The United Nations System of Environmental-Economic Accounting Ecosystem Accounting (SEEA-EA) aims to quantify the contributions of ecosystems to economies and incorporate the contribution of nature into economic decision-making. However, a key challenge for the SEEA-EA ecosystem service (ES) accounts is distinguishing the actual use from potential supply, especially for regulating services like river flood mitigation service, and connecting it to different beneficiaries in a consistent unit. To address this challenge, integrated models method for river flood mitigation service accounts within the SEEA-EA framework was proposed to incorporating beneficiaries. The method comprises four steps: 1) determining serviceshed, service providing areas and service benefiting areas, 2) assessing potential ecosystem flood mitigation service, 3) simulating spatial flow, connecting it with beneficiaries, and valuation, and 4) compiling Supply and Use table of actual use in alignment with SEEA-EA framework. We applied this framework to the flood-prone Wanquanhe Basin on Hainan Island, China, to assess flood mitigation service provided by forest ecosystems (e.g., natural forests, rubber, and gardens) and the Niululing Reservoir. The SWAT and Random Forest models estimated potential ES, land use maps identified beneficiaries, and the HEC-RAS model simulated the spatial flow of floodwater. Data from these models were then integrated to quantify the ES actual use, which were subsequently incorporated into the SEEA-EA accounts. The case study revealed that in 2020, the upstream forest ecosystems and the reservoir provided 15.52 and 2.9 billion Yuan worth of actual flood mitigation service, respectively, to downstream beneficiaries. This study demonstrates how our integrated models method for river flood mitigation service effectively distinguishes between potential and actual ES use, connects services to beneficiaries, and constructs comprehensive SEEA-EA accounts. This method can be replicated in other regions to compile SEEA-EA accounts for river flood mitigation service, which can inform nature-based decision-making for flood risk management.

Ecological restoration programs have significantly enhanced carbon sequestration across key mountainous regions, yet the sustainability of these gains remains uncertain in hydro-climatic transition zones. This study develops an integrated assessment framework to quantify the concurrent evolution of net ecosystem production (NEP), soil moisture (SM), and vegetation resilience from 2001 to 2020. By synthesizing multi-source remote sensing data, field observations, and dynamic vegetation model outputs within a machine-learning architecture, we evaluated the eco-hydrological stability of restoration-impacted landscapes under ongoing hydro-climatic change and ecological transition. Methodologically, segmented regression was used to identify NEP trajectories and structural breakpoints, while vegetation resilience was quantified from STL-decomposed NDVI residuals using sliding autocorrelation and variance. Piecewise structural equation modeling was further employed to disentangle the climate-moisture-productivity cascades. NEP increased widely across the Qinling region, with a mean gain of 10.3 g C m- 2, despite concurrent regional drying, as precipitation and soil moisture declined relative to the 1990-2000 baseline at rates of -0.5064% yr- 1 and - 0.1840% yr- 1, respectively. Critically, random forest partial-dependence analysis identified a nonlinear moisture threshold: when SM depletion exceeds -0.02 m3 m- 3, particularly within the critical moisture-depletion interval of -0.04 to -0.06 m3 m- 3, the vegetation resilience indicator declines from >= 1.0 & times; 10-3 to <= 0.6 & times; 10-3. These findings indicate that current restoration strategies, while successful in achieving short-term carbon targets, may approach a moistureconstrained degradation threshold in specific sensitivity belts. This study provides an evidence-based reference for environmental impact assessments, suggesting a shift from productivity-centric goals toward climateadaptive and moisture-resilient management to ensure the long-term sustainability of ecological restoration.

In highly urbanized areas, rapid expansion of construction land often undermines ecological restoration benefits. Traditional ecological restoration frameworks struggle to meet the growing demand for ecosystem services (ES) during socio-economic development. Using Beijing as a case study, this research integrates socio-economic-natural complex ecosystem theory, spatial homogeneity theory, and restoration limitation hypothesis to propose a framework for simulating ecosystem service potential. This framework provides broader spatial options and quantitative reference targets for ecological restoration. The results showed that ecosystem service value (ESV) in Beijing has generally declined over the past 20 years, though it has increased annually by 1.46% in the last five years. Topographical factors significantly influence the spatial differentiation of ecosystem services across the city. When considering comprehensive ecological benefit restoration, the potential ESV for Beijing is estimated at 311821.57 million CNY, representing a 10.25% potential growth over current levels, with a recoverable area of 11446.20 km2. The value enhancement potential of individual ecosystem services ranges from 2.28% to 25.00%, with corresponding recoverable areas ranging from 981.69 km2 to 12693.01 km2. Forests and cropland play a significant role in restoring ecosystem service value potential. The mountain forests within the ecological conservation development zones and the wetlands across the city remain key areas for restoring ecosystem service value potential. This study enhances existing spatial identification approaches and provides quantitative reference targets for ecological restoration based on ecosystem service potential, supporting sustainable urban ecological planning.

