u003ed\u003c/strong\u003e, Statistical comparisons of vegetation greening effects on fire for global, tropical, arid, temperate, and cold regions. Bars and error bars indicate the ensemble mean and standard error of median contributions across products, respectively. \u003cstrong\u003ee\u003c/strong\u003e, Comparisons of vegetation greening effects with climate change on fire activity. Bars indicate multi-product ensemble means of median effects of vegetation greening (LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e) and climate change (T and PRE) on fire (δBF\u003csup\u003eX\u003c/sup\u003e) for global, tropical, arid, temperate, and cold regions. Error bars indicate the standard error across products.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-5467904/v1/dc578804b80b656306378578.png"},{"id":71946397,"identity":"5c39708d-24db-4b47-9084-7c41ff671b0f","added_by":"auto","created_at":"2024-12-20 03:23:19","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1239230,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eWater availability and vegetation growth regulate the spatial variabilities of ∂BF/∂LAI\u003c/strong\u003e\u003csub\u003e\u003cstrong\u003edead\u003c/strong\u003e\u003c/sub\u003e\u003cstrong\u003e and ∂BF/∂LAI\u003c/strong\u003e\u003csub\u003e\u003cstrong\u003elive\u003c/strong\u003e\u003c/sub\u003e\u003cstrong\u003e.\u003c/strong\u003e \u003cstrong\u003ea−b\u003c/strong\u003e, PFT-scale relationships between ∂BF/∂LAI\u003csub\u003edead\u003c/sub\u003e and aridity index (\u003cstrong\u003ea\u003c/strong\u003e) and LAI\u003csub\u003elive\u003c/sub\u003e (\u003cstrong\u003eb\u003c/strong\u003e). The points relate ensemble means of median sensitivities with median values of factors for each PFT. The error bars indicate spatial standard deviation of factors (horizontal) and standard error of sensitivities across products (vertical). \u003cstrong\u003ec−d\u003c/strong\u003e, PFT-scale relationships between ∂BF/∂LAI\u003csub\u003elive\u003c/sub\u003e and aridity index (\u003cstrong\u003ec\u003c/strong\u003e) and LAI\u003csub\u003elive\u003c/sub\u003e (\u003cstrong\u003ed\u003c/strong\u003e). The relationships were fitted by natural logarithmic function. \u003cstrong\u003ee−f\u003c/strong\u003e, Threshold of aridity index (\u003cstrong\u003ee\u003c/strong\u003e) or LAI\u003csub\u003elive\u003c/sub\u003e (\u003cstrong\u003ef\u003c/strong\u003e) partitions positive and negative ∂BF/∂LAI\u003csub\u003elive\u003c/sub\u003e. The fractions of positive (red) and negative (blue) sensitivities in each factor bin (50 bins) are shown. Gray indicates the overlapping area. Note that higher aridity index indicates wetter environment.\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-5467904/v1/a8784dfe6a716a6e48238092.png"},{"id":71946387,"identity":"1624c422-aefe-4f86-9dde-57a8748d0ea2","added_by":"auto","created_at":"2024-12-20 03:23:18","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":2007660,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComparisons of fire sensitivities to vegetation greening between satellite observation and seven FireMIP models for tropical (a), arid (b), temperate (c), and cold (d) regions.\u003c/strong\u003e The white bar indicates the satellite result, which was derived from satellite LAI driven BEPS NPP and MODIS burned area products. The colored bars indicate the results from individual FireMIP models. The white bar with stripes denotes the multi-model mean.\u003c/p\u003e","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-5467904/v1/77bcb16d4d0ca0b95fb21044.png"},{"id":71946415,"identity":"8014190a-df02-4f66-813e-aa301a744d00","added_by":"auto","created_at":"2024-12-20 03:23:22","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":2244731,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFuture projections of burned fraction (BF) variations induced by vegetation greening until the end of the 21st century.\u003c/strong\u003e \u003cstrong\u003ea–d\u003c/strong\u003e, Global patterns of BF differences induced by vegetation greening between 2081–2100 and 2001–2020 for SSP126 (\u003cstrong\u003ea\u003c/strong\u003e), SSP245 (\u003cstrong\u003eb\u003c/strong\u003e), SSP370 (\u003cstrong\u003ec\u003c/strong\u003e), and SSP585 (\u003cstrong\u003ed\u003c/strong\u003e) scenarios. ‘+’ and ‘-’ represent amplification and damping, respectively. \u003cstrong\u003ee\u003c/strong\u003e, Average differences in BF induced by vegetation greening between 2081–2100 and 2001–2020 for four climate zones. Error bars indicate the standard error across models. \u003cstrong\u003ef\u003c/strong\u003e, Projected temporal trends of BF variations relative to 2001 induced by greening at four SSP scenarios. Shaded areas indicate the standard deviation across models.\u003c/p\u003e","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-5467904/v1/b794e4cf3a44c8daa6b3f292.png"},{"id":91097308,"identity":"e87841eb-7d62-4938-baf9-4343f6a72173","added_by":"auto","created_at":"2025-09-11 14:15:49","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":13815475,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5467904/v1/6ab94553-335b-4d14-aa2a-ac5b9ef59538.pdf"},{"id":71947080,"identity":"1a1f2ebf-3268-45cf-b735-101fd4ed6ef2","added_by":"auto","created_at":"2024-12-20 03:31:19","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":10600053,"visible":true,"origin":"","legend":"Supplementary Information","description":"","filename":"GreeningOnFireSI.docx","url":"https://assets-eu.researchsquare.com/files/rs-5467904/v1/b0a5cc5dd1f0ed086f83206e.docx"},{"id":71946402,"identity":"dc4eb48f-8f49-4c09-806b-10c22e9b1896","added_by":"auto","created_at":"2024-12-20 03:23:20","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":20130714,"visible":true,"origin":"","legend":"","description":"","filename":"ExtendedDataFig.docx","url":"https://assets-eu.researchsquare.com/files/rs-5467904/v1/e51f86371e8d6f3f3abddabe.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Vegetation greening enhances global fire activity","fulltext":[{"header":"Main Text","content":"\u003cp\u003eIn 2023, Canada experienced a significant rise in mega-fire incidents\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e, closely following the unprecedented wildfire outbreaks in Australia from 2019 to 2020 and Siberia in 2021. Global terrestrial ecosystems have recently experienced a substantial amplification of extreme wildfire events\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e, posing considerable threats to biodiversity\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e, terrestrial carbon storage\u003csup\u003e\u003cspan additionalcitationids=\"CR15\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e, air quality\u003csup\u003e\u003cspan additionalcitationids=\"CR18\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e, and thus regional-to-global climate\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. Wildfires occur when certain conditions are met, including fuel availability and dryness, ignition source and favorable weather conditions\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Both local and regional evidences reveal that climate change has prolonged fire season\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, intensified extreme fire weather conditions\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e, and exacerbated the extent\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e, intensity\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e, severity\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e, and emission\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e of wildfires in many ecosystems worldwide. Moreover, fuel availability also plays a crucial role in shaping fire regimes across many regions, as shown by natural fuel (aridity) gradients\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e and human activity\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e. There are substantial local-to-regional evidences of fire regime changes due to human modifications of fuels by land management\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e or forestry\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e,\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e, such as increased landscape fragmentation and reduced fuel continuity due to agricultural expansions in tropical areas\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e, enhanced fuel build-up driven by rural abandonment in southern Europe\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e, by active fire exclusion in western US\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e, and by forestry plantations in Chile and Portugal\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. However, evidence of fuel role based on how climate change and CO\u003csub\u003e2\u003c/sub\u003e fertilization impact vegetation growth at the global scale remains limited\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Vegetation dynamics can regulate fire-climate interactions, either exacerbating or mitigating fire risk\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. Therefore, understanding the mechanisms that control the response of fire to vegetation dynamics is key to gaining insights into how fire regimes will change under climate change.\u003c/p\u003e \u003cp\u003eIn response to elevated CO\u003csub\u003e2\u003c/sub\u003e concentration, climate change, nitrogen deposition, and land-use change, our planet is greening on large part of vegetated areas, as documented by satellite-based observations of a persistent increase in leaf area index (LAI) over the past four decades\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Vegetation greening can enhance green leaf cover, productivity and connectivity by accelerating photosynthesis and increasing carbon sequestration during the growing season\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e,\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. This greening-induced fuel build-up can increase fire activity in some low-productivity arid regions in which fire regimes are regarded as fuel-limited\u003csup\u003e\u003cspan additionalcitationids=\"CR41\" citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. However, larger green biomass may also create shadier environments (e.g. in forest), making fire more difficult to penetrate and thereby reducing fire activity, especially in moist environments. Additionally, vegetation greening is associated with increased production of dead fuels\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e, leading to an accumulation of dead fine fuel on the ground. This occurs through processes such as leaf senescence and shedding in forest ecosystems and higher grass productivity generating more dead grasses through seasonal drying in savannas. How these biogeochemical processes induced by vegetation greening combine to influence global fire activity, however, remains elusive.\u003c/p\u003e \u003cp\u003eIn this study, we tested whether vegetation greening, including increase in canopy live foliage and dead fine fuel (litter and dead grass) due to senescence and shedding, can exacerbate or mitigate global fire regimes, and which process dominants across global terrestrial ecosystems. We expected that: (ⅰ) at a global scale, most fires are initially fueled by dead fine fuel on the ground before spreading to other components (e.g. live fine fuel in the crown). Therefore, dead fine fuel dominants the greening effects on fire and can amplify global fire activity due to its higher flammability and lower moisture content. (ⅱ) For canopy greening, there are regional contrasting effects on fires, with an amplifying effect in fuel-limited regions due to increased availability and connectivity of live fine fuel (e.g. some arid and non-forest ecosystems), and a dampening effect in moisture-limited regions due to wetter and shadier environments (e.g. some forest ecosystems). This leads to two greening-induced fundamental processes that were tested in this study: increasing crown live fine fuel and ground dead fine fuel.\u003c/p\u003e \u003cp\u003eTo test these hypotheses, we proposed a framework to calculate two indices: LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e. LAI\u003csub\u003elive\u003c/sub\u003e represents the interannual variation of live fine fuel and is defined as the maximum three-consecutive-month averaged LAI. LAI\u003csub\u003edead\u003c/sub\u003e, a relative index for the interannual variation of dead fine fuel, is defined as the reduction of LAI during the senescence period, assuming that negative seasonal variations in LAI reflect the transfer of living leaf biomass to the litter pool and dead grass (Methods; Extended Data Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). We combined an ensemble of satellite-based global burned area [aggregated to grid cell-level burned fraction (BF) with unit of %; Fig. S2], fire radiative power (FRP) and LAI products, as well as long-term climatic variables (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Next, we explored the control effects of water and productivity conditions on determining how fire responds to vegetation greening. Moreover, we examined the fire response to vegetation greening in state-of-the-art fire-vegetation models participating in the Fire Model Intercomparison Project (FireMIP; Table S2) to compare simulations with observation-driven assessments. Based on the estimated fire sensitivities to vegetation greening from satellite observations, we further projected the changes in fire regime induced by long-term vegetation greening until the end of the 21st century under four Shared Socioeconomic Pathway (SSP) scenarios (Table S3).