Amid accelerating global environmental change, assessing ecological vulnerability is critical for sustainability science. Focusing on the Yellow River Source Region (YRSR)—a key water source and ecological shield in China—this study develops an integrated assessment system based on the “Pressure–State–Response” (PSR) framework, incorporating 29 indicators. A combined weighting approach integrating analytic hierarchy process (AHP) with entropy-based objective weighting characterizes the spatiotemporal patterns, drivers, and future trajectories of ecological vulnerability. Key findings reveal: (1) heterogeneous warming–wetting trends with stronger humidification in the south and relative stability in the north drive divergent hydrological responses, highlighting the limitations of single-climate metrics in explaining vulnerability dynamics; (2) vulnerability patterns are primarily shaped by climatic factors—especially temperature and potential evapotranspiration—with anthropogenic pressures serving as secondary modulators, reinforcing the foundational role of thermal and moisture regimes in alpine ecosystem resilience; and (3) scenario projections consistently identify the northeast as a persistently high-vulnerability zone, yet show that balanced socioeconomic development can reconcile ecological protection with development needs. Based on these insights, a four-tier ecological zoning scheme and a governance framework comprising three strategies—strict conservation, adaptive regulation, and sustainable utilization—are proposed. This work offers actionable scientific guidance for tailored ecological conservation in the YRSR and contributes methodological advancements for vulnerability assessment and adaptive management of high-elevation ecosystems globally.

Haloxylon Bunge is a key restoration genus in drylands because it supports windbreak formation, sand fixation, and carbon sequestration. However, ongoing warming may reshape its habitat suitability across the global dryland belt. Here, we used Biomod2 to model global habitat suitability of Haloxylon under four CMIP6 climate scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5) across the 2030s, 2050s, 2070s, and 2090s. Under the historical baseline, suitable habitat covered 7.79 × 106 km2 and extended mainly from North Africa through the Middle East and Central Asia to north-western China. The dominant environmental controls were precipitation of the warmest quarter, UV-B seasonality, and temperature seasonality, contributing 33.82%, 14.14%, and 6.21%, respectively. Across all future scenario-period combinations, suitable area declined to 6.37-7.57 × 106 km2, equivalent to losses of 2.74%-18.22% relative to the baseline. This pattern indicates a persistent contraction of climatically suitable habitat, with larger declines generally occurring under higher-emission pathways. Low-suitability habitat changed little, whereas moderate- and high-suitability habitats declined consistently. Stable suitable and stable unsuitable areas dominated future change dynamics, while contractions were concentrated along the margins of the Sahara-Arabian region, the Iranian Plateau, and the Indo-Pakistan drylands. The suitability centroid remained within the Iranian Plateau, with a maximum net shift of 281.64 km, indicating regional redistribution within the existing dryland belt rather than large-scale geographic relocation. These findings show that future climate change will affect Haloxylon primarily through overall range contraction, erosion of higher-quality habitat and internal spatial reorganization within its current dryland distribution. This spatially explicit framework can help identify stable core areas, track vulnerable margins and cautiously assess frontier zones for future restoration and climate-adaptive planning.

Climate change is reshaping plant species distributions, posing challenges for drought-adapted taxa with restricted native ranges. This study employed an optimized MaxEnt ecological niche model, using GBIF occurrence records and a suite of bioclimatic and UV-B radiation variables, to project the current and future potential distribution of Caryopteris Bunge under four climate scenarios (SSP126, SSP245, SSP370, and SSP585). The model exhibited relatively high predictive performance, forecasting a consistent expansion of highly suitable habitats across all future pathways, with the SSP585 scenario projecting a 24.41

