Global cold and wet ecosystems, such as permafrost, northern peatlands, Arctic tundra, boreal wetlands, and alpine swamp meadows, store large amounts of soil organic carbon (SOC) and are typically water-rich. While it is well recognized that these ecosystems are highly vulnerable to climate warming as it accelerates SOC decomposition, how soil water levels regulate SOC decomposition and CO2 emissions specifically by constraining oxygen (O2) availability during the growing season remains poorly understood at large spatial scales. Here, we integrate field observations, global data sets, and process-based models to quantify how soil water dynamics influence CO2 emissions by regulating O2 diffusion across these ecosystems. Results from 107 field sites consistently reveal a protective effect of high soil water levels against SOC decomposition and CO2 release during the peak growing season (representing one-third of the year), suggesting that, beyond warming, the loss of this protection due to soil moisture reduction under near-saturated conditions is a critical driver of increased CO2 emissions. However, global data-driven data sets and process-based model simulations show divergent correlations between soil CO2 emissions and soil water levels. Many models, which rely on simplified soil moisture scalars or empirical functions to represent soil water level effects, inadequately reproduce the effects of soil water levels on SOC decomposition and CO2 emissions in cold and wet ecosystems. Our study underscores the urgent need to incorporate more mechanistic representations of soil water-mediated SOC decomposition into models and to improve spatially explicit hydraulic parameters, thereby enhancing projections of soil carbon-climate feedback.

With global warming, rising atmospheric vapor pressure deficit (VPD) has emerged as a critical driver of vegetation productivity and the terrestrial carbon sink. Yet, the thresholds beyond which VPD constrains tree growth remain poorly quantified. Using 2,953 tree-ring sites and combining generalized additive models with threshold regression models, we demonstrate that growth-related VPD thresholds are widespread. Approximately 63% of the studied species groups exhibited identifiable thresholds, which were generally lower in cool, humid regions and higher in warm, dry environments. For the same tree species occurring under contrasting aridity conditions, thresholds also differed substantially. As VPD has risen, the proportion of sites exceeding these thresholds has increased rapidly since about 1970 and is projected to rise further under different emissions scenarios. Under SSP5-8.5 in particular, approximately 79.8% of the sites are projected to be exposed to VPD above their thresholds by the end of the century. Although potential upward threshold shifts associated with acclimation or adaptation could partly reduce future exceedance, this buffering effect becomes increasingly limited under stronger climate forcing. Without effective climate mitigation, rising VPD will increasingly exceed the tolerance thresholds of trees, placing more trees under sustained growth constraints and undermining terrestrial carbon sequestration.

The algal-bacterial symbiotic communities within the submerged macrophyte phyllosphere exhibit significant potential for lake restoration. However, their response mechanisms to environmental heterogeneity remain unclear, as traditional experiments or models typically overlook the complexity of cross-kingdom microbial networks. To address this, we established a cross-scale framework that integrates controlled mesocosm experiments with field lake surveys across trophic gradients. Using multilevel network analysis, we found that increased environmental heterogeneity promoted stochastic assembly and niche differentiation within phyllosphere communities. This enhanced the functional metabolic complementarity of algal-bacterial networks, thereby strengthening the ecosystem resilience. These findings challenge the traditional view that homogeneous environments favor microbial functional redundancy. Notably, machine learning models trained on experimental data showed high predictive accuracy but exhibited systematic biases when applied to natural lakes, highlighting the scale-dependent complexity of in situ microbial networks. Our study identifies heterogeneity-driven microbial insurance as a critical stabilizing mechanism and advocates for incorporating this ecological complexity into cross-scale restoration strategies.

Urban populations are increasingly exposed to severe and disproportionate heatwaves. While existing studies address urban heatwaves and intra-urban disparities, there remain gaps in understanding of the impact of urbanization on heatwaves and the associated inequalities. In this study, we analyzed urban compound heatwave (UCHW) and associated inequality across 936 global cities. Our findings reveal a sustained increase in UCHW under global urbanization, accompanied by a general decline in associated inequalities from 2003 to 2019. This trend is particularly pronounced in the Global South, where the intensification of UCHWs has outpaced that in the Global North, accompanied by a more significant reduction in related inequalities. Urbanization intensifies UCHWs by increasing impervious surfaces and reducing urban greenery, while concurrently decreasing their spatial heterogeneity and thus lowering UCHW inequality. Our study highlights the impact of the urbanization on UCHWs and associated inequalities, which is crucial for sustainability of cities.