\u003c/p\u003e\n\u003ch3\u003eGlobal fire sensitivity to vegetation greening\u003c/h3\u003e\n\u003cp\u003eOur estimation of fire sensitivity to vegetation greening at each grid cell across the globe (Method) suggests that an increase in LAI\u003csub\u003edead\u003c/sub\u003e magnified BF in ~\u0026thinsp;80% of global vegetated land. Areas with positive fire sensitivity to LAI\u003csub\u003edead\u003c/sub\u003e (\u0026part;BF/\u0026part;LAI\u003csub\u003edead\u003c/sub\u003e) spanned over most of the globe, with an overall sensitivity of 3.94\u0026thinsp;\u0026plusmn;\u0026thinsp;0.53%/m\u003csup\u003e2\u003c/sup\u003e m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea), indicating the widespread amplifying effect of increased LAI\u003csub\u003edead\u003c/sub\u003e on fire activity. Hotspots of positive sensitivity are located in North American boreal area, Eurasian steppe, Siberia, and large arid areas in the southern hemisphere (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). This positive \u0026part;BF/\u0026part;LAI\u003csub\u003edead\u003c/sub\u003e shows a bimodal distribution along the latitudinal gradients, with the strongest effect in the southern hemisphere and the high-latitude boreal area (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb). This global pattern of fire sensitivity to LAI\u003csub\u003edead\u003c/sub\u003e was also supported by using fire radiative power (FRP), indicating that an increase in dead fine fuel also tends to magnify global fire intensity (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea and c).\u003c/p\u003e \u003cp\u003eOn the other hand, our analysis revealed contrasting fire sensitivities to LAI\u003csub\u003elive\u003c/sub\u003e (\u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e) across different regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed). Approximately 47% of global vegetated areas showed positive fire responses to canopy greening, mainly distributed in Alaska, west coast North America, African savanna, and midwestern Australia. In contrast, the sensitivities were opposite in the remaining 53% of vegetated areas, including the high-latitude northern ecosystems and the tropics. This contrasting fire response to increased LAI\u003csub\u003elive\u003c/sub\u003e is also supported by the latitudinal distributions of \u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee) and by using FRP to indicate variations in fire intensity (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb and c).\u003c/p\u003e \u003cp\u003eThese spatial patterns of the interactions between fire activity and LAI\u003csub\u003edead\u003c/sub\u003e and LAI\u003csub\u003elive\u003c/sub\u003e were further confirmed by the climatological gradients of fire sensitivities (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec and f). The consistently positive \u0026part;BF/\u0026part;LAI\u003csub\u003edead\u003c/sub\u003e were observed across climatological gradients, with higher sensitivities at moderate and low precipitation levels [mean annual total precipitation (PRE)\u0026thinsp;\u0026lt;\u0026thinsp;1500 mm] (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec). For LAI\u003csub\u003elive\u003c/sub\u003e, a positive response emerged in warm and dry regions [mean annual air temperature (T)\u0026thinsp;\u0026gt;\u0026thinsp;10 ℃ and PRE\u0026thinsp;\u0026lt;\u0026thinsp;1000 mm], and a negative one in cold or moist areas (T\u0026thinsp;\u0026lt;\u0026thinsp;10 ℃ or PRE\u0026thinsp;\u0026gt;\u0026thinsp;1000 mm) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ef). This finding is supported by a previous study across tropical continents, quantifying the precipitation threshold cross which fire regimes switch from fuel- to moisture-limited\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e. This pattern is also consistent with the intermediate fire-productivity hypothesis\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e suggesting that in warm and dry regions fire is mostly limited by fuel, whose production is positively related to LAI\u003csub\u003elive\u003c/sub\u003e. Conversely, in more humid regions, fire is limited by the occurrence of favorable dry climatic conditions. Higher LAI\u003csub\u003elive\u003c/sub\u003e, which is generally associated with wetter and shadier environments, is typically less favorable for fires.\u003c/p\u003e \u003cp\u003eComparatively, we found stronger fire sensitivities to LAI\u003csub\u003edead\u003c/sub\u003e than LAI\u003csub\u003elive\u003c/sub\u003e in all climate zones (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea and d). Higher positive fire sensitivity to LAI\u003csub\u003edead\u003c/sub\u003e were observed than LAI\u003csub\u003elive\u003c/sub\u003e in arid regions (7.92\u0026thinsp;\u0026plusmn;\u0026thinsp;0.43 vs. 3.25\u0026thinsp;\u0026plusmn;\u0026thinsp;0.64%/m\u003csup\u003e2\u003c/sup\u003e m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e). Additionally, the magnitudes of positive effects of LAI\u003csub\u003edead\u003c/sub\u003e were larger in tropical, temperate, and cold regions (3.01\u0026thinsp;\u0026plusmn;\u0026thinsp;0.52, 2.30\u0026thinsp;\u0026plusmn;\u0026thinsp;0.28, and 4.75\u0026thinsp;\u0026plusmn;\u0026thinsp;0.88%/m\u003csup\u003e2\u003c/sup\u003e m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e, respectively), compared to the dampening effects of canopy greening (\u0026minus;\u0026thinsp;2.04\u0026thinsp;\u0026plusmn;\u0026thinsp;0.32, \u0026minus;\u0026thinsp;1.30\u0026thinsp;\u0026plusmn;\u0026thinsp;0.16, and \u0026minus;\u0026thinsp;1.18\u0026thinsp;\u0026plusmn;\u0026thinsp;1.04%/m\u003csup\u003e2\u003c/sup\u003e m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e, respectively). Overall, the magnitudes of fire responses to the increase in LAI\u003csub\u003edead\u003c/sub\u003e were stronger and more spatially consistent than those to increased LAI\u003csub\u003elive\u003c/sub\u003e. The stronger sensitivity of fire to LAI\u003csub\u003edead\u003c/sub\u003e is likely due to the unidirectional positive effect of this parameter on fire. In fact, the LAI\u003csub\u003edead\u003c/sub\u003e is both a relative index of available fuel amount and of conditions favorable to fire, because the moisture of dead fine fuel is lower and more susceptible to dry weather conditions\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e. On the contrary, LAI\u003csub\u003elive\u003c/sub\u003e is related to two antagonistic processes. On one side it positively correlates with biomass and therefore fuel amount and continuity (positive effect for fire), while on the other side, it relates to shady environments and conditions of higher wetness that are negative for fire. Which of these two antagonistic effects prevails is likely to depend on the local background climate and vegetation, and on the limiting factors for fire occurrence.\u003c/p\u003e \u003cp\u003eWe further compared the \u0026part;BF/\u0026part;LAI\u003csub\u003edead\u003c/sub\u003e and \u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e among different plant functional types (PFTs) and biomes. For all PFTs and biomes, fire activity exhibited stronger responses to LAI\u003csub\u003edead\u003c/sub\u003e than LAI\u003csub\u003elive\u003c/sub\u003e. We also found the strongest positive sensitivities in shrubland. The response in needleleaf forests was stronger than that in broadleaf and mixed forests, and stronger for deciduous than evergreen forests (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea). Moreover, the response was stronger in dry forests than in moist ones, and more pronounced in semi-arid, arid, boreal forest and tundra ecosystems (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb). Additionally, significantly different \u0026part;BF/\u0026part;LAI\u003csub\u003edead\u003c/sub\u003e and \u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e were observed between forest and non-forest ecosystems (\u003cem\u003ep\u003c/em\u003e-value\u0026thinsp;\u0026lt;\u0026thinsp;0.01). For LAI\u003csub\u003edead\u003c/sub\u003e, sensitivity in non-forest was approximately twice that in forest (6.89\u0026thinsp;\u0026plusmn;\u0026thinsp;0.39 vs. 3.90\u0026thinsp;\u0026plusmn;\u0026thinsp;0.73%/m\u003csup\u003e2\u003c/sup\u003e m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e) (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec). BF in forest and non-forest responded to LAI\u003csub\u003elive\u003c/sub\u003e change in opposite ways, with sensitivities of \u0026minus;\u0026thinsp;2.23\u0026thinsp;\u0026plusmn;\u0026thinsp;0.73 and 1.04\u0026thinsp;\u0026plusmn;\u0026thinsp;0.33%/m\u003csup\u003e2\u003c/sup\u003e m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e, respectively (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed). That is in forest, greening generates shadier conditions that inhibit fires, while in open-canopy (non-forest) ecosystems, it contributes to fuel build-up (live and dead) and increases fire activity.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eVegetation greening as a key role in global fire activity\u003c/h2\u003e \u003cp\u003eWe further quantified the actual global trend in fire activity induced by the long-term trends in LAI over the last two decades (Methods). From an ensemble of satellite-based LAI products, we found a widespread greening trend for LAI\u003csub\u003elive\u003c/sub\u003e with a global mean of 0.013 m\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e, and for LAI\u003csub\u003edead\u003c/sub\u003e of 0.011 m\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Correspondingly, the global mean greening-induced trend in burned fraction (δBF\u003csup\u003eLAI\u0026minus;total\u003c/sup\u003e, total effects of LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e) was 0.014\u0026thinsp;\u0026plusmn;\u0026thinsp;0.004% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec and d), although burned area declined during the last two decades (Fig. S2). This strengthening effect was attributed to the imbalance between a larger amplifying effect of LAI\u003csub\u003edead\u003c/sub\u003e (δBF\u003csup\u003eLAI\u0026minus;dead\u003c/sup\u003e, 0.047\u0026thinsp;\u0026plusmn;\u0026thinsp;0.009% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e) and a slightly dampening effect of LAI\u003csub\u003elive\u003c/sub\u003e (δBF\u003csup\u003eLAI\u0026minus;live\u003c/sup\u003e, \u0026minus;\u0026thinsp;0.018\u0026thinsp;\u0026plusmn;\u0026thinsp;0.007% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea-d).\u003c/p\u003e \u003cp\u003eHowever, spatial variabilities emerge across climate zones (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed). The BF trend induced by greening was largest in arid regions (0.034\u0026thinsp;\u0026plusmn;\u0026thinsp;0.011% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e) due to the amplifying effect of both LAI\u003csub\u003edead\u003c/sub\u003e and LAI\u003csub\u003elive\u003c/sub\u003e, followed by cold regions (0.023\u0026thinsp;\u0026plusmn;\u0026thinsp;0.007% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e), especially in Siberia and Alaska regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec). The effect of LAI\u003csub\u003edead\u003c/sub\u003e dominates the amplifying effect across cold areas (0.070\u0026thinsp;\u0026plusmn;\u0026thinsp;0.014% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e), partly offset by the dampening effect of LAI\u003csub\u003elive\u003c/sub\u003e change (\u0026minus;\u0026thinsp;0.020\u0026thinsp;\u0026plusmn;\u0026thinsp;0.016% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e). The amplifying effect of LAI\u003csub\u003edead\u003c/sub\u003e was larger than the dampening effect of LAI\u003csub\u003elive\u003c/sub\u003e in the tropics, resulting in an overall increase in fire activity. The effect of greening on fire was minimum in temperate regions.