Groundwater resources, both the quantity and quality, are vital to ecosystems and livelihoods in arid and semiarid regions. Natural and anthropogenic activities greatly impact groundwater quality and stability, posing a potential threat to the ecological environment. Using hydrochemical diagramming and stable isotope tracing, the hydrochemical properties, recharge sources, transformation processes, and driving mechanisms of groundwater-surface water (rivers) were evaluated for the period between 2004 and 2024 in an oasis-desert system of northwest China. The groundwater and surface water (rivers) were slightly alkaline, with pH ranges of 6.90-8.40, and 6.94-7.84, respectively. Compared to 2004, groundwater electrical conductivity (4097.08 mu S center dot cm-1) and total dissolved solids (2622.13 mg center dot L-1) increased in 2024, with salinization intensifying along groundwater flow paths. The surface water was mainly HCO-3-Ca2+ type, while the groundwater was SO2-4-Na+, Cl-- Na+, and Cl-- SO2-4-Ca2+-Mg2+ types. Abrupt changes in groundwater quality in oasis-desert systems were identified in 2009. The groundwater hydrochemical components were predominantly determined by cation exchange, rock weathering, and evaporation-crystallization. Mean values of delta 18O (delta D) in shallow and deep groundwater were-12.12 %o (-82.86 %o) and-11.84 %o (-83.59 %o), respectively. A close hydraulic connection exists between shallow and deep groundwater, thus pushing the latter to transform into shallow groundwater. Finite-element simulations indicate that as pumping rates increase and freshwater recharge decreases, groundwater quality deteriorates. The salinization range has expanded across the oasis-desert system, with the maximum distance extension toward the freshwater side reaching 26.32-55.26 %. Thus, it is recommended to control pumping rates to monitor groundwater deterioration.

Glaciers of the Qinghai-Tibetan Plateau (QTP) are vital ecological and economic assets, playing a crucial role in regional ecosystems and socio-economic development. As global warming accelerates cryospheric retreat, it becomes increasingly important to accurately quantify changes in glacier ecosystem services and develop adaptive management strategies. This study focuses on glaciers in the QTP, utilizing data from the First (1978-2002) and Second (2004-2011) Chinese Glacier Inventories, multi-source remote sensing imagery, and socio-economic statistical yearbooks. Employing economic valuation methods such as shadow pricing and market valuation, this study comprehensively quantifies glacier service values in terms of freshwater supply, hydropower generation, climate regulation, runoff regulation, habitat support, aesthetic and recreational value, and scientific research. Additionally, glacier vulnerability across the different basins of the QTP is analyzed through dynamic indicators. The results indicate that the total glacier ecosystem service value on the QTP amounts to approximately 1.34 trillion CNY, with regulatory services accounting for 95.04 %. Compared to the First Glacier Inventory, service values decreased by 22.07 % in the Second Inventory. Service values are higher in the south and west, with the Brahmaputra and Tarim Basins contributing 27.45 % and 26.56 %, respectively. However, rising temperatures and reduced precipitation increase glacier vulnerability, particularly in high-value regions like the Indus Basin, where vulnerability is also influenced by population growth. The findings highlight the urgency of implementing adaptive management strategies, such as glacier disaster risk warning, glacier resource utilization, social participation, and industrial development, to mitigate ecological and economic risks and achieve sustainable regional development. This study provides a comprehensive assessment of glacier service values and vulnerabilities on the QTP, offering a holistic understanding of the impacts of glacier retreat in the region and informing more effective decision-making and management strategies.

Climate-groundwater interactions are the dominant drivers of plant-environment feedback processes in drylands worldwide. However, the responses of dryland ecosystems to acute atmospheric drought and chronic groundwater decline, as well as the underlying feedback loops that control state stability or transition of such ecosystems across different soil properties and plant salt tolerance remain uncertain. To address this knowledge gap, we introduces a comprehensive methodology that integrates a minimalist stochastic species-dependent soil water-salt dynamics model, a novel plant salinity-dependent water stress model, and a robust modelling framework for water and salt stress-vegetation feedbacks. Using lysimeter experiments and field transect studies, these models were showcased in the salt-tolerant Haloxylon ammodendron (H. ammodendron) ecosystem with fine-textured soils and salt-sensitive Haloxylon persicum (H. persicum) ecosystem with coarse-textured soils in the Gurbantunggut Desert, China. Our results indicate that over time water-salt imbalances are becoming more pronounced, driven by shifting precipitation regimes, declining groundwater tables, depleting soil moisture and intensifying salinization. The enhanced feedback between changing water and salt regimes and vegetation is forcing both Haloxylon ecosystems into a new stable state (i.e., bare or sparsely vegetated land), with shift in water-salt stress of the component species contributing to this transition. Furthermore, the tradeoff between resistance and resilience of such ecosystems is declining from the desert margins to the desert hinterlands. The H. ammodendron ecosystem shows lower resistance and resilience (indicating a decrease in stability), while the H. persicum ecosystem exhibits higher resistance and resilience. Although the water and salt stress-vegetation feedback in both Haloxylon ecosystems is driven by atmospheric and groundwater conditions and mediated by plant salt tolerance, their changing condition is ultimately determined by soil properties. These findings have major implications for the conservation and restoration of similar dryland ecosystems worldwide, especially in the context of a changing climate.