Global changes are reshaping the structure and function of ecosystems, exerting profound impacts on the maintenance and enhancement of terrestrial ecosystem resilience. However, the future dynamics of ecosystem resilience in response to ongoing global warming remain unclear. Here, we investigate changes in terrestrial ecosystem resilience under global warming of 1.5 °C (GW1.5) and 2 °C (GW2) on the basis of leaf area index (LAI) data from CMIP6 simulations. Compared to the historical period (1985–2014), limiting global warming to 1.5 °C instead of 2 °C could reduce the proportion of regions experiencing declines in resilience by approximately 12% (SSP2-4.5) and 15% (SSP5-8.5). Compared with the SSP5-8.5 scenario, the resilience of boreal regions shows an increasing trend at GW1.5 under SSP2-4.5. An additional 0.5 °C of warming is projected to increase ecosystem resilience in tropical regions under SSP2-4.5, while exacerbating declines in the resilience of boreal regions under both scenarios. Long-term trend analysis indicates that 70% (SSP2-4.5) of regions and 79% (SSP5-8.5) will experience shifts from resilience gains to losses by the end of the 21st century, which are likely regulated by CO2 fertilization effects. Our findings highlight that even a half-degree difference in warming can markedly reshape terrestrial ecosystem resilience by mitigating overall declines and altering regional trajectories, underscoring the necessity of limiting global warming to 1.5 °C.

The Köppen–Geiger climate classification is widely used to represent global eco-climatic patterns, yet its application to high-altitude regions such as the Qinghai–Tibet Plateau (QTP) remains limited because tundra climates (ET) are defined primarily by temperature, overlooking moisture constraints on vegetation. Here we present a revised Köppen–Geiger framework that incorporates the aridity index (AI) as an intermediary to represent coupled hydrothermal conditions while preserving the original temperature–precipitation structure. This refinement subdivides the ET climate into three sub-classes: arid (ETw), semi-arid (ETm), and semi-humid (ETf) tundra. Validation against satellite-derived NDVI shows improved correspondence with vegetation patterns. The revised classification reveals systematic reorganization of climate zones across the QTP, highlighting the role of coupled thermal and moisture constraints in shaping eco-climatic gradients under recent climate change. More broadly, this approach provides a transferable pathway to refine climate classifications in high-altitude and cold–dry regions where temperature-based schemes alone are insufficient.

Ecosystem resilience is a key indicator of proximity to tipping points in terrestrial ecosystems, with declining resilience generally reflecting a heightened risk of abrupt transitions between ecosystem states. However, quantitative assessments of resilience change remain highly uncertain, owing to the inherently multidimensional nature of ecosystem resilience, as well as inconsistencies among assessment methodologies and ecosystem state variables used to infer resilience change. By synthesizing core ecosystem resilience concepts and evaluating multiple indicators and satellite-based vegetation state variables within established theoretical frameworks, we reveal substantial inconsistencies in inferred resilience trends: 59.6% and 42.8% of pixels show disagreement under the frameworks of critical slowing down and flickering, respectively, and 69.3% of show inconsistent trends across four satellite-based vegetation state variables (EVI, CSIF, LAI and kNDVI). Integrating dynamic global vegetation model simulations with observations further identifies climate change as the dominant driver of global ecosystem resilience changes, with temperature playing a key role. Together, these findings indicate that under ongoing global warming, neglecting conceptual differences in resilience and inconsistencies in representing critical transitions can lead to systematic misinterpretation of ecosystem stability and tipping-point risks, underscoring the urgent need for ecosystem-specific resilience assessment.

The Beijiang River Basin is an important ecological security protection area and water source supply area in Guangdong Province. This study assesses the spatiotemporal distribution characteristics of watershed water quality based on on-site monitoring data and multivariate statistical analysis. The results indicate that PO43-P concentrations peak during the flood season, whereas pH, NO3--N, and total nitrogen (TN) reach their highest levels during the autumn normal-flow period. Spatially, water quality follows a gradient of upstream > downstream > midstream, with the midstream region identified as the primary zone of water quality degradation. Future non-point source (NPS) pollution characteristics in the Beijiang River Basin are influenced by land use/cover change (LUCC) and climate change, showing significant variation across Shared Socioeconomic Pathway (SSP) scenarios. Under SSP126, precipitation increases at the slowest rate, with a peak annual value of 1599.77 mm during 2031-2040 and an average basin temperature of 19.61 degrees C. In contrast, SSP245 exhibits a marked increase in precipitation, reaching 1802.92 mm by 2061-2070. Under SSP585, annual precipitation rises to 2200.04 mm, with temperatures approximately 0.5 degrees C higher than those under SSP126. Simulations based on the improved ESP-PLUS model indicate that, under the natural development scenario (NDS), expansion of construction land increases urban runoff pollution by 32.97%. Under the economic development scenario (EDS), 1023 km(2) of ecological land is lost, significantly weakening pollution interception capacity, while construction land increases by 26.01%. In contrast, the coordinated development scenario (CDS) reduces ecological land loss by more than 60% compared to EDS through balanced development and conservation, thereby maintaining the basin's pollutant purification function. Overall, future nitrogen and phosphorus loads in the watershed are projected to first decrease and then increase. Accordingly, differentiated management strategies are recommended, emphasizing the coordinated development of economic growth and ecological protection, and providing a scientific basis for controlling NPS pollution under changing climatic conditions.