Land-use change and associated landscape pattern modifications are key drivers of biodiversity dynamics. However, the influence of landscape heterogeneity on soil microbial diversity remains poorly understood. This study analyzed soil microbial communities (bacteria and fungi) and soil properties (organic carbon, pH, available potassium, nitrogen, and phosphorus) across 74 sites in Hainan, China, covering six land-use types (rubber, areca, banana, farmland, longan, and mango). We evaluated the effects of both landscape composition (habitat coverage and patch richness) and configuration heterogeneity (patch density and largest patch index) on soil microbial community structure and diversity. Results show that land-use types and landscape patterns weakly affect microbial composition but strongly shape diversity in a taxon-specific way. Bacterial Chao1 richness and fungal Shannon diversity varied significantly among land-use types, mainly due to soil available phosphorus. Rubber coverage positively correlated with bacterial and fungal diversity, while the largest patch index negatively affected only fungal diversity. Available phosphorus reduced this negative effect. This study demonstrates that land-use change and landscape heterogeneity differentially influence soil microbial composition and diversity, with available phosphorus emerging as a key determinant. Our findings highlight the need for careful landscape planning and soil nutrient management to preserve soil microbial diversity.

Accurate estimation of forest canopy height is critical for understanding forest dynamics and ecosystem functions in complex mountainous regions. Traditional optical remote sensing methods are widely used for estimating forest canopy height at large scales. However, challenges persist in vast mountainous areas, where rugged terrain and dense, high-biomass forest cover impede both human access and field survey. This study integrates Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) LiDAR, Sentinel-1 microwave, and Sentinel-2 optical data within a random forest machine learning framework, and a scale generalization methodology is applied to estimate canopy height in the alpine forests of Morihe, China. We conclude the following: (a) The ICESat-2 data were calibrated by unmanned-aerial-vehicle-based observations; the 95th percentile of relative height from ICESat-2 displays the strongest correlation (r = 0.80) among all measured parameters. (b) Incorporating multisource data, including microwave variables (VV and VH) and topographic variables (altitude and slope), improves model performance relative to that of the optical-only model, with R2 increasing from 0.77 to 0.83 and the root mean square error and mean absolute error decreasing by 0.21 and 0.10, respectively. (c) Compared with existing products, our product was better suited for canopy height estimation in complex mountainous areas, with R2 increasing from 0.34 to 0.83 and the root mean square error and mean absolute error decreasing by 0.63 and 0.52, respectively. Overall, multisource remote sensing data were used to produce wall-to-wall canopy height maps at a 10-m resolution, which reliably capture height distributions in complex mountainous areas, demonstrating the advantage of data fusion for enhancing mapping accuracy and ecosystem monitoring.

The persistent mismatch between the supply of ecosystem services (ES) and stakeholder demand remains a critical obstacle to effective ecosystem management. Previous studies have often failed to integrate the objective supply of ES with subjective stakeholder preferences within a unified spatial framework. In this study, we addressed this gap using the Shennongjia Forest Region (SNJFR) in China as a case study. We quantified the spatial supply of five key ESs (carbon sequestration, water yield, soil conservation, habitat quality, and leisure tourism) and evaluated the preferences for 23 ES types among four principal stakeholder groups (community residents, individual operators, government officials, and tourists; n = 120) using Q-methodology. A composite index based on multi-criteria decision analysis (MCDA) was then constructed to assess the spatial alignment between supply and demand. Our results revealed a distinct spatial gradient in the SNJFR, where the supply of the five key ESs was concentrated in the central region, diminishing toward the periphery. Conversely, stakeholder preferences clustered around four thematic perspectives: tourism culture, liveability, production-living, and ecological conservation. Notably, the tourism culture perspective received the strongest support (45.92%), indicating that cultural services are highly prioritized by stakeholders compared to regulating and supporting services. Crucially, the spatial matching analysis highlighted discrepancies between high-preference areas and actual service provision. Based on the MCDA, we delineated the region into three functional zones: comprehensive urban development, ecotourism-dominated, and ecological barrier protection. Our findings reveal divergent stakeholder feedback on zoning and development priorities. Decision-makers must reconcile these disparities during the zoning process to foster sustainable management, particularly in resource-rich and spatially heterogeneous regions.