The scaling of renewable energy infrastructure and the conservation of biodiversity are crucial for achieving sustainable development and climate goals. However, the spatial extent of potential biodiversity exposure to photovoltaic solar (PV) infrastructure-defined here as the spatial overlap of species habitats with areas beyond core construction zones during both construction and operational phases-remains insufficiently quantified, especially under climate change. Here we address this gap by assessing this exposure in the arid region of Northwest China (ANWC), integrating three complementary models (system dynamics (SD), patch-generating land use simulation (PLUS), and BIOMOD2 ensemble) to project PV expansion and spatial overlap with biodiversity under current conditions and three future climate scenarios (Ecological Protection, Trend Continuation, and Economic Growth). Currently, PV coverage in this dryland region is 300 km2 and is projected to expand to 2600-5200 km2 by 2100, representing a 14-fold increase under the Economic Growth scenario. Spatial overlap between projected PV infrastructure and biodiversity hotspots remains limited under both present and future scenarios, with the proportion of modelled species distributions exposed to PV expansion being less than 1% at the regional scale. This low proportion reflects current and projected siting patterns (i.e., PV development predominantly avoids high-suitability habitats) rather than the magnitude of species-level responses. This study provides a spatially explicit quantification of potential exposure, not direct ecological impacts. Our findings highlight the need for proactive siting and mitigation strategies to ensure that renewable energy expansion proceeds in alignment with biodiversity conservation goals and the United Nations Sustainable Development Goals (SDGs 7, 13, and 15).

Near-surface microhabitats play a critical role in regeneration in high-altitude coniferous forests, yet quantitative evidence on how changes in these characteristics reshape controls on seedling survival remains scarce. Here, we examined Picea crassifolia forests in the Qilian Mountains on the northeastern Qinghai-Tibet Plateau using data from 124 seedling subplots within a 10.2 ha forest dynamics plot. Annual survival of established seedlings (>= 3 cm height) was monitored alongside measurements of moss layer structure, soil physicochemical properties, topography, and canopy structure. Partial correlation analysis, threshold regression, and Bayesian generalized linear mixed models were applied to assess nonlinear survival responses to moss thickness. Results revealed: (1) A significant threshold relationship between survival and moss thickness. Below 2.8 cm, survival increased with thickness; beyond this threshold, survival plateaued. (2) Environmental conditions differed markedly across the threshold: low-moss plots (<2.8 cm) exhibited lower moss biomass and cover, lower soil ammonium (NH4+), and higher bulk density and nitrate (NO3-) compared to high-moss plots (>= 2.8 cm). (3) Key environmental correlates of survival shifted across the threshold. Under low moss thickness, survival was primarily associated with soil organic carbon, NH4+, and slope; under high moss thickness, these associations weakened substantially. These results indicate that moss layer thickness acts as a key state variable regulating seedling survival in high-altitude coniferous forests, and that ecologically meaningful threshold responses can occur even when overall survival rates are high.

Soil inorganic carbon (SIC) accounts for more than 80 % of total soil carbon stocks in arid regions, yet its formation pathways remain poorly understood, particularly the role of soil organic carbon (SOC) mineralization in producing CO2 that precipitates as pedogenic inorganic carbon (PIC). Using a delta 13C end-member mixing approach in riparian soils of the lower Heihe River, Northwest China, the contribution of SOC-derived CO2 to PIC formation was quantitatively resolved across soil textures and depths. On average, 41 % of the CO2 contributing to PIC formation originated from SOC mineralization, with biogenic contributions declining with depth and remaining consistently higher in loam and loamy sand than in sandy soils. The estimated amount of SOC-derived CO2 associated with PIC formation (mCO2, defined here as the CO2 input required for carbonate precipitation rather than the stoichiometric CO2 content of PIC) exhibited strong positive relationships with SIC content and electrical conductivity. This pattern highlights a previously underrecognized coupling among soil ionic conditions, SOC availability, and carbonate precipitation processes. These findings provide direct isotopic evidence for the mechanistic role of SOC mineralization in facilitating PIC formation and underscore the importance of incorporating biogenic CO2 pathways into assessments of dryland carbon budgets and carbon-management strategies.