\u003c/p\u003e \u003cp\u003eThe methodology used to quantify the effects of long-term trends in LAI on BF trend was also applied to each climate factor (T and PRE) to assess the relative contributions of vegetation greening and climate change (Methods). We found that although warming dominated global fire dynamics, increasing global BF at a rate of 0.035\u0026thinsp;\u0026plusmn;\u0026thinsp;0.0005% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e, more than twice the total greening effect (0.014\u0026thinsp;\u0026plusmn;\u0026thinsp;0.004% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e), changes in LAI\u003csub\u003edead\u003c/sub\u003e still played a key role in controlling trends of BF, as evidenced by the dominant contribution of this parameter at global scale (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ee). The contribution of LAI\u003csub\u003elive\u003c/sub\u003e was relatively lower at global level (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ee), resulting from the large spatial complementarity across different climate zones (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). In cold regions, LAI\u003csub\u003edead\u003c/sub\u003e had the greatest effect on the variations of BF, resulting from a combination of high fire sensitivity to LAI\u003csub\u003edead\u003c/sub\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea) and large increased trend in LAI\u003csub\u003edead\u003c/sub\u003e (Extend Data Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed). This may be indirectly supported by the fact that in boreal forest and tundra, litter and soil organic materials, whose accumulations are related to annual transfer from live biomass to dead (LAI\u003csub\u003edead\u003c/sub\u003e in this study), account for more than 80% of fuel consumption during fires\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eDrivers of spatial variabilities of ∂BF/∂LAI and ∂BF/∂LAI\u003c/h3\u003e\n\u003cp\u003eIn order to interpret the spatial variability of the fire sensitivity to LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e, we explored the underlying mechanisms that explain the divergent responses of fire activity (sign and magnitude) to vegetation change (LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e) across different regions. The possible drivers include water availability, quantified by aridity index and relative humidity (RH), and vegetation growth, quantified by LAI\u003csub\u003elive\u003c/sub\u003e and tree cover. We found significant negative correlations between \u0026part;BF/\u0026part;LAI\u003csub\u003edead\u003c/sub\u003e and aridity index (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.88, \u003cem\u003ep\u003c/em\u003e-value\u0026thinsp;\u0026lt;\u0026thinsp;0.01) and LAI\u003csub\u003elive\u003c/sub\u003e (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.94, \u003cem\u003ep\u003c/em\u003e-value\u0026thinsp;\u0026lt;\u0026thinsp;0.01) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea and b). Similar relationships were observed for RH and tree cover (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). These results reveal a stronger positive fire response to LAI\u003csub\u003edead\u003c/sub\u003e in arid and low-productivity areas than that in moist and high-productivity areas.\u003c/p\u003e \u003cp\u003eWe also found that water availability and vegetation growth had a strong capability to regulate the spatial variabilities of \u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e, revealed by the significant negative correlations between \u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e and aridity index and LAI\u003csub\u003elive\u003c/sub\u003e (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.70 and \u0026minus;\u0026thinsp;0.78, respectively, both \u003cem\u003ep\u003c/em\u003e-value\u0026thinsp;\u0026lt;\u0026thinsp;0.01) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec and d). These controlling effects were further confirmed by the thresholds of aridity index (0.364\u0026thinsp;\u0026plusmn;\u0026thinsp;0.002) and LAI\u003csub\u003elive\u003c/sub\u003e (2.25\u0026thinsp;\u0026plusmn;\u0026thinsp;0.035 m\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e) that can partition positive and negative \u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee and f). These findings were further supported by RH and tree cover (Extended Data Fig.\u0026nbsp;6). These results revealed that an increased LAI\u003csub\u003elive\u003c/sub\u003e could increase fire activity in arid and low-productivity areas (fuel build-up limited), while the contrary happens in moist and high-productivity areas (fuel-moisture limited). This finding is supported by the intermediate fire-productivity hypothesis, which describes changes in fire activity along with productivity and moisture gradients regulated by critical thresholds\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e,\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e,\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e. The observed thresholds of water availability and productivity are relevant for projecting the changes in fire regimes from fuel build-up limited to fuel-moisture limited or vice versa under greening and warming scenarios.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eDiagnosis of global fire sensitivity to vegetation greening in FireMIP models\u003c/h3\u003e\n\u003cp\u003eFire sensitivities to vegetation greening derived from satellite observations are a valuable benchmark to diagnose the ability of models to simulate the fire-vegetation interactions. We thus investigated whether state-of-the-art fire-vegetation models participating in the FireMIP can capture the observed fire sensitivities to vegetation greening (including increases in canopy live foliage and dead fine fuel). As not all models provide monthly LAI outputs, we used alternative net primary productivity (NPP) outputs from models to depict vegetation greening. The sensitivities from satellite were derived from satellite LAI driven BEPS NPP and MODIS burned area products. Given that there presented strong positive relationships between the sensitivities from NPP and those from LAI (Fig. S4), the satellite results from NPP can be used to diagnose the capability of models.\u003c/p\u003e \u003cp\u003eWe found that fire-vegetation models captured the direction of fire response to change in dead fine fuel with positive sensitivities over four climate zones, but underestimated their magnitudes on average (multi-model mean vs. satellite) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), particularly in arid and cold regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb and d). The models misrepresented the fire sensitivity to changes in canopy live foliage in arid regions with four out of seven models showing negative responses (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb), which may result from the misrepresentation of drylands LAI in the models\u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. Moreover, all models underestimated the magnitude of fire responses to both canopy live foliage and dead fine fuel in high-latitude cold regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed). These discrepancies between model simulations and observations may result from the failure of models in reproducing the magnitude and spatial pattern of interannual variability of burned area\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. Models also have difficulty in replicating the length of the growing season\u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e, which may bias the representations of accumulated live and dead fuels. Amongst seven FireMIP models, the CLM overestimated the fire responses to vegetation greening in tropical and temperate regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea and c). A recent study also found that CLM exhibited the largest fire impacts on the carbon cycle\u003csup\u003e\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e. These evidences suggest that CLM may be too sensitive to changes in terrestrial carbon storage. Overall, the FireMIP models underestimated the interactions between fire activity and vegetation greening, particularly in arid and cold regions. This finding is supported by an evaluation showing that FireMIP models generally underestimate the sensitivities to pre-season vegetation productivity in semi-arid ecosystems\u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eProjection of future greening-induced fire changes\u003c/h3\u003e\n\u003cp\u003eGiven the estimated fire sensitivities to LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e from satellite observations (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea and d) and the ongoing persistent greening projected for the future (Fig. S5)\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e, we further projected the variations of fire activity (indicated by BF) induced by vegetation greening until the end of the 21st century using four SSP scenarios (SSP126, 245, 370, and 585) of LAI, considering two processes of greening effect (live and dead fine fuels) (Methods). Assuming constant sensitivities for the next decades, we found a projected increase of greening-induced fire activity in 65% of global vegetated areas over 2081\u0026ndash;2100 compared with 2001\u0026ndash;2020, and increasingly severe from low to high emission scenarios (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea\u0026ndash;d). The arid and cold regions were projected to experience larger amplifications with BF increase of 0.67\u0026thinsp;\u0026plusmn;\u0026thinsp;0.18% and 0.96\u0026thinsp;\u0026plusmn;\u0026thinsp;0.17% at SSP126, respectively, and correspondingly of 2.28\u0026thinsp;\u0026plusmn;\u0026thinsp;0.89% and 2.17\u0026thinsp;\u0026plusmn;\u0026thinsp;0.36% at SSP585 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee). Moreover, vegetation greening could lead to increased BF at a rate of 0.010\u0026ndash;0.021% y\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e during 2015\u0026ndash;2100 under the four SSP scenarios (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ef).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eTowards understanding the fire-greening feedback\u003c/h3\u003e\n\u003cp\u003eOur study offers insights into how global fire activity responds to vegetation dynamics driven by global change factors (CO\u003csub\u003e2\u003c/sub\u003e fertilization, climate change, and land use change, etc.). We provide evidence of a positive feedback between vegetation greening and fire, which is contributing to the intensification of fire regimes, increasing both the area burned (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) and the intensity of fires (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Elevated atmospheric CO\u003csub\u003e2\u003c/sub\u003e can enhance plant growth and litter production\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Therefore, the greening effects on fire are achieved through two processes, i.e. increasing live and dead fine fuels (represented by LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e, respectively). More greening during the growing season results in more dead litters and dry grasses due to senescence and shedding during the senescence period (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb and d). Our results highlight the importance of CO\u003csub\u003e2\u003c/sub\u003e fertilization and climate change-induced modification of vegetation growth, and consequently fuels, for predicting changes in fire regimes.\u003c/p\u003e \u003cp\u003eWe argue that LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e represent live and dead fine fuels, and do not include coarse woody components such as stem, branch, and bark. Fine fuels, either alive (foliage) or dead (litter and dry grass), are the most flammable components, driving fire activity and emission across many ecosystems worldwide\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e. Moreover, many factors may induce uncertainties in the LAI\u003csub\u003edead\u003c/sub\u003e calculation. The seasonal reduction of LAI can also be caused by disturbances or human activities, such as insect hazards, grazing, and timber harvest, which may not result in the formation of dead fuels. Despite these limitations, we found a strong effect of LAI\u003csub\u003edead\u003c/sub\u003e changes in fire activity. In addition, LAI\u003csub\u003edead\u003c/sub\u003e is a relative index indicating the interannual variation of dead fine fuel rather than its absolute quantity, which is demonstrated by comparing with global litterfall production dataset (Fig. S6) and site-level ground litterfall measurements (Fig. S7 and Table S4). The natural formation of dead fine fuel is related to leaf area and longevity, turnover time, the rate of decomposition, and other biogeochemical processes\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e,\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e\u003c/sup\u003e. Improving the representation of these processes may benefit the modeling of dead fuel accumulation in fire-enable dynamics global vegetation models.