Previous studies have established how regional climate variability regulates local terrestrial gross primary productivity (GPP), yet the hemispheric-scale spatial organization of GPP, coordinated by large-scale atmospheric circulation, remains poorly understood. Here, using multi-source observations and numerical simulations, we show that anthropogenic shifts in Northern Hemisphere westerlies fundamentally reorganize terrestrial GPP patterns. Around 2000, westerly curvature reversed from a southward to a northward bend over eastern Europe, Northeast Asia, and western North America, while exhibiting opposite changes over central Asia and central North America. Spatial patterns of GPP trends during 1982-2018 closely match GPP responses to westerly curvature variations. Sensitivity analyses using CESM1 large-ensemble simulations and single-forcing experiments identify greenhouse gas forcing as the dominant driver of these changes, thereby reshaping GPP through surface climatic factors. Under the RCP8.5 scenario, continued curvature changes are projected to enhance GPP growth across northern Europe, Northeast Asia, and western North America, while suppressing productivity in southern Europe and central North America. These results reveal anthropogenic forcing influences terrestrial carbon uptake via large-scale atmospheric circulation, with important implications for predicting future carbon-climate feedback.

Fire regulates the carbon, water and nutrient cycles, especially in regions with distinct wet and dry seasons. The dramatic vegetation response to climatic factors complicates fire preconditions in Eurasian drylands, leaving our understanding of climate-fuel-fire patterns limited. Here we identify eight pyromes with unique combinations of fire characteristics of size, expansion, frequency and duration. We find these pyromes are geographically contiguous and exhibit varied response to climate, fuel load, fuel flammability and human activities. Fuel load is closely related to fast-spreading fires in Eurasian Steppe, which mostly belongs to the large-fast-frequent-long (LFFL) and medium-fast-intermediate-short (MFIS) pyromes. There, antecedent fine fuel accumulation during the growing season dries out in the subsequent dry season, is ignited by human activity and spreads rapidly. High-latitude regions are dominated by slow-spreading but long-lasting fires (large-slow-rare-long (LSRL) and medium-slow-intermediate-long (MSIL) pyromes), where changes in fuel flammability due to long-term climate stress determine fire extent. These findings provide initial guidance of which controls should be considered for improving predictions of fire activity and fire risk assessment across Eurasian drylands.

Urbanized karst watersheds are characterized by strong heterogeneity in groundwater recharge, aquifer storage change, and river–aquifer exchange. However, under the combined influence of karstification and urbanization, the temporal response relationships among recharge input, aquifer storage dynamics, and river–aquifer exchange remain insufficiently quantified. In this study, these coupled processes were investigated in the Nanming River Basin, a representative urbanized karst watershed in southwestern China. A coupled SWAT–MODFLOW model was developed to simulate groundwater recharge, aquifer storage change, and river–aquifer exchange. The model reproduced the main seasonal patterns of streamflow and groundwater levels, with streamflow R2 and NSE values of 0.55–0.85 and 0.51–0.83, respectively, and groundwater-level BIAS values of −0.020 m and −0.352 m at the two observation wells. The results show that groundwater recharge exhibited pronounced temporal pulsing and spatial heterogeneity. High recharge cells accounted for 25.03% of the active model area but contributed 79.38% of the total recharge. Recharge generated by SWAT and transferred to MODFLOW showed strong temporal consistency, indicating that the coupled framework effectively propagated recharge signals. In contrast, river–aquifer exchange varied more gradually and lagged groundwater recharge by 2 days in the coupled-model outputs. Across the 24 selected month-end output days, the Nanming River generally functioned as a gaining river, with positive net exchange occurring in 95.8% of the analyzed output periods and a cumulative net exchange of 3.61 × 106 m3. River–aquifer exchange also showed clear spatial concentration, with a few key river-connected subbasins contributing most of the cumulative net exchange. These findings indicate that groundwater recharge, aquifer storage change, and river–aquifer exchange do not respond as a simple synchronous system. The buffering role of aquifer storage and the detailed mechanisms underlying the asynchronous response require further verification.