Accurately quantifying the sensitivity of alpine vegetation to climate change is a key prerequisite for formulating regional climate change adaptation policies. The sensitivity of the fragile alpine grasslands on the Tibetan Plateau to climate change has received widespread attention. However, the spatiotemporal dynamics and driving mechanisms of this sensitivity are still unclear under continuous warming and wetting. This study, based on MODIS_NDVI and meteorological data from 2000 to 2023, constructed a dynamic Vegetation Sensitivity Index (VSI) framework and integrated Random Forest (RF) and eXtreme Gradient Boosting (XGBoost) models with Shapley Additive exPlanations (SHAP) attribution analysis to reveal the spatiotemporal evolution characteristics and driving mechanisms of vegetation sensitivity on the Tibetan Plateau. The results show that (1) the VSI of alpine grasslands exhibited a spatial pattern of higher values in the southwest and lower values in the northeast, with an overall upward trend. Specifically, 56.31% of the region showed an increase in the VSI, with the upward trend being more pronounced in the northern plateau. (2) The dominant role of different climate factors varied regionally; vegetation sensitivity to precipitation increased in the northern plateau, and temperature sensitivity decreased in the central plateau, while sensitivity to solar radiation significantly increased in the central plateau. (3) SHAP attribution analysis indicated that elevation was the core factor driving VSI differentiation, showing a higher sensitivity at higher elevations, with lower growth rates. These findings reveal the dynamic evolution of vegetation sensitivity under the warming and wetting climate trend and its elevation-regulated mechanism, providing important scientific insights for regional ecological adaptation management.

Industry accounts for 70% of China's carbon emissions, making it the most critical sector for achieving national carbon neutrality. However, high-resolution understanding of the spatiotemporal dynamics of China's industrial carbon emissions remains limited. Here, we present a mapping framework based on land-use probability and machine learning to generate 30 m resolution industrial carbon emission maps for the period 2000-2020. Furthermore, we analyze the spatiotemporal evolution of industrial emissions and their relationship with urban scale. Results show that the proposed method achieves high modeling accuracy (R2 = 0.77) and strong spatial precision. We find that industrial emissions have become more spatially concentrated, with increasing regional polarization. We also identify significant regional heterogeneity, marked by a clear north-south divide: northern cities generally emit more industrial carbon than southern cities of similar population size, due to differences in industrial legacy and energy structure. This study provides a more accurate data foundation for understanding China's industrial carbon emissions, delivers essential scientific support for formulating region-specific emission reduction policies critical to achieving the dual carbon goals, and offers a transferable methodological framework for global carbon emission monitoring.

IntroductionIn extreme arid regions, Populus euphratica (P. euphratica) and Tamarix ramosissima (T. ramosissima) play vital ecological and landscape roles, but their survival and regeneration are severely limited by water scarcity. Since soil evaporation and plant transpiration represent the major pathways of water loss in these ecosystems, quantifying soil evaporation rates and clarifying plant water-use strategies are crucial for supporting vegetative growth and maintaining population stability.MethodsThis study investigated water movement in P. euphratica, T. ramosissima, and their mixed stand using multi-source isotope data collected from June to September.ResultsSoil evaporation exhibited consistent patterns across all stands, being most pronounced in the shallow layer (0–40 cm) and declining exponentially with depth, while the deep soil layer (>200 cm) was primarily recharged by groundwater. Although vegetation type introduced some variability in evaporation rates, these differences did not reach statistical significance. Both species primarily extracted water from the 100–300 cm soil layer and groundwater, with P. euphratica relying more heavily on soil water at 100–200 cm depth and T. ramosissima utilizing more groundwater, reflecting its greater drought tolerance. In mixed stands, this divergence in water use intensified interspecific competition, resulting in lower soil moisture and a more rapid decline in groundwater levels compared to pure stands, a phenomenon linked to differences in root distribution.DiscussionTo maintain the structure of desert riparian forests, ecological water conveyance must be carefully managed to sustain groundwater levels within a suitable range for vegetation. Such management would help prevent T. ramosissima dominance and facilitate the survival, regeneration, and sustainable development of P. euphratica.

Human-wildlife conflict in drylands is worsening due to climate change and increasing human pressures, yet its spatiotemporal dynamics remain poorly understood. Here, we present an integrated modeling framework combining system dynamics, the Patch-generating Land Use Simulation (PLUS) model, and the BIOMOD2 species distribution platform to project future potential conflict trends under three socioeconomic pathways (i.e., SSP1-2.6, SSP2-4.5, and SSP5-8.5). Our results show that 23,900 km2 of drylands are currently potentially affected by spatial overlap with conflict-prone areas, with projections under SSP5-8.5 suggesting that potential conflict areas could nearly double by 2100. Spatial simulations highlight oasis-desert transition zones, particularly near protected areas, as key hotspots where increasing aridity and habitat fragmentation drive heightened resource competition. Our analysis suggests that management aligned with Nature-based Solutions principles—including land-use restrictions, ecological restoration and protected area management—could reduce spatial overlap between human land use and wildlife habitat, supporting biodiversity and sustainable development. These findings underscore the need to integrate such strategies into global sustainability frameworks, like the Kunming-Montreal Global Biodiversity Framework, to enhance the resilience of dryland ecosystems amid rapid environmental change.