\u003c/p\u003e \u003cp\u003eAlthough temperature and precipitation can partly explain the variations of fire weather and foliar and dead fuel moisture\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u003c/sup\u003e, this oversimplification in the attribution model (see Methods) may overlook the direct effects of fire weather and water deficits in the soil and atmosphere. Additionally, the annual aggregated climate variables may obscure the influence of seasonal climate factors on fires\u003csup\u003e\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e, such as high temperature and low precipitation during the fire season, which are more directly related to the hot and dry conditions that drives fires. Therefore, we also used fire season maximum temperature and precipitation as the climatic drivers and obtained robust consistent findings (Fig. S8a and b). Additionally, we directly employed the fire weather index (FWI) and climatic water deficit (CWD) during the fire season, with CWD calculated as the difference between potential and actual evapotranspiration and more closely related to the water deficit that determines fuel flammability\u003csup\u003e\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e. The results remained unchanged (Fig. S8c and d). Vapor pressure deficit (VPD) can largely explain the changes in extreme FWI\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e and serve as a reliable predictor of dead fuel moisture content\u003csup\u003e\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e (Fig. S9). We also induced surface soil moisture (indicating soil water supply) and relative humidity (indicating atmospheric water demand) into the regression models, and obtained robust results (Fig. S10).\u003c/p\u003e \u003cp\u003eOur findings were also robust when using nonlinear multiple regression model (Fig. S11), different sources of satellite LAI data (Fig. S12), using different spatial moving windows to increase sample size of model training (Fig. S13), as well as different size of temporal window to calculate LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e (Fig. S14). We also constructed the bioclimatic spaces to qualitatively analyze the fire responses to vegetation greening and the control effects of moisture and productivity on their spatial variabilities (Fig. S15 and S16). Additionally, we also estimated the sensitivities using random forest and explainable machine learning (SHapley Additive exPlanations, SHAP), which further strengthens the robustness of the results obtained with traditional statistical approaches (Fig. S17).\u003c/p\u003e \u003cp\u003eAs the fire-enabled dynamics global vegetation models in the FireMIP are designed for future burned area projections under climate change, their abilities are crucial for future projections of fire dynamics and impacts and terrestrial carbon budget, depending on how to represent relationships between climate, vegetation, socio-economics and burned area in the models. Our findings provide a benchmark to examine whether the models can replicate the fire-greening interactions from satellite observations. Previous analysis using the sensitivity runs of FireMIP models with CO\u003csub\u003e2\u003c/sub\u003e concentration fixed indicated that most models did not show a clear fire response to CO\u003csub\u003e2\u003c/sub\u003e fertilization, except for LPJ-GUESS-SPITFIRE and JSBACH-SPITFIRE showing that CO\u003csub\u003e2\u003c/sub\u003e fertilization considerably contributed to an increase in burned area\u003csup\u003e\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e\u003c/sup\u003e. As CO\u003csub\u003e2\u003c/sub\u003e fertilization impacts fire activity mainly through altering vegetation dynamics, the sensitivity runs of models can indirectly support our findings that current models underestimate the fire-greening interactions. Altogether, the biases shown by models go in the direction of underestimating the positive feedback between greening and fire, potentially leading to an underestimation of future fire regimes and an overoptimistic projection of the future terrestrial carbon budget.\u003c/p\u003e \u003cp\u003eVegetation greening is one of the highly credible evidence of anthropogenic climate change\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Our study provides a new perspective to understanding the impact of vegetation greening, i.e. potential intensification of fire regimes (area burnt and intensity), which has received relatively less attention in previous studies. Global burned area declined on average over the past two decades, primarily driven by human-induced declines in savannas\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. However, in many other ecosystems, burned area is increasing such as the western America and boreal forests\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e, in which vegetation greening plays an important role driving the amplification of fire activity. Our results call on the urgent need to monitor ecosystems with a potential strong coupling between greening and fire, such as high-latitude northern ecosystems\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. The Arctic tundra is experiencing greening that is associated with the shift of vegetation composition, such as tundra shrub expansion\u003csup\u003e\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e\u003c/sup\u003e. This expansion may amplify fire activity because of the largest positive sensitivities to LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e of shrubland compared with other PFTs (Extended Data Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea). Altogether, greening-induced intensification of fire regimes may threaten the historic carbon stored in vegetation, soil and peatland in boreal forest and Arctic tundra ecosystems, and release large amounts of carbon into the atmosphere\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e,\u003cspan additionalcitationids=\"CR58\" citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e\u003c/sup\u003e. This mechanism may indeed be considered as a positive feedback in the climate system that is partially offsetting the negative feedback driven by the fertilization effect of CO\u003csub\u003e2\u003c/sub\u003e on greening and primary productivity.\u003c/p\u003e \u003cp\u003eRelatedly, the massive afforestation actions committing to mitigate climate change (e.g. Bonn Challenge, Great Green Wall, and the Three Billion Tree Pledge, etc.) have contributed to the vegetation greening, especially in dry areas and savannas\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e,\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e,\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e\u003c/sup\u003e. Our results suggest that these actions could also produce counterproductive side effects, that is, potential amplification of fire regimes and a consequent increase of the vulnerability of the land carbon stock\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e,\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e,\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e\u003c/sup\u003e. Our study reinforces that these afforestation actions would require extensive forest management (e.g. fuel management, especially dead fine fuel) to provide any beneficial effects, which is not guaranteed in low-income regions where most afforestation are foreseen. The strategies of fuel management (e.g. prescribed burning) should be environmentally sustainable and optimized to maximize the benefits of fuel reduction to mitigate fire hazard, while minimizing the adverse effects on ecosystem services (e.g. carbon sequestration and soil functioning) and biodiversity\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn summary, we conclude that while global burned area has been declining primarily driven by agricultural expansion and intensification over the past two decades, vegetation greening is contributing to an intensification of global fire regimes in addition to climate warming. This amplification is primarily attributed to the increased dead fine fuel, which can counteract the potential dampening effect of increased canopy live foliage. Vegetation greening has the potential to enhance leaf cover and connectivity, and it also leads to an increase in dead fine fuel due to leaf and grass senescence. These biogeochemical feedbacks from vegetation greening could further exacerbate the fire regime changes driven by climate change. As vegetation effect on fire is identified as a main deficiency of current fire-vegetation models\u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e, future efforts should focus on better characterizing the processes linking fire activity and vegetation dynamics in data-driven and process-based fire models to improve predictions on future trajectories of the terrestrial carbon budget. In addition, this improvement will enhance our ability to predict fire dynamics, provide early fire warnings, and implement effective mitigation strategies to reduce their impacts on climate, ecosystems and human societies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgments:\u003c/strong\u003e We are deeply indebted to the data providers and the managers of the FireMIP models output data. We also give our sincere thanks to all data providers listed in Table S1 for continuous efforts and for sharing their data. We thank Xishuangbanna Station for Tropical Rainforest Ecosystem Studies for providing litterfall ground measurements at Bubeng and Menglun sites, Dr. M. Mund for providing litterfall data at Hainich site, and Dr. M. Detto for providing litterfall data at BCI site. Other litterfall field data can be acquired online.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This work was funded by\u0026nbsp;the National Natural Science Foundation of China (42125105), and\u0026nbsp;the National Key Research and Development Program of China (2019YFA0606601 and 2019YFA0606603). J.P. was funded by the TED2021-132627B-I00 grant funded by the Spanish MCIN, AEI/10.13039/501100011033, the European Union NextGenerationEU/PRTR, the Fundación Ramón Areces project CIVP20A6621 and the Catalan government grant SGR221-1333.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions:\u003c/strong\u003e Y.G.Z. designed the research. G.K.L., Y.G.Z. and C.Y.W. wrote the first draft of the manuscript. G.K.L. performed the analyses and visualization. All authors assessed the research analyses and contributed to the writing of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests:\u003c/strong\u003e The authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData and materials availability:\u003c/strong\u003e All data used in this study are available online. The specific links for each observation dataset, FireMIP models' outputs, and CMIP6 outputs are presented in Table S1, Table S2, and Table S3, respectively.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003ePiao, S. \u003cem\u003eet al.\u003c/em\u003e Characteristics, drivers and feedbacks of global greening. Nature Reviews Earth \u0026amp; Environment 1, 14\u0026ndash;27 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu, Z. \u003cem\u003eet al.\u003c/em\u003e Greening of the Earth and its drivers. Nature Climate Change 6, 791\u0026ndash;795 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, C. \u003cem\u003eet al.\u003c/em\u003e China and India lead in greening of the world through land-use management. Nat Sustain 2, 122\u0026ndash;129 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCanadell, J. G. \u003cem\u003eet al.\u003c/em\u003e Multi-decadal increase of forest burned area in Australia is linked to climate change. Nat Commun 12, 6921 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSenande-Rivera, M., Insua-Costa, D. \u0026amp; Miguez-Macho, G. Spatial and temporal expansion of global wildland fire activity in response to climate change. Nature Communications 13 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCunningham, C. X., Williamson, G. J. \u0026amp; Bowman, D. M. J. S. Increasing frequency and intensity of the most extreme wildfires on Earth. Nature Ecology \u0026amp; Evolution 8, 1420\u0026ndash;1425 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJones, M. W. \u003cem\u003eet al.\u003c/em\u003e Global rise in forest fire emissions linked to climate change in the extratropics. Science 386, eadl5889 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAndela, N. \u003cem\u003eet al.\u003c/em\u003e A human-driven decline in global burned area. Science 356, 1356\u0026ndash;1362 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJain, P. \u003cem\u003eet al.\u003c/em\u003e Drivers and Impacts of the Record-Breaking 2023 Wildfire Season in Canada. Nature Communications 15, 6764 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eByrne, B. \u003cem\u003eet al.\u003c/em\u003e Carbon emissions from the 2023 Canadian wildfires. Nature (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJones, M. W. \u003cem\u003eet al.