Long-term intensive tillage has reduced soil organic carbon (SOC) and threatened sustainable maize production in Northeast China. Yet how tillage regulates SOC-yield relationship across climatic gradients remains unclear, limiting climate-smart cropland management. Here, we combined field observations with the DSSAT model to simulate long-term effects of conventional tillage (CT) and no-tillage (NT) on SOC and maize yield (MY) across the maize belt of Jilin Province, and qualified SOC-MY synergy/trade-off along aridity gradients. NT consistently increased SOC region-wide and improve MY on average, but yield benefits showed clear climatic thresholds. Yield gains were concentrated in moderately dry environments with intermediate water-heat availability (aridity index, AI ≈ 0.28-0.40). In humid zones (AI > 0.67), NT still promoted SOC accumulation but often failed to sustain MY, indicating a shift from co-benefit to trade-off. The area exhibiting simultaneous SOC and MY increases expanded rapidly during early adoption, stabilized in years 5-20, and gradually decline thereafter. These results demonstrate that NT co-benefit depend jointly on climate suitability and adoption stage, providing transferable thresholds for climate-informed promotion of conservation agriculture in temperate maize system.

The coupled effects of socio-economic and natural complex factors, both in time and space, collectively shape diverse ecosystem patterns, with different ecosystems exhibiting differentiated expansion probabilities in space. However, existing spatial selection frameworks for ecological restoration give limited consideration to this. Taking Beijing as an example, this study integrated ecosystem services value (ESV) assessment and land use expansion probability simulation to propose a framework for determining re-naturalization pathways aimed at enhancing natural benefits, intended to guide spatial selection in ecological restoration practices. The results showed that, over the past 20 years, the expansion of construction land has primarily encroached upon cropland. At the same time, historical ecological restoration projects have contributed to the recovery of forest, shrub, and wetland areas by 2.70%, 73.86%, and 9.9%, respectively. Terrain factors (such as DEM and slope) significantly influence the spatial expansion probabilities of various ecosystems. The overall ESV of the city has decreased by 2.84% over the past 20 years, with wetlands exhibiting the highest average ESV of 83.47 CNY/m2. Furthermore, the ESV supply capacity of shrubland, grassland, and wetlands has significantly increased, while the ESV supply capacity of forests and construction land has declined. Based on the principle of enhancing average ESV and expansion probabilities, this study identified 18 re-naturalization pathways with significant spatial differences and highlights the spatially extreme importance regions for each pathway. This framework provided spatial decision support for ecological restoration that combines the ecosystem services enhancement with socio-economic-natural coupling mechanisms.

Traditional ecological conditions indicators struggle to meet the demands of diverse ecosystem management. Ecosystem regulation services value (RSV) quantifies the monetized overall ecosystem benefits and conditions, having more intuitive and broader application prospects. However, the spatiotemporal fluctuations of actual meteorological conditions (AMC) significantly impact the objectivity and accuracy of RSV in characterizing ecosystem conditions. Sliding average meteorological conditions (SMC) often reduce the kurtosis of meteorological data, making them unsuitable for assessing ecosystem services to extreme meteorological conditions. This study used Beijing as a case study to propose the concept of comparable meteorological conditions (CMC) and determination method, then evaluated RSV from 2000 to 2020 under AMC, SMC, and CMC. Furthermore, this study compared RSV under different meteorological conditions with traditional ecological quantification indicator, Ecosystem Quality Index (EQI), and conducted spatiotemporal reliability verification using linear regression model (LR), geographically weighted regression model (GWR) and trend scoring methods. The results showed that the CMC for temperature, precipitation, evaporation, wind speed, and air humidity were 2004, 2020, 1994, 2009, and 2007, respectively. The spatial-temporal variation of CMC-based RSV exhibited smaller magnitude, better aligning with the stability of ecosystem. Spatial correlation between CMC-based RSV and EQI was significantly higher than that of AMC (LR: 1.51% to 5.44%; GWR: 0.90% to 1.39%) and SMC (LR: 1.64% to 5.40%; GWR: 0.97% to 1.34%). Temporal trend score between CMC-based RSV and EQI is 6.18% and 6.02% higher than under AMC and SMC, respectively. These results indicated that CMC provides a more scientific data basis for management-oriented RSV assessments.