Spring precipitation directly impacts the growth and development of vegetation in dryland ecosystems, which in turn affects the biomass of the growing season and fundamentally maintains the ecosystem stability. However, the effects of recently frequent variabilities in spring precipitation on ecosystem productivity are still unclear. Here, we investigate the spatiotemporal patterns of precipitation and gross primary productivity (GPP) in the spring, as well as the impact of spring excessive precipitation on GPP over dryland ecosystems in northern China, based on remote sensing datasets of precipitation and GPP at eight-day intervals from 2000 to 2018. We also assess the sensitivity and stability of GPP to spring excessive precipitation. Spring excessive precipitation had a unidirectional driving effect on vegetation growth and an evident lag effect of an eight-day period across more than 87% of the drylands in northern China. This indicates that dryland ecosystems in northern China respond quickly to excessive precipitation in the spring. Vegetation sensitivity to excessive precipitation is the weakest at an aridity index of around 0.1. Drylands with high temperatures and large precipitation exhibit positive sensitivity, strong resistance, and high resilience of vegetation to excessive precipitation. There is a synchronous change between the resistance and resilience of GPP to excessive precipitation. Our results provide an innovative understanding of the interaction mechanism between precipitation variations and dryland ecosystems, offering theoretical support for ecological restoration, sustainable development, and carbon neutrality in dryland regions.

Natural drylands often exhibit multistability—distinct ecosystem configurations sustained under similar climates by internal feedbacks. Here, we test whether such multistability persists in human-managed drylands and evaluate how water management stabilizes ecosystem states. Focusing on China’s pan-Hexi Corridor—a representative oasis–desert mosaic spanning a long history of water reallocation—we integrate multi-source datasets to quantify an ecosystem-quality index (EQI). Combining natural water resources with composite human-activity intensity, we use potential-landscape and resilience diagnostics to identify multistability and apply regression models to quantify drivers and generate projections. Potential-landscape analysis reveals two natural low-EQI states coexisting, while a high-EQI state emerges where water and human-activity intensity are sufficient. Overall, human activity raises EQI but reduces resilience. Sectorally, forestry and environmental water are positively associated with EQI, whereas aquaculture and domestic uses are negative. Under CMIP6 scenarios, projections indicate regional EQI declines over the next 50 years, with 0.69–0.91

Spring precipitation directly impacts the growth and development of vegetation in dryland ecosystems, which in turn affects the biomass of the growing season and fundamentally maintains the ecosystem stability. However, the effects of recently frequent variabilities in spring precipitation on ecosystem productivity are still unclear. Here, we investigated the spatiotemporal patterns of precipitation and Gross Primary Productivity (GPP) in spring, as well as the impacts of spring excessive precipitation on GPP over dryland ecosystems in the northern China, based on remote sensing datasets of precipitation and GPP at 8-day intervals from 2000 to 2018. We also assessed the sensitivity and stability of GPP to spring excessive precipitation. Spring excessive precipitation had a unidirectional driving effect on vegetation growth and an evident lag effect of an 8-day period across more than 87% of the drylands in northern China. It indicates that dryland ecosystems in northern China responded quickly to excessive precipitation in spring. Vegetation sensitivity to excessive precipitation is the weakest at aridity index (AI) around 0.1. The drylands with high temperatures and large precipitation exhibit positive sensitivity, strong resistance, and high resilience of vegetation to excessive precipitation. There is a synchronous change between the resistance and resilience of GPP to excessive precipitation. Our results can provide innovative understanding of the interaction mechanism between precipitation variation and dryland ecosystems, offering theoretical support for ecological restoration, sustainable development, and carbon neutrality in dryland regions.