\u003c/em\u003e State of Wildfires 2023\u0026ndash;2024. Earth Syst. Sci. Data 16, 3601\u0026ndash;3685 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKelly, L. T. \u003cem\u003eet al.\u003c/em\u003e Fire and biodiversity in the Anthropocene. Science 370 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGrau-Andr\u0026eacute;s, R., Moreira, B. \u0026amp; Pausas, J. G. Global plant responses to intensified fire regimes. \u003cem\u003eGlobal Ecology and Biogeography\u003c/em\u003e, e13858 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng, B. \u003cem\u003eet al.\u003c/em\u003e Record-high CO2 emissions from boreal fires in 2021. Science 379, 912\u0026ndash;917 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan der Velde, I. R. \u003cem\u003eet al.\u003c/em\u003e Vast CO2 release from Australian fires in 2019\u0026ndash;2020 constrained by satellite. Nature 597, 366\u0026ndash;369 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eByrne, B. \u003cem\u003eet al.\u003c/em\u003e Unprecedented Canadian forest fire carbon emissions during 2023. \u003cem\u003ePREPRINT (Version 1) available at Research Square\u003c/em\u003e (30 November 2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBowman, D. \u003cem\u003eet al.\u003c/em\u003e Wildfires: Australia needs national monitoring agency. Nature 584, 188\u0026ndash;191 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBowman, D. M. J. S. \u0026amp; Sharples, J. J. Taming the flame, from local to global extreme wildfires. Science 381, 616\u0026ndash;619 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSolomon, S. \u003cem\u003eet al.\u003c/em\u003e Chlorine activation and enhanced ozone depletion induced by wildfire aerosol. Nature 615, 259\u0026ndash;264 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao, J. \u003cem\u003eet al.\u003c/em\u003e Forest fire size amplifies postfire land surface warming. Nature 633, 828\u0026ndash;834 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePausas, J. G. \u0026amp; Keeley, J. E. Wildfires and global change. Frontiers in Ecology and the Environment 19, 387\u0026ndash;395 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJolly, W. M. \u003cem\u003eet al.\u003c/em\u003e Climate-induced variations in global wildfire danger from 1979 to 2013. Nat Commun 6, 7537 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJain, P., Castellanos-Acuna, D., Coogan, S. C. P., Abatzoglou, J. T. \u0026amp; Flannigan, M. D. Observed increases in extreme fire weather driven by atmospheric humidity and temperature. Nature Climate Change 12, 63\u0026ndash;70 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDescals, A. \u003cem\u003eet al.\u003c/em\u003e Unprecedented fire activity above the Arctic Circle linked to rising temperatures. Science 378, 532\u0026ndash;537 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurton, C. \u003cem\u003eet al.\u003c/em\u003e Global burned area increasingly explained by climate change. Nature Climate Change (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBalch, J. K. \u003cem\u003eet al.\u003c/em\u003e Warming weakens the night-time barrier to global fire. Nature 602, 442\u0026ndash;448 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFlannigan, M. \u003cem\u003eet al.\u003c/em\u003e Global wildland fire season severity in the 21st century. Forest Ecology and Management 294, 54\u0026ndash;61 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePausas, J. G. \u0026amp; Paula, S. Fuel shapes the fire-climate relationship: evidence from Mediterranean ecosystems. Global Ecology and Biogeography 21, 1074\u0026ndash;1082 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBowman, D. M. J. S. \u003cem\u003eet al.\u003c/em\u003e Vegetation fires in the Anthropocene. Nature Reviews Earth \u0026amp; Environment 1, 500\u0026ndash;515 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDubinin, M., Luschekina, A. \u0026amp; Radeloff, V. C. Climate, Livestock, and Vegetation: What Drives Fire Increase in the Arid Ecosystems of Southern Russia? \u003cem\u003eEcosystems\u003c/em\u003e 14, 547\u0026ndash;562 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePausas, J. G. \u0026amp; Fern\u0026aacute;ndez-Mu\u0026ntilde;oz, S. Fire regime changes in the Western Mediterranean Basin: from fuel-limited to drought-driven fire regime. Climatic Change 110, 215\u0026ndash;226 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCovington, W. W. \u0026amp; Moore, M. M. Southwestern Ponderosa Forest Structure: Changes Since Euro-American Settlement. Journal of Forestry 92, 39\u0026ndash;47 (1994).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeverkus, A. B., Thorn, S., Lindenmayer, D. B. \u0026amp; Pausas, J. G. Tree planting goals must account for wildfires. Science 376, 588\u0026ndash;589 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHagmann, R. K. \u003cem\u003eet al.\u003c/em\u003e Evidence for widespread changes in the structure, composition, and fire regimes of western North American forests. Ecological Applications 31, e02431 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDodge, M. Forest Fuel Accumulation\u0026mdash;A Growing Problem. Science 177, 139\u0026ndash;142 (1972).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG\u0026oacute;mez-Gonz\u0026aacute;lez, S., Ojeda, F. \u0026amp; Fernandes, P. M. Portugal and Chile: Longing for sustainable forestry while rising from the ashes. Environmental Science \u0026amp; Policy 81, 104\u0026ndash;107 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJones, M. W. \u003cem\u003eet al.\u003c/em\u003e Global and Regional Trends and Drivers of Fire Under Climate Change. Reviews of Geophysics 60, e2020RG000726 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKeenan, T. F. \u003cem\u003eet al.\u003c/em\u003e Net carbon uptake has increased through warming-induced changes in temperate forest phenology. Nature Climate Change 4, 598\u0026ndash;604 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, J. M. \u003cem\u003eet al.\u003c/em\u003e Vegetation structural change since 1981 significantly enhanced the terrestrial carbon sink. Nat Commun 10, 4259 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePausas, J. G. \u0026amp; Ribeiro, E. The global fire-productivity relationship. Global Ecology and Biogeography 22, 728\u0026ndash;736 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eForkel, M. \u003cem\u003eet al.\u003c/em\u003e Recent global and regional trends in burned area and their compensating environmental controls. Environmental Research Communications 1 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePausas, J. G. \u0026amp; Bradstock, R. A. Fire persistence traits of plants along a productivity and disturbance gradient in mediterranean shrublands of south-east Australia. Global Ecology and Biogeography 16, 330\u0026ndash;340 (2007).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, S. \u003cem\u003eet al.\u003c/em\u003e Benchmark estimates for aboveground litterfall data derived from ecosystem models. Environmental Research Letters 14 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlvarado, S. T., Andela, N., Silva, T. S. F., Archibald, S. \u0026amp; Poulter, B. Thresholds of fire response to moisture and fuel load differ between tropical savannas and grasslands across continents. Global Ecology and Biogeography 29, 331\u0026ndash;344 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatthews, S. Dead fuel moisture research: 1991\u0026ndash;2012. International Journal of Wildland Fire 23 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan Wees, D. \u003cem\u003eet al.\u003c/em\u003e Global biomass burning fuel consumption and emissions at 500 m spatial resolution based on the Global Fire Emissions Database (GFED). Geoscientific Model Development 15, 8411\u0026ndash;8437 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbatzoglou, J. T., Williams, A. P., Boschetti, L., Zubkova, M. \u0026amp; Kolden, C. A. Global patterns of interannual climate-fire relationships. Glob Chang Biol 24, 5164\u0026ndash;5175 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHantson, S. \u003cem\u003eet al.\u003c/em\u003e Quantitative assessment of fire and vegetation properties in simulations with fire-enabled vegetation models from the Fire Model Intercomparison Project. Geoscientific Model Development 13, 3299\u0026ndash;3318 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLasslop, G. \u003cem\u003eet al.\u003c/em\u003e Global ecosystems and fire: Multi-model assessment of fire‐induced tree‐cover and carbon storage reduction. Global Change Biology 26, 5027\u0026ndash;5041 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eForkel, M. \u003cem\u003eet al.\u003c/em\u003e Emergent relationships with respect to burned area in global satellite observations and fire-enabled vegetation models. Biogeosciences 16, 57\u0026ndash;76 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan der Werf, G. R. \u003cem\u003eet al.\u003c/em\u003e Global fire emissions estimates during 1997\u0026ndash;2016. Earth System Science Data 9, 697\u0026ndash;720 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eResco de Dios, V. \u003cem\u003eet al.\u003c/em\u003e A semi-mechanistic model for predicting the moisture content of fine litter. Agricultural and Forest Meteorology 203, 64\u0026ndash;73 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLittell, J. S. Drought and Fire in the Western USA: Is Climate Attribution Enough? Current Climate Change Reports 4, 396\u0026ndash;406 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eClarke, H. \u003cem\u003eet al.\u003c/em\u003e Forest fire threatens global carbon sinks and population centres under rising atmospheric water demand. Nat Commun 13, 7161 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTeckentrup, L. \u003cem\u003eet al.\u003c/em\u003e Response of simulated burned area to historical changes in environmental and anthropogenic factors: a comparison of seven fire models. Biogeosciences 16, 3883\u0026ndash;3910 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMyers-Smith, I. H. \u003cem\u003eet al.\u003c/em\u003e Complexity revealed in the greening of the Arctic. Nature Climate Change 10, 106\u0026ndash;117 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWalker, X. J. \u003cem\u003eet al.\u003c/em\u003e Increasing wildfires threaten historic carbon sink of boreal forest soils. Nature 572, 520\u0026ndash;523 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFan, L. \u003cem\u003eet al.\u003c/em\u003e Siberian carbon sink reduced by forest disturbances. Nature Geoscience 16, 56\u0026ndash;62 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePellegrini, A. F. A. \u003cem\u003eet al.\u003c/em\u003e Fire effects on the persistence of soil organic matter and long-term carbon storage. Nature Geoscience 15, 5\u0026ndash;13 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHermoso, V., Regos, A., Moran-Ordonez, A., Duane, A. \u0026amp; Brotons, L. Tree planting: A double-edged sword to fight climate change in an era of megafires. Glob Chang Biol 27, 3001\u0026ndash;3003 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG\u0026oacute;mez-Gonz\u0026aacute;lez, S. \u003cem\u003eet al.\u003c/em\u003e Afforestation and climate mitigation: lessons from Chile. Trends in Ecology \u0026amp; Evolution 39, 5\u0026ndash;8 (2024).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eObservation-based global vegetation and fire dynamics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe leaf area index (LAI) is an essential structural parameter for the description of plant canopies, often used as a proxy of vegetation greenness\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e1\u003c/span\u003e,62\u003c/sup\u003e. We used observations from seven satellite-based LAI products over the last two decades to investigate how fire responds to widespread greening. These products are derived from different optical sensors with diverse observational capabilities and methodologies of processing and retrieval. This allowed us to explore the robust relationships between fire activity and vegetation greening that are independent of sensors and retrieval methods. Specifically, we utilized the latest version of Global Land Surface Satellite LAI (GLASS V6, 2001\u0026ndash;2020)\u003csup\u003e63\u003c/sup\u003e, the MODIS LAI (MOD15A2H V6, 2001\u0026ndash;2020)\u003csup\u003e64\u003c/sup\u003e, the long-term Global Mapping LAI (GLOBMAP V3, 2001\u0026ndash;2020)\u003csup\u003e65\u003c/sup\u003e, the latest Global Inventory Modeling and Mapping Studies LAI (GIMMS LAI4g, 2001\u0026ndash;2020)\u003csup\u003e66\u003c/sup\u003e, the NOAA Climate Data Record LAI (TCDR V4, 2001\u0026ndash;2020)\u003csup\u003e67\u003c/sup\u003e, and the European Geoland2 version 2 LAI derived from SPOT/VEGETATION \u0026amp; PROBA-V (GEOV2-VGT, 2001\u0026ndash;2019)\u003csup\u003e68\u003c/sup\u003e and AVHRR (GEOV2-AVHRR, 2001\u0026ndash;2019)\u003csup\u003e69\u003c/sup\u003e. These LAI products were averaged to monthly and resampled to 0.25\u0026deg; by the method of bilinear interpolation.