Urban trees constitute a critical component of urban forest carbon sinks, and accurate quantification of their aboveground carbon stock (AGCS) is essential for advancing carbon-neutrality strategies and optimizing urban green-space planning. Although pixel-based remote sensing approaches remain the dominant paradigm for cityscale biomass estimation, their performance is often constrained in dense urban cores by tree-understory mixing, species-specific structural variability, and spectral interference from impervious surfaces. Focusing on Beijing's Second Ring Road, this study develops a crown-level, species-aware framework for individual-tree AGCS assessment by integrating the YOLOv7-seg deep learning model with species-specific crown-DBH allometric equations, explicitly mitigating mixed-pixel and background confounding effects. (1) The proposed crown-level model achieved strong detection performance in complex urban environments, identifying 183,144 individual trees, with an overall detection accuracy exceeding 0.69 and class-level AP@0.5 reaching 0.85 for Platycladus orientalis. (2) Species-aware AGCS estimation based on crown-DBH relationships (R-2 = 0.67-0.92) yielded a total urban tree carbon stock of 24.14 Gg C and a mean carbon density of 3.86 Mg C ha(-1) . (3) Compared with traditional pixel-based spectral machine learning approaches, the crown-level deep learning framework produced AGCS estimates more consistent with field measurements across diverse background conditions and species compositions, achieving higher explanatory power (R-2 = 0.909 vs. 0.633) and significantly reducing estimation bias (P < 0.05). By conducting carbon estimation at the individual-tree scale through precise crown delineation and species-specific allometric modeling, this study effectively minimizes interference from mixed pixels and non-tree vegetation, providing a more robust, interpretable, and transferable methodological foundation for high-resolution urban forest carbon monitoring and management.

Abstract Forest resilience is vital for ecosystem sustainability under environmental changes. While planted forests show lower resilience than natural forests, their response mechanisms remain unclear. We assess the resilience dynamics of natural and planted forests across China from 2001 to 2020 using lag-1-month autocorrelation. Natural forests showed a significant resilience increase (+0.0057 year-1), while planted forests declined (-0.0038 year-1). This contrast stems from distinct responses to key environmental drivers, including water availability factors (precipitation/PRE, soil moisture/SM), atmospheric stress factors (temperature/TEM, vapor pressure deficit/VPD) factors, and vegetation structural factors (vegetation continuous fields /VCF). In water availability-dominated regions, natural forest resilience adapted to lower water, while planted forests depended on higher water. This arose from the positive regulation of resilience by PRE and SM, forming a stable moisture buffer. In planted forests, the path effects from PRE to SM and from SM to resilience were relatively weak, and VPD had a stronger negative impact. In atmospheric stress-dominated regions, natural forest resilience was sensitive to TEM and VPD but maintained broad tolerance through multi-pathway regulation, with positive effects from TEM, SM, and VCF. Planted forest resilience lagged behind and lacked moisture buffering, resulting in a sharper decline in response to TEM and VPD changes. In vegetation structure-dominated regions, moderate VCF (40-70%) enhanced resilience, with natural forest resilience enhancing by VCF through direct effects and VPD buffering via SM. Planted forests relied more on climate-driven VPD mitigation, with weaker VCF effects and a stronger dependence on TEM and PRE for resilience. This study highlights climate-vegetation resilience pathways in natural and planted forests, emphasizing the role of moisture buffering and vegetation in enhancing resilience.

The Qinghai-Tibet Plateau (QTP), characterized by its unique geography and climate, plays an irreplaceable role in global biodiversity conservation and the provision of ecosystem services (ES). Protected areas (PAs) are essential for maintaining habitats and preventing ecosystem degradation; however, under climate change, the protection efficiency of PAs with fixed boundaries is increasingly constrained. Optimizing PA networks to enhance climate resilience while balancing biodiversity conservation and ES provision remains a major challenge. Here, we develop a multi-objective spatial optimization framework to improve PA effectiveness and support the joint conservation of biodiversity and ES under future climate scenarios. Our results show that: (1) current PAs cover only a limited proportion of biodiversity and s, and their future gains in protection efficiency remain modest under climate change; (2) prior to optimization, protection efficiency for biodiversity, water retention, and wind erosion control increases only marginally (approximately 0.8%), while carbon sequestration shows only a slight improvement; and (3) after optimization, biodiversity protection Climate-Resilient Conservation; Protected Area Optimization; Biodiversity; Ecosystem Services; Synergistic improvement; Qinghai-Tibet Plateau increases by 28.96% and +ecosystem service efficiency by 15.34%, while the protection efficiency of biodiversity–ES synergy areas increases by 42.7%. These findings demonstrate that synergistic optimization of biodiversity and ES is both feasible and effective for advancing climate-resilient conservation on the QTP. The proposed co-protection framework provides clear guidance for improving PA efficiency and spatial configuration, offering a practical approach for adaptive conservation in climate-sensitive regions, particularly in high-latitude and high-mountain ecosystems.