Global warming has exhibited pronounced regional heterogeneity and temporal asymmetry, yet the underlying drivers and consequences of seasonal imbalance in temperature change remain insufficiently understood. In particular, asymmetric seasonal warming can fundamentally alter regional energy balance, hydrological pathways, and ecohydrological processes, potentially triggering cascading impacts across natural and human systems. Central Asia, as a typical arid region highly sensitive to climate change, provides an ideal case for investigating these processes. This study presents a comprehensive analysis of seasonal temperature variations over the past half century (1960-2020) and examines their ecohydrological implications. Our results reveal a marked shift in the dominant season contributing to long-term warming. While winter warming played a leading role during the earlier period (1960-1991), its contribution to annual mean temperature increase has substantially declined in recent decades. In contrast, spring warming has intensified significantly, with its contribution rising from 8.0% to 59.2% between the two periods, surpassing winter as the primary driver of regional warming. This transition reflects a fundamental reorganization of the seasonal thermal regime in Central Asia. The enhanced spring warming has induced profound changes in cryospheric and hydrological processes. Rising spring temperatures reduce snowfall fractions and accelerate snowmelt in mountainous regions, leading to earlier and more rapid release of water resources. Consequently, hydrological pathways have been altered, with a noticeable advance in the timing of spring runoff peaks and a decline in summer runoff. These changes disrupt the natural regulation of water availability, increasing the mismatch between water supply and demand during the growing season. In addition, intensified spring warming accelerates soil moisture depletion during early growing stages, significantly weakening the buffering capacity of soil water reservoirs. This process exacerbates water stress in subsequent months and contributes to the increasing frequency and severity of summer extreme events, including droughts, heatwaves, and hot-dry wind episodes. Such compound climate extremes pose substantial risks to agricultural productivity and ecosystem stability in this water-limited region. From an ecological perspective, spring warming also reshapes the coupling between thermal and hydrological conditions. It modifies the onset, duration, and intensity of the growing season, thereby influencing vegetation phenology and ecosystem productivity. However, these potential gains in early-season growth are often offset by intensified water limitations later in the season. The resulting imbalance amplifies ecosystem vulnerability and may trigger cascading ecological risks, including vegetation degradation and reduced resilience to climate extremes. Overall, this study provides a systematic assessment of the mechanisms through which enhanced spring warming influences the ecohydrological system in Central Asia. By integrating perspectives from energy balance, hydrological processes, and ecological responses, it highlights the critical role of seasonal warming asymmetry in driving regional environmental change. The findings offer new insights into the evolution of water resources and the emergence of ecological risks under ongoing climate change, and underscore the necessity of incorporating seasonal dynamics into climate impact assessments and adaptation strategies in arid regions.

The expansion of urban built-up areas and loss of urban greenspace alter the surface energy balance, exacerbating heat stress. While the greenspace loss reduces biophysical cooling, urbanization can indirectly promote vegetation growth, partially offsetting this effect. Here, using 1 km resolution air temperature data, satellite observations, and city-based random forest model, we quantified the cooling potential of indirect growth effects (IGEs) on compound heatwaves across 499 cities globally. On average, IGEs reduced cumulative heat by 0.35 degrees C, offsetting 4.6 % of urbanization-amplified compound heatwaves. Cooling effects were stronger in Global North cities (-0.48 degrees C) than in Global South cities (-0.25 degrees C). Scenario simulation show that a sustained increase in IGEs could double the cooling potential: if all cities attained the 90th percentile value, IGEs could mitigate up to 1.16 degrees C, offsetting 18.3 % of urbanization-amplified compound heatwaves. These findings highlight the significant potential for advancing heat risk management and promoting sustainable cities through Nature-based Solutions.

Rapid-onset flash droughts may cause devastating impacts on both ecosystems and human communities. We found that widely used multispectral satellite methods typically failed to capture the initial physiological response of forests to flash drought. Using a 3D radiative transfer model, we showed that canopy reflectance remained largely unchanged in the early stages of flash drought, even when upper-canopy leaves were wilting. We developed a new method based on a two-source energy balance model (TSEB-SM) driven by satellite infrared and soil moisture observations to capture water-flux dynamics in forests affected by flash drought. TSEB-SM tracked early-stage flash drought dynamics in China, where 8-daily transpiration declined by mean values of 14% (summer) and 6% (autumn) during the first two weeks of flash drought. Our results demonstrate that satellite-driven energy balance models can accurately track the temporal response of forests to flash drought, providing new tools for forest management in a warming climate. A new model that integrates satellite-based thermal and soil-moisture observations into a two-source energy balance framework accurately captures rapid forest responses to flash droughts, as validated by multi-source field data.