Introduction:Understanding responses of soil fungal community characteristics to vegetation restoration is essential for optimizing artificial restoration strategies in alpine mining ecosystems. Despite its ecological significance, current comprehension regarding the structure composition and assembly mechanisms of soil fungal communities following vegetation restoration in these fragile ecosystems remains insufficient. Methods:We used the high-throughput sequencing and null model analysis to determine the variations and environmental drivers of soil fungal community structures and assembly processes across different restoration chronosequences (natural plant sites, unrestored sites, 2-year restoration sites, and 6-year restoration sites) in a semiarid alpine coal mining region. Results:Artificial vegetation restoration significantly enhanced the α diversity of soil fungal communities while reducing β diversity. However, with prolonged restoration duration, we observed a significant decrease in α diversity accompanied by a corresponding increase in β diversity. Moreover, artificial restoration induced substantial modifications in soil fungal community composition. Taxonomic analysis demonstrated a distinct shift in dominant specialist species from Ascomycota in unrestored, natural plant, and 2-year restoration sites to Glomeromycota in 6-year restoration sites. Dispersal limitation and homogeneity selection were the predominant mechanism governing soil fungal community assembly, with its relative contributions varying significantly across restoration stages. In natural plant communities and unrestored sites, the structure of soil fungal community was primarily governed by dispersal limitation. The 2-year restoration sites exhibited a marked transition, with homogeneous selection emerging as the dominant assembly process, primarily influenced by soil sand content, total phosphorus (TP), total potassium (TK), and belowground biomass (BGB). This transition was accompanied by a significant reduction in the contribution of dispersal limitation. Discussion:As restoration progressed, the importance of homogeneous selection gradually decreased, while dispersal limitation regained prominence, with community structure being predominantly regulated by soil clay content, soil moisture content (SMC), and TP. Our results underscore the critical role of soil texture and phosphorus availability in shaping soil fungal community dynamics throughout the revegetation process.

Artemisia ordosica can prevent desertification and increase carbon sequestration, and it has been extensively planted in the Mu Us Desert, China. Evapotranspiration (ET) plays an important role in the survival of Artemisia ordosica. However, the controlling factors of ET remain unknown. To investigate the influencing factors on the actual evapotranspiration, we set up a weighing lysimeter with Artemisia ordosica in the Mu Us Desert. We collected data of air temperature (Ta), net radiation (Rn), wind speed (WS), soil moisture (θ), vapor pressure deficit (VPD), and heat flux (HF). The multiple linear regression model was used to quantify the influence of the six environmental factors on the ET. In addition, we applied the boosted regression tree (BRT) method to quantify the relative contribution of these environmental factors to ET. Our results show that annual ET was 444.46 mm, which was mainly influenced by the VPD during the dry season and Rn during the rainy season. This is different from the previous results which emphasized the importance of θ and Ta. The BRT results show that VPD and Rn are the most contributors to ET in the research area. In addition, ET significantly decreased when the soil moisture was less than 0.063 cm3/cm3. ET can increase by an average of 90% after a rainfall event. Our results have significance for the hydrological cycle and ecological environment protection.

Global climate change is driving shrub expansion in alpine grassland ecosystems. It is well known that shrubs play a crucial role in alleviating environmental stress and maintaining multiple ecosystem functions, especially under extreme conditions. However, the impacts of shrubs on soil multifunctionality (hereafter "EMF") at the alpha and R-scales, along with the patterns and underlying mechanisms across soil depth and precipitation gradients, remain unclear. Here, we investigated the effects of dominant shrubs on EMF (at alpha and R-scales) along a precipitation gradient (five sites ranging from 502 to 739 mm) and across three soil depths (0-15 cm, 15-30 cm, and 30-45 cm), and further identified the driving factors and regulatory mechanisms. We found that both alpha and R-EMF significantly decreased with increasing soil depth, and exhibited a hump-shaped distribution along the precipitation. The positive effect of shrubs on alpha-EMF was only evident under extreme precipitation conditions (either too low or too high), regardless of soil depth. Meanwhile, in surface soil (0-15 cm), shrubs significantly reduced R-EMF at the site with 544 mm of precipitation, whereas in deeper soil layers (30-45 cm), shrubs also led to a significant reduction in R-EMF at sites with 502 mm and 739 mm of precipitation. Shrubs, soil depth, and precipitation affected alpha-EMF primarily through abiotic pathways (soil moisture and pH), whereas their effects on R-EMF were mediated through both abiotic pathways (soil moisture and pH heterogeneity) and biotic pathways (bacterial R-diversity). Our study provides a comprehensive understanding of shrub impacts on ecosystem functions at different scales across various soil depths in alpine ecosystems, and emphasizes the regulatory role of precipitation, enhancing our understanding of the potential impacts of shrub expansion.