Atmospheric deposition of phosphorus (P) has multitudes of environmental implications, but quantifying and validating its spatial and temporal patterns remain highly challenging. Here, we integrated a detailed P emission inventory with atmospheric chemistry transport modeling and a temporally augmented mass-conserving downscaling approach to estimate terrestrial P deposition at 0.1° resolution during 2000-2019. The results reveal a 16.5% increase in global terrestrial P deposition, from 174.6 g ha-1 yr-1 (2.31 Tg yr-1) in 2000 to 202.6 g ha-1 yr-1 (2.69 Tg yr-1) in 2019. Among the natural sources, mineral dust predominated with emissions ranging from 1.31 to 1.50 Tg P yr-1, followed by primary biological aerosol particles (0.14-0.15 Tg P yr-1). Among anthropogenic sources, emissions were led by fossil fuel (1.12-1.62 Tg P yr-1) and biofuel combustion (0.39-0.52 Tg P yr-1), which significantly exceeded contributions from agricultural activities (0.07-0.10 Tg P yr-1). Spatially, the net global increase in P deposition was primarily driven by increased anthropogenic emissions in China (+4.09% per year) and India (+2.98% per year), which outpaced concurrent P emission reductions in the US (-0.40% per year) and Europe (-0.04% per year). Despite the implementation of a series of clean air policies in the two countries, anthropogenic P emissions continued to climb or level off, underscoring the need for systematic monitoring to mitigate potential environmental risks.

The global cryosphere is retreating under ongoing climate change. The Third Pole (TP) of the Earth, which serves as a critical water source for two billion people, is also experiencing this decline. However, the interplay between rising temperatures and increasing precipitation in the TP results in complex cryospheric responses, introducing uncertainties in the future budget of TP cryospheric water (including glacier and snow water equivalents and frozen soil moisture). Using a calibrated model that integrated multiple cryospheric-hydrological components and processes, we projected the TP cryospheric water budgets under both low and high climatic forcing scenarios for the period 2021–2100 and assessed the relative impact of temperature and precipitation. Results showed (1) that despite both scenarios involving simultaneous warming and wetting, under low climatic forcing, the total cryospheric budget exhibited positive dynamics (0.017 mm yr ^−1 with an average of 1.77 mm), primarily driven by increased precipitation. Glacier mass loss gradually declined with the rate of retreat slowing, accompanied by negligible declines in the budget of snow water equivalent and frozen soil moisture. (2) By contrast, high climatic forcing led to negative dynamics in the total cryospheric budget (−0.056 mm yr ^−1 with an average of −1.08 mm) dominated by warming, with accelerated decreases in the budget of all cryospheric components. These variations were most pronounced in higher-altitude regions, indicating elevation-dependent cryospheric budget dynamics. Overall, our findings present alternative futures for the TP cryosphere, and highlight novel evidence that optimistic cryospheric outcomes may be possible under specific climate scenarios.

Soil conservation service (SCS) plays a fundamental role in maintaining land productivity, preventing land degradation, and safeguarding long-term food and ecological security. Traditional assessments of SCS typically focus on individual pixels or plots, often overlooking the influence of landscape connectivity on SCS delivery. This study aims to quantify the hidden effects of landscape patterns governing SCS improvement during a large-scale ecological restoration program. Using multivariate statistical methods, we analyzed land use, landscape metrics, and observed sediment yield from 30 watersheds participating in China's Grain for Green Program (GGP) on the Loess Plateau between 2000 and 2020. Our results revealed four key findings: (1) The GGP drove significant land use transitions inducing a critical shift toward greater landscape aggregation; (2) 93.3 % of watersheds showed significant improvements in SCS closely synchronized with increased landscape