The Arctic has experienced rapid and profound changes due to its heightened sensitivity to global warming and growing regional human pressures. While past research has advanced our understanding of these transformations, a comprehensive assessment within a unified analytical framework is still needed to quantify the ecological impacts of human activity across this fragile region. In this study, we systematically assessed the expansion of human activity and its ecological effects across Arctic and sub-Arctic regions from 2000 to 2020. We combined satellite-based land-cover datasets, vegetation resilience indicator (i.e., lag-1 month temporal autocorrelation of remotely sensed greenness), and species distribution data to track and analyze these changes and impacts. Our findings show that areas affected by human activity—mainly cultivated lands and artificial surfaces—expanded by nearly 13,000 km², equivalent to a rate of 1.8 % per decade. This growth was largely driven by the increase in artificial surfaces (∼77.2 %) and extended to higher latitude. As a result, natural habitats became increasingly fragmented, vegetation resilience declined, and risks of ecological tipping points rose. These impacts threatened the habitats of approximately 97.5 % of Arctic species, including 111 species listed as vulnerable or endangered. Our results highlight that, beyond the effects of climate change, the continued expansion of human activity is intensifying ecological risks in the Arctic. This underscores an urgent need for enhanced ecological protection and transformative social strategies to safeguard the region's future.

The impact of elevated atmospheric vapor pressure deficit (VPD) on vegetation productivity is well-documented at monthly and annual scales. However, the influence of daytime VPD (VPDday) and nighttime VPD (VPDnight) is often overlooked. Using multiple long-term remote sensing proxies of vegetation productivity, we reveal distinct effects of VPDday and VPDnight on growing season vegetation productivity over the extratropical Northern Hemisphere (> 25° N). VPDday was negatively associated with vegetation productivity in 73.2% of vegetated pixels, and robustness analyses across alternative datasets and methods yielded a range of 67.4%-75.6%. By contrast, positive effects of VPDnight were detected in 51.8% of vegetated pixels, with corresponding estimates ranging from 36.5% to 55.7% across robustness analyses. This contrast was strongly related to aridity conditions. Vegetation productivity in drylands is more vulnerable to the double negative effects of high VPDday and VPDnight, while in humid regions, it benefits from increased VPDnight. Sap-flow observations helped explain this contrast from the perspective of plant hydraulic transport. In humid regions, relatively ample soil moisture allowed nocturnal water transport to be maintained under elevated VPDnight, helping restore plant water status overnight and providing favorable hydraulic conditions for daytime carbon uptake and vegetation productivity. In drylands, sap flow declined more strongly under high atmospheric demand and limited moisture during both daytime and nighttime, suggesting stronger hydraulic limitation and reduced overnight recovery, and thereby creating less favorable hydraulic conditions for vegetation productivity. These findings underscore the different roles of VPDday and VPDnight in regulating vegetation productivity and highlight the importance of incorporating both into models to improve predictions of climate change impacts on terrestrial ecosystems.

Tropical forests are increasingly affected by drought, yet the factors that control post-drought ecosystem resilience—the capacity to withstand disturbances—are not fully understood. Here we use temporal autocorrelation of satellite-derived vegetation greenness to quantify ecosystem resilience following 142,444 severe drought events across tropical forests from 2003 to 2022. We show that resilience declined in 68.8% of areas after droughts, particularly in dry environments, whereas 20.3% of areas with increased resilience were located in moist tropical forests. More intense and prolonged droughts led to a pronounced decline in resilience. Mean annual precipitation was identified as the most important regulator influencing resilience changes after drought, while soil phosphorus was the most consistent regulator across forest biomes, exhibiting widespread mitigating effects on resilience loss. Along decreasing precipitation gradients, the mitigating effect of soil phosphorus on post-drought resilience loss intensified. These findings provide insights into how tropical forests respond to drought and offer practical guidance for region-specific, adaptive forest management under a changing climate. Drought disturbances are reducing the recovery capacity of tropical forests, especially in drier conditions, but soil phosphorus can mitigate this impact, according to a satellite-based analysis of ecosystem resilience.