\u003c/p\u003e\n\u003cp\u003eSnow contamination significantly degrades the accuracy of satellite LAI products, particularly in the mid-high latitudes of Northern Hemisphere\u003csup\u003e70\u003c/sup\u003e, which will further influence the calculation of LAI\u003csub\u003edead\u003c/sub\u003e. Amongst these LAI products, only GLASS and GLOBMAP have processed the snow-contaminated pixels and filled the data gaps\u003csup\u003e63,71\u003c/sup\u003e. Thus, we reconstructed a monthly LAI basis from 2001 to 2020 for the month with snow cover by averaging these two products. The month with snow cover was identified as the month with average air temperature below 0 ℃\u003csup\u003e72\u003c/sup\u003e based on the ERA5-Land monthly 2 m air temperature dataset. Then, the snow-contaminated LAI values in other five LAI products (MOD15A2H, GIMMS, TCDR, GEOV2-VGT, and GEOV2-AVHRR) were replaced by the reconstructed LAI basis. After that, these monthly LAI products were partitioned to yearly LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e to represent live and dead fine fuels, respectively (see Derivation of annual LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e).\u003c/p\u003e\n\u003cp\u003eWe used burned area (BA) to represent global fire activity over the period 2001\u0026ndash;2020. The two-decade BA detection was derived from two MODIS-based products, NASA\u0026rsquo;s standard BA product (MCD64A1 V6)\u003csup\u003e73\u003c/sup\u003e, and European Fire_cci version 5.1 (FireCCI51)\u003csup\u003e74\u003c/sup\u003e. The MCD64A1 V6 provides monthly 500-m global BA observations. We aggregated original monthly 500-m BA records into yearly 0.25\u0026deg; resolution by calculating burned fraction (BF) as the percentage of burned pixels over a whole year within a 0.25\u0026deg; \u0026times; 0.25\u0026deg; grid cell. We focused on fires that occurred in natural vegetated areas, thus any burned areas detected in croplands and non-vegetated areas were masked. The FireCCI51 provides monthly 250-m global BA observations, which comprise three data layers: date of the first detection (JD), confidence level (CL), and land cover of burned pixels (LC) based on the Land Cover CCI maps. We selected high-confidence (CL\u0026thinsp;\u0026gt;\u0026thinsp;50%) burned pixels (JD\u0026thinsp;\u0026gt;\u0026thinsp;0) with natural vegetation covering (excluding croplands and non-vegetated pixels by LC layer) to calculate yearly BF within 0.25\u0026deg; \u0026times; 0.25\u0026deg; grid cells.\u003c/p\u003e\n\u003cp\u003eWe also used another proxy to represent fire intensity, fire radiative power (FRP), which indicates the rate of emission of fire radiative energy and biomass consumption\u003csup\u003e75\u003c/sup\u003e. The FRP was derived from MCD14ML V6 active fire product over the period 2001\u0026ndash;2020\u003csup\u003e76\u003c/sup\u003e. This product provides information on hotspot detections, including coordinates, FRP, acquisition time, detection confidence, and fire type. For each year, we allocated these fire detections into global 0.25\u0026deg; \u0026times; 0.25\u0026deg; grid cells with the WGS84 coordinate system, and averaged FRP to represent FRP per detection. We only retained the fire detections labelled as vegetation fires and with confidence larger than 50%. Thus, we obtained the yearly global grid cell product of FRP per detection with 0.25\u0026deg; resolution for 2001\u0026ndash;2020.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClimate data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe climate conditions during the period 2001\u0026ndash;2020 were characterized using the ERA5-Land monthly averaged products with 0.1\u0026deg; resolution provided by the European Centre for Medium-Range Weather Forecasts (ECMWF)\u003csup\u003e77\u003c/sup\u003e. The climatic variables include air temperature (T) and dewpoint temperature (Td) at 2 m above the surface, total precipitation (PRE), and surface soil moisture (SM) at a layer of 0\u0026ndash;7 cm. The T and PRE represent background climate conditions. The T and Td were used to calculate vapor pressure deficit (VPD) and relative humidity (RH) following ref.\u003csup\u003e78\u003c/sup\u003e to represent atmospheric water demand and drought. These monthly climatic variables were annually averaged to yearly for T, VPD, RH and SM, and annually accumulated for PRE.\u003c/p\u003e\n\u003cp\u003eWe also used monthly maximum temperature (T\u003csub\u003emax\u003c/sub\u003e) and climatic water deficit (CWD) derived from TerraClimate dataset over 2001\u0026ndash;2020, with a 1/24\u0026deg; spatial resolution\u003csup\u003e79\u003c/sup\u003e. Fire season T\u003csub\u003emax\u003c/sub\u003e can serve to indicate the extreme hot condition driving the fire occurrence and spread. CWD, calculated as the difference between potential and actual evapotranspiration, is closely related to the water demand and availability that drive fuel flammability\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e. Fire weather index (FWI), obtained from ERA5-based global meteorological wildfire danger dataset\u003csup\u003e80\u003c/sup\u003e, was also used to directly represent fire weather conditions. These climate variables were upscaled to 0.25\u0026deg; using bilinear interpolation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuxiliary data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eResults of fire sensitivities to vegetation greening were explored for different climate zones (tropical, arid, temperate, and cold), derived from a K\u0026ouml;ppen-Geiger climate classification map (Fig. S3)\u003csup\u003e81\u003c/sup\u003e, for different plant functional types (PFTs) (Fig. S18) based on the IGBP classification scheme from the MCD12Q1 V6 land cover type product\u003csup\u003e82\u003c/sup\u003e, and different biomes (Fig. S19) from the Terrestrial Ecoregions of the World\u003csup\u003e83\u003c/sup\u003e. The MCD12Q1 V6 product over the period 2001\u0026ndash;2020 was also used to mask the MCD64A1 V6 burned pixels located in croplands and non-vegetated areas.\u003c/p\u003e\n\u003cp\u003eThe 20-year (2001\u0026ndash;2020) averaged RH derived from ERA5-Land monthly data, and the 30-year (1970\u0026ndash;2000) climatological averaged aridity index derived from Global Aridity Index and Potential Evapotranspiration (ET0) Climate Database V2\u003csup\u003e84\u003c/sup\u003e were regarded as the proxies of water availability. The 20-year averaged tree cover component of vegetation continuous field from MOD44B V6.1 product\u003csup\u003e85\u003c/sup\u003e represented vegetation growth. The proxies of water availability and vegetation growth were used to explore the underlying mechanism driving the spatial variability of fire response to vegetation greening. All above data were upscaled to 0.25\u0026deg; using bilinear interpolation, except for K\u0026ouml;ppen-Geiger climate classification map and MCD12Q1 land cover map, which used the nearest neighbor method.\u003c/p\u003e\n\u003cp\u003eTo project global variations of fire activity induced by vegetation greening until the end of the 21st century under future scenarios, we also used monthly LAI (historical: 2000\u0026ndash;2014; future: 2015\u0026ndash;2100) from three CMIP6 models under SSP126, 245, 370 and 585 scenarios (Fig. S5 and Table S3). Future monthly LAI was aggregated to yearly LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e. The future projections were conducted under a common 1\u0026times;1 degree spatial resolution.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDerivation of annual LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe chose LAI over other indices of biomass (including foliage and woody components) for the following reasons: (ⅰ) LAI is a widely used proxy for vegetation growth, and much evidence for global greening is provided by the widespread, persistent increase in LAI\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. (ⅱ) Long-term LAI products allow the investigation of how fire activity responds to vegetation change at spatiotemporal scales that cannot be explored with biomass data. (ⅲ) Different LAI products derived from different sensors and retrieval methods allow for robust estimation of the effects of vegetation greening on fires. (ⅳ) We only consider the leaf parts of live and dead fuels, which can be well estimated from seasonal changes in LAI\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Fine fuel is also the most available to fire and the component that drives fires in many ecosystems. (ⅴ) At the inter-annual scale, changes in woody biomass are rather limited and therefore may not be relevant for assessing greening-induced fire changes. (ⅵ) LAI provides information on the short-term dynamics of any type of ecosystems (i.e. changes in leaf biomass), but this is not the case for woody biomass. (ⅶ) Vegetation greenness variables have previously been used as proxies for fuel loads in fire modelling\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e41\u003c/span\u003e,86\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eFrom the perspective of pyrogeography, vegetation greening can potentially increase fuel availability and connectivity (both vertical and horizontal). Considering the different fuel components, we proposed a framework to clarify the effects of greening on fire activity: (ⅰ) greening leads to flourishing canopy during the growing season, which increases live foliage biomass and continuity (live fine fuel); (ⅱ) more flourishing leaves lead to more litter and dead grass (dead fine fuel) due to leaf senescence and shedding during the senescence period. Both the enhancements of canopy live foliage and dead fine fuel can influence fire activity.\u003c/p\u003e\n\u003cp\u003eTo represent these two components, we calculated LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e (Extended Data Fig. 1). We considered consecutive two-year LAI seasonal variations. The LAI\u003csub\u003elive\u003c/sub\u003e was calculated as the maximum three-consecutive-month averaged LAI, which approximately represents the available amount of canopy foliage biomass for burning (live fine fuel). The LAI\u003csub\u003edead\u003c/sub\u003e was the difference between maximum and the subsequent minimum three-consecutive-month averaged LAI (Extended Data Fig. 1). This negative seasonal variation of LAI is related to a transfer of living leaf biomass to the litter pool, including the reductions due to natural leaf shedding (e.g. in forests) and annual drying (e.g. grasslands in temperate steppes and tropical savanna ecosystems). That is, the method should capture fuel dynamics in both crown-fire regimes and grass-fueled surface-fire regimes. According to this method, we calculated annual LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e over the period 2002\u0026ndash;2020 (Extended Data Fig. 2 and Fig. \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e).\u003c/p\u003e\n\u003cp\u003eWe used a long-term global litterfall production dataset from 2002\u0026ndash;2013 to examine the relationship between LAI\u003csub\u003edead\u003c/sub\u003e and litterfall production. We found that LAI\u003csub\u003edead\u003c/sub\u003e has a good spatiotemporal consistency with litterfall production over a large part of vegetated area, particularly in high-latitude boreal forest and non-forest areas. But for some tropical, subtropical and temperate forest areas, their relationships were undesired (Fig. S6). Thus, we further collected 6 site-level litterfall ground measurements at forest ecosystems, and found significantly positive relationship between the detrended LAI\u003csub\u003edead\u003c/sub\u003e and detrended litterfall measurements across different forest types (see Supplementary Text S1, Fig. S7 and Table S4). These validations demonstrated that LAI\u003csub\u003edead\u003c/sub\u003e is a robust relative index that can indicate the inter-annual variations of litterfall rather than its absolute quantity. The detailed datasets and methods for validations were presented in Supplementary Text S1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSensitivity of fire activity to vegetation greening\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFire activity, measured as yearly BF in our study, is not a phenomenon caused by a single factor but rather triggered by many concurrent conditions, including high fuel availability, high fuel aridity, and fire weather. To disentangle the LAI contribution to fire activity from multiple drivers, we used ridge regression at the interannual scale to calculate fire sensitivity to changes in LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e. Ridge regression is designed to solve the collinearity problem between predictors of ordinary multiple linear regression\u003csup\u003e87\u003c/sup\u003e. Following the method of ref.\u003csup\u003e62\u003c/sup\u003e, we related year-to-year variations of BF (\u0026Delta;BF) with year-to-year variations of two components of LAI (\u0026Delta;LAI\u003csub\u003elive\u003c/sub\u003e and \u0026Delta;LAI\u003csub\u003edead\u003c/sub\u003e) and background climate (\u0026Delta;T and \u0026Delta;PRE) in the attribution model (Eq. 1):\u003c/p\u003e\n\u003cdiv id=\"Equ1\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e$$\\:\\varDelta\\:{BF}_{y}=\\:{\\beta\\:}_{0}+\\:\\frac{\\partial\\:BF}{\\partial\\:{LAI}_{live}}\\varDelta\\:{LAI}_{live,y}+\\frac{\\partial\\:BF}{\\partial\\:{LAI}_{dead}}\\varDelta\\:{LAI}_{dead,y}+\\frac{\\partial\\:BF}{\\partial\\:T}\\varDelta\\:{T}_{y}+\\frac{\\partial\\:BF}{\\partial\\:PRE}\\varDelta\\:{PRE}_{y}$$\u003c/div\u003e\n \u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\n\u003c/div\u003e\n\u003cp\u003ewhere the \u0026Delta; operator indicates the difference between two consecutive years (year-to-year variation). This treatment can disentangle the resulting signal from possible long-term dependencies on covariates\u003csup\u003e88\u003c/sup\u003e. We only considered the pixels in which fire changes in two consecutive years when training the attribution model based on long-term series of observation and reanalysis data, thus \u0026Delta; operator was conducted in three situations: fire\u0026loz;fire; no fire\u0026loz;fire; and fire\u0026loz;no fire. The sensitivities of BF to LAI\u003csub\u003elive\u003c/sub\u003e (\u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e) and LAI\u003csub\u003edead\u003c/sub\u003e (\u0026part;BF/\u0026part;LAI\u003csub\u003edead\u003c/sub\u003e) were calculated as the partial derivatives of the attribution model. The positive (negative) sensitivity means that an increase in canopy live foliage or dead fine fuel amplifies (dampens) fire activity. To obtain a more reliable and spatially consistent estimates by increasing sample size, we used a 9 \u0026times; 9 spatial moving window (2.25\u0026deg; \u0026times; 2.25\u0026deg;) to estimate the sensitivities of centered grid cells. In addition to the burned area, we also used FRP in the model to quantify the effect of greening on fire intensity (Extended Data Fig. 3).\u003c/p\u003e\n\u003cp\u003eLong-term fire variations related to long-term trends in LAI (\u0026delta;BF\u003csup\u003eLAI\u003c/sup\u003e) were calculated as Eq. 2:\u003c/p\u003e\n\u003cdiv id=\"Equ2\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ2\" name=\"EquationSource\"\u003e$$\\:\\delta\\:{BF}^{LAI}=\\frac{\\partial\\:BF}{\\partial\\:LAI}\\times\\:\\delta\\:LAI$$\u003c/div\u003e\n \u003cdiv class=\"EquationNumber\"\u003e2\u003c/div\u003e\n\u003c/div\u003e\n\u003cp\u003ewhere \u0026part;BF/\u0026part;LAI is the fire sensitivity to LAI\u003csub\u003elive\u003c/sub\u003e or LAI\u003csub\u003edead\u003c/sub\u003e estimated in Eq. 1, and \u0026delta;LAI is the long-term trend in LAI\u003csub\u003elive\u003c/sub\u003e or LAI\u003csub\u003edead\u003c/sub\u003e, calculated by the Mann\u0026ndash;Kendall test and Theil\u0026ndash;Sen slope estimator (MK-TS). Similarly, to obtain spatially consistent trends in LAI, we calculated the trend of centered grid cells after averaging LAI within a 9 \u0026times; 9 spatial moving window (2.25\u0026deg; \u0026times; 2.25\u0026deg;). The total effects of vegetation greening on fire activity (\u0026delta;BF\u003csup\u003eLAI\u0026minus;total\u003c/sup\u003e), including the effects of two processes, were the sum of \u0026delta;BF\u003csup\u003eLAI\u0026minus;live\u003c/sup\u003e and \u0026delta;BF\u003csup\u003eLAI\u0026minus;dead\u003c/sup\u003e. When \u0026delta;BF\u003csup\u003eLAI\u003c/sup\u003e \u0026gt; 0 (\u0026lt;\u0026thinsp;0), this indicates that long-term trends in LAI over the last two decades amplify (dampen) fire activity. This method was also applied for the climatic variables in the regression model to calculate the BF trends induced by long-term trends in T (\u0026delta;BF\u003csup\u003eT\u003c/sup\u003e) and PRE (\u0026delta;BF\u003csup\u003ePRE\u003c/sup\u003e) to compare the relative contributions of vegetation greening and climate change.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDrivers of spatial variability of\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e\u0026part;BF/\u003c/strong\u003e\u003cstrong\u003e\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e and\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e\u0026part;BF/\u003c/strong\u003e\u003cstrong\u003e\u0026part;LAI\u003csub\u003edead\u003c/sub\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAccording to the intermediate fire\u0026ndash;productivity hypothesis, fire activity exhibits a unimodal relationship with productivity\u003csup\u003e28,40\u003c/sup\u003e. This model is regulated by whether fire activity is dominated by fuel productivity (fuel-limited; in low-productivity environments) or fuel moisture (moisture-limited; in high-productivity environments). Inspired by this, we investigated the control effects of water availability (aridity index and RH) and vegetation growth (LAI\u003csub\u003elive\u0026nbsp;\u003c/sub\u003eand tree cover) on the spatial variability of\u0026nbsp;\u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e and \u0026part;BF/\u0026part;LAI\u003csub\u003edead\u003c/sub\u003e.\u003c/p\u003e\n\u003cp\u003eThere was an apparent opposite\u0026nbsp;\u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e across the globe (Fig. 1d), thus, we identified the critical thresholds of water availability and vegetation growth that can partition the positive and negative sensitivity. We searched for the optimal threshold based on the criterion that maximizes the product of the proportion of positive sensitivity below the threshold and the proportion of negative sensitivity above the threshold\u003csup\u003e89\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMulti-product ensemble\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo make the estimates of fire response to vegetation greening robust and reliable, we developed a multi-product ensemble with eight members that includes any combinations of four LAI (GLASS V6, MOD15A2H V6, GLOBMAP V3 and GIMMS LAI4g) and two BA products (MCD64A1 V6 and FireCCI51) over the period 2001\u0026ndash;2020. Fire sensitivities to LAI, fire variations related to long-term LAI trends, and critical thresholds (Fig. S20 and S21) that regulate the spatial variability of\u0026nbsp;\u0026part;BF/\u0026part;LAI\u003csub\u003elive\u003c/sub\u003e were represented by the multi-product ensemble mean. The uncertainties were represented by the standard error across products.\u003c/p\u003e\n\u003cp\u003e\u003cspan type=\"BoldUnderline\" class=\"BoldUnderline\" name=\"Emphasis\"\u003e\u003cstrong\u003eRobustness of fire sensitivities to vegetation greening\u003c/strong\u003e\u003c/span\u003e\u003c/p\u003e\n\u003cp\u003eWe validated the robustness of our estimates of fire sensitivity to vegetation greening (including LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e) through the following aspects: (ⅰ) The simplification of only considering temperature and precipitation as climatic driver in the attribution model (Eq.\u0026nbsp;1) may overlook the direct effects of fire weather and water deficits on fires. Moreover, the annual aggregated climatic variables may obscure the influence of seasonal climatic factors on fires\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e, such as high temperature and low precipitation during the fire season, which are more directly related to the hot and dry conditions that drives fires. Therefore, we also used fire season T\u003csub\u003emax\u003c/sub\u003e and PRE (Fig. S8a and b) or FWI and CWD (Fig. S8c and d) as the climatic drivers. The fire season was defined as the three consecutive months centered around the month with maximum burned area (Fig. S8e). (ⅱ) We considered VPD as the climatic driver (Fig. S9). Because VPD can largely explain the variations of extreme fire weather\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e and serve as a reliable predictor of dead fuel moisture content\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e. (ⅲ) We also considered surface soil moisture (indicating soil water supply) and relative humidity (indicating atmospheric water demand) to replace precipitation in the model, as they are more related to water balance deficit that drives fuel flammability (Fig. S10). (ⅳ) We used the model (Eq.\u0026nbsp;3) that includes the interaction terms between vegetation dynamics and climate (e.g. LAI co-varies with temperature) to validate our estimates independent of the possible interactions among drivers (Fig. S11). Although Eq.\u0026nbsp;3 is not a strictly nonlinear model that completely describes the interactions in the climate\u0026ndash;vegetation\u0026ndash;fire system, the result suggests that the impacts of interaction terms are relatively small. (ⅴ) To validate the robustness of estimates from different satellite sensors and observational period, we used LAI products of AVHRR-based TCDR v4 for 2001\u0026ndash;2020, and GEOV2-VGT and GEOV2-AVHRR for 2001\u0026ndash;2019 to calculate sensitivities (Fig. S12). (ⅵ) To validate the robustness of estimates from different spatial window sizes, we tested 7 \u0026times; 7 (1.75\u0026deg; \u0026times; 1.75\u0026deg;) and 11 \u0026times; 11 (2.75\u0026deg; \u0026times; 2.75\u0026deg;) spatial moving windows (Fig. S13). (ⅶ) We also calculated LAI\u003csub\u003elive\u003c/sub\u003e as the maximum monthly LAI, and LAI\u003csub\u003edead\u003c/sub\u003e as the difference between maximum and the subsequent minimum monthly LAI (Fig. S14). The results confirmed that our findings were independent of the size of temporal window to calculate LAI\u003csub\u003elive\u003c/sub\u003e and LAI\u003csub\u003edead\u003c/sub\u003e. (ⅷ) We additionally used the bioclimatic spaces to qualitatively analyze fire responses to LAI changes (Supplementary Text S2 and Fig. S15 and S16). (ⅸ) We also used random forest and explainable machine learning (SHapley Additive exPlanations, SHAP) methods, and the consistent findings strengthen the robustness of the results than using traditional statistical method (Supplementary Text S3; Fig. S17). The above validations consistently demonstrated the robustness of our results that fire activity has globally consistent positive responses to increased LAI\u003csub\u003edead\u003c/sub\u003e and regional contrasting responses to increased LAI\u003csub\u003elive\u003c/sub\u003e.\u003c/p\u003e\n\u003cdiv id=\"Equ3\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ3\" name=\"EquationSource\"\u003e\u003cimg src=\"https://myfiles.space/user_files/58894_9946feeafa4c1df7/58894_custom_files/img1734586317.png\" width=\"752\" height=\"69\"\u003e\u003cbr\u003e\u003c/div\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cspan type=\"BoldUnderline\" class=\"BoldUnderline\" name=\"Emphasis\"\u003e\u003cstrong\u003eFireMIP models\u003c/strong\u003e\u003c/span\u003e\u003c/p\u003e\n\u003cp\u003eTo examine whether state-of-the-art fire-vegetation models enable to reproduce the observed sensitivities of global fire activity to vegetation greening, we used the available outputs of monthly net primary productivity (NPP) (not all models provide monthly LAI outputs) and burned fraction from seven models participating in the FireMIP\u003csup\u003e90\u003c/sup\u003e, including CLM, JSBACH-SPITFIRE, LPJ-GUESS-SPITFIRE, ORCHIDEE-SPITFIRE, CTEM, JULES-INFERNO, and LPJ-GUESS-SIMFIRE-BLAZE (Table S2). We used the model output of NPP given its availability and strong positive relationship with LAI, especially in drylands\u003csup\u003e91\u003c/sup\u003e. The FireMIP aims to improve our understanding of fire processes and their representation in global models, and further projections of global fire dynamics and impacts on ecosystems and human societies\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e48\u003c/span\u003e,92\u003c/sup\u003e. We used the similar method to represent the two processes of vegetation greening influencing fire activity, i.e. increasing live and dead fine fuels, and then used ridge regression to estimate fire sensitivities to greening after eliminating the effects of temperature and total precipitation. The comparisons were conducted during the overlapping period between satellite observations and FireMIP models of 2001\u0026ndash;2012.\u003c/p\u003e\n\u003cp\u003eTo correspond with the models\u0026apos; outputs, we used the global dataset of satellite LAI data driven NPP during 2001-2012\u003csup\u003e39,93\u003c/sup\u003e and MODIS burned area product to derive satellite-based results. Daily NPP predictions are based on the process-based Boreal Ecosystem Productivity Simulator (BEPS) and driven by satellite-based LAI, clumping index, land cover, meteorological data, atmospheric CO\u003csub\u003e2\u003c/sub\u003e concentration. etc.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. The daily NPP was averaged to monthly and then partitioned into yearly live and dead components. Given that there showed strong positive relationships between the fire sensitivities to vegetation greening derived from NPP and those from LAI (Fig. S4), the satellite results from NPP can be used to diagnose the capability of FireMIP models.\u003c/p\n\u003cp\u003e\u003cb\u003eReferences\u003c/b\u003e\u003c/p\u003e\n\u003cp\u003e62 \u0026nbsp; \u0026nbsp;Forzieri, G., Alkama, R., Miralles, D. G. \u0026amp; Cescatti, A. Satellites reveal contrasting responses of regional climate to the widespread greening of Earth. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e356\u003c/strong\u003e, 1180-1184 (2017).\u003c/p\u003e\n\u003cp\u003e63 \u0026nbsp; \u0026nbsp;Ma, H. \u0026amp; Liang, S. Development of the GLASS 250-m leaf area index product (version 6) from MODIS data using the bidirectional LSTM deep learning model. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e \u003cstrong\u003e273\u003c/strong\u003e (2022).\u003c/p\u003e\n\u003cp\u003e64 \u0026nbsp; \u0026nbsp;Myneni, R., Knyazikhin, Y. \u0026amp; Park, T. MOD15A2H MODIS/Terra Leaf Area Index/FPAR 8-Day L4 Global 500m SIN Grid V006. https://doi.org/10.5067/MODIS/MOD15A2H.006 (2015).\u003c/p\u003e\n\u003cp\u003e65 \u0026nbsp; \u0026nbsp;Liu, Y., Liu, R. \u0026amp; Chen, J. M. Retrospective retrieval of long-term consistent global leaf area index (1981\u0026ndash;2011) from combined AVHRR and MODIS data. \u003cem\u003eJournal of Geophysical Research: Biogeosciences\u003c/em\u003e \u003cstrong\u003e117\u003c/strong\u003e, G04003 (2012).\u003c/p\u003e\n\u003cp\u003e66 \u0026nbsp; \u0026nbsp;Cao, S.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Spatiotemporally consistent global dataset of the GIMMS Leaf Area Index (GIMMS LAI4g) from 1982 to 2020. \u003cem\u003eEarth Syst. Sci. Data Discuss.\u003c/em\u003e \u003cstrong\u003e2023\u003c/strong\u003e, 1-31 (2023).\u003c/p\u003e\n\u003cp\u003e67 \u0026nbsp; \u0026nbsp;Claverie, M., Matthews, J., Vermote, E. \u0026amp; Justice, C. A 30+ Year AVHRR LAI and FAPAR Climate Data Record: Algorithm Description and Validation. \u003cem\u003eRemote Sensing\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e (2016).\u003c/p\u003e\n\u003cp\u003e68 \u0026nbsp; \u0026nbsp;Verger, A., Baret, F. \u0026amp; Weiss, M. Near Real-Time Vegetation Monitoring at Global Scale. \u003cem\u003eIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 3473-3481 (2014).\u003c/p\u003e\n\u003cp\u003e69 \u0026nbsp; \u0026nbsp;Verger, A., Baret, F. \u0026amp; Weiss, M. Algorithm Theoretical Basis Document - GEOV2/AVHRR: Leaf Area Index (LAI), Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) and Fraction of green Vegetation Cover (FCOVER) from LTDR AVHRR. (Available at https://www.theia-land.fr/wp-content/uploads/2022/03/THEIA-SP-44-0207-CREAF_I2.50-1.pdf). \u0026nbsp;(2020).\u003c/p\u003e\n\u003cp\u003e70 \u0026nbsp; \u0026nbsp;Yan, K.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Evaluation of MODIS LAI/FPAR Product Collection 6. Part 2: Validation and Intercomparison. \u003cem\u003eRemote Sensing\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e (2016).\u003c/p\u003e\n\u003cp\u003e71 \u0026nbsp; \u0026nbsp;Chen, J. M., Feng, D. \u0026amp; Mingzhen, C. Locally adjusted cubic-spline capping for reconstructing seasonal trajectories of a satellite-derived surface parameter. \u003cem\u003eIEEE Transactions on Geoscience and Remote Sensing\u003c/em\u003e \u003cstrong\u003e44\u003c/strong\u003e, 2230-2238 (2006).\u003c/p\u003e\n\u003cp\u003e72 \u0026nbsp; \u0026nbsp;Wu, C.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Increased drought effects on the phenology of autumn leaf senescence. \u003cem\u003eNature Climate Change\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 943-949 (2022).\u003c/p\u003e\n\u003cp\u003e73 \u0026nbsp; \u0026nbsp;Giglio, L., Boschetti, L., Roy, D. P., Humber, M. L. \u0026amp; Justice, C. O. The Collection 6 MODIS burned area mapping algorithm and product. \u003cem\u003eRemote Sens Environ\u003c/em\u003e \u003cstrong\u003e217\u003c/strong\u003e, 72-85 (2018).\u003c/p\u003e\n\u003cp\u003e74 \u0026nbsp; \u0026nbsp;Lizundia-Loiola, J., Ot\u0026oacute;n, G., Ramo, R. \u0026amp; Chuvieco, E. A spatio-temporal active-fire clustering approach for global burned area mapping at 250 m from MODIS data. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e \u003cstrong\u003e236\u003c/strong\u003e (2020).\u003c/p\u003e\n\u003cp\u003e75 \u0026nbsp; \u0026nbsp;Freeborn, P. H., Wooster, M. J., Roy, D. P. \u0026amp; Cochrane, M. A. Quantification of MODIS fire radiative power (FRP) measurement uncertainty for use in satellite-based active fire characterization and biomass burning estimation. \u003cem\u003eGeophysical Research Letters\u003c/em\u003e \u003cstrong\u003e41\u003c/strong\u003e, 1988-1994 (2014).\u003c/p\u003e\n\u003cp\u003e76 \u0026nbsp; \u0026nbsp;Giglio, L., Schroeder, W. \u0026amp; Justice, C. O. The collection 6 MODIS active fire detection algorithm and fire products. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e \u003cstrong\u003e178\u003c/strong\u003e, 31-41 (2016).\u003c/p\u003e\n\u003cp\u003e77 \u0026nbsp; \u0026nbsp;Mu\u0026ntilde;oz-Sabater, J.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e ERA5-Land: a state-of-the-art global reanalysis dataset for land applications. \u003cem\u003eEarth System Science Data\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 4349-4383 (2021).\u003c/p\u003e\n\u003cp\u003e78 \u0026nbsp; \u0026nbsp;Yuan, W.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Increased atmospheric vapor pressure deficit reduces global vegetation growth. \u003cem\u003eScience Advances\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, eaax1396 (2019).\u003c/p\u003e\n\u003cp\u003e79 \u0026nbsp; \u0026nbsp;Abatzoglou, J. T., Dobrowski, S. Z., Parks, S. A. \u0026amp; Hegewisch, K. C. TerraClimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958-2015. \u003cem\u003eSci Data\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 170191 (2018).\u003c/p\u003e\n\u003cp\u003e80 \u0026nbsp; \u0026nbsp;Vitolo, C.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e ERA5-based global meteorological wildfire danger maps. \u003cem\u003eScientific Data\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e (2020).\u003c/p\u003e\n\u003cp\u003e81 \u0026nbsp; \u0026nbsp;Beck, H. E.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Present and future Koppen-Geiger climate classification maps at 1-km resolution. \u003cem\u003eSci Data\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 180214 (2018).\u003c/p\u003e\n\u003cp\u003e82 \u0026nbsp; \u0026nbsp;Sulla-Menashe, D., Gray, J. M., Abercrombie, S. P. \u0026amp; Friedl, M. A. Hierarchical mapping of annual global land cover 2001 to present: The MODIS Collection 6 Land Cover product. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e \u003cstrong\u003e222\u003c/strong\u003e, 183-194 (2019).\u003c/p\u003e\n\u003cp\u003e83 \u0026nbsp; \u0026nbsp;Olson, D. M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e 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. \u003cem\u003eBioScience\u003c/em\u003e \u003cstrong\u003e51\u003c/strong\u003e, 933-938 (2001).\u003c/p\u003e\n\u003cp\u003e84 \u0026nbsp; \u0026nbsp;Trabucco, A. \u0026amp; Zomer, R. Global Aridity Index and Potential Evapotranspiration (ET0) Climate Database v2. https://doi.org/10.6084/m9.figshare.7504448.v3 (2019).\u003c/p\u003e\n\u003cp\u003e85 \u0026nbsp; \u0026nbsp;DiMiceli, C. M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Annual global automated MODIS vegetation continuous field (MOD44B) at 250 m spatial resolution for data years beginning day 65, 2000-2014, collection 5 percent tree cover, version 6. https://doi.org/10.5067/MODIS/MOD44B.061 (2017).\u003c/p\u003e\n\u003cp\u003e86 \u0026nbsp; \u0026nbsp;Knorr, W., Kaminski, T., Arneth, A. \u0026amp; Weber, U. Impact of human population density on fire frequency at the global scale. \u003cem\u003eBiogeosciences\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 1085-1102 (2014).\u003c/p\u003e\n\u003cp\u003e87 \u0026nbsp; \u0026nbsp;Dormann, C. F.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. \u003cem\u003eEcography\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 27-46 (2013).\u003c/p\u003e\n\u003cp\u003e88 \u0026nbsp; \u0026nbsp;Forzieri, G.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Increased control of vegetation on global terrestrial energy fluxes. \u003cem\u003eNature Climate Change\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 356-362 (2020).\u003c/p\u003e\n\u003cp\u003e89 \u0026nbsp; \u0026nbsp;Guan, K.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Photosynthetic seasonality of global tropical forests constrained by hydroclimate. \u003cem\u003eNature Geoscience\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 284-289 (2015).\u003c/p\u003e\n\u003cp\u003e90 \u0026nbsp; \u0026nbsp;Hantson, S.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Model outputs: Quantitative assessment of fire and vegetation properties in historical simulations with fire-enabled vegetation models from the Fire Model Intercomparison Project [Data set]. \u003cem\u003eZenodo\u003c/em\u003e https://doi.org/10.5281/zenodo.3555562 (2019).\u003c/p\u003e\n\u003cp\u003e91 \u0026nbsp; \u0026nbsp;Pan, N., Wang, S., Wei, F., Shen, M. \u0026amp; Fu, B. Inconsistent changes in NPP and LAI determined from the parabolic LAI versus NPP relationship. \u003cem\u003eEcological Indicators\u003c/em\u003e \u003cstrong\u003e131\u003c/strong\u003e (2021).\u003c/p\u003e\n\u003cp\u003e92 \u0026nbsp; \u0026nbsp;Rabin, S. S.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e The Fire Modeling Intercomparison Project (FireMIP), phase 1: experimental and analytical protocols with detailed model descriptions. \u003cem\u003eGeoscientific Model Development\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 1175-1197 (2017).\u003c/p\u003e\n\u003cp\u003e93 \u0026nbsp; \u0026nbsp;He, Q.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Drought Risk of Global Terrestrial Gross Primary Productivity Over the Last 40 Years Detected by a Remote Sensing‐Driven Process Model. \u003cem\u003eJournal of Geophysical Research: Biogeosciences\u003c/em\u003e \u003cstrong\u003e126\u003c/strong\u003e (2021).\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5467904/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5467904/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eGlobal terrestrial ecosystems have witnessed increased vegetation greenness\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e and intensified fire regimes\u003csup\u003e\u003cspan additionalcitationids=\"CR5 CR6\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e in many ecosystems worldwide, but the potential connections between them remain elusive. We quantify the impact of vegetation greening on global fire activity by examining changes in live and dead fine fuels based on multiple long-term satellite-based datasets. We show that, despite the recently observed human-driven decline in global burned area\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e, vegetation greening has led to an increase in global burned fraction at a rate of 0.014\u0026thinsp;\u0026plusmn;\u0026thinsp;0.004% per year over 2001\u0026ndash;2020. This amplifying effect is primarily driven by the increase in dead fine fuel (0.047\u0026thinsp;\u0026plusmn;\u0026thinsp;0.009% per year), partially offset by the dampening effect of increased canopy live foliage (-0.018\u0026thinsp;\u0026plusmn;\u0026thinsp;0.007% per year). Notably, current fire-vegetation models inaccurately represent the interactions between fire and greening, resulting in underestimations of fire responses to vegetation greening, particularly in arid and cold regions. Our findings highlight the widespread amplification of global fire activity caused by the ongoing trend of vegetation greening. They underscore the importance of considering this biogeochemical positive feedback in the land-climate system and support the efforts to mitigate its impact on ecosystems and societies.\u003c/p\u003e","manuscriptTitle":"Vegetation greening enhances global fire activity","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-12-20 03:15:12","doi":"10.21203/rs.3.rs-5467904/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"86671821-3b80-4a68-834f-d763a952c2ed","owner":[],"postedDate":"December 20th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":41780556,"name":"Earth and environmental sciences/Climate sciences/Climate change/Climate and Earth system modelling"},{"id":41780557,"name":"Earth and environmental sciences/Climate sciences/Climate change/Climate-change impacts"},{"id":41780558,"name":"Biological sciences/Ecology/Fire ecology"}],"tags":[],"updatedAt":"2025-09-11T14:07:28+00:00","versionOfRecord":[],"versionCreatedAt":"2024-12-20 03:15:12","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5467904","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5467904","identity":"rs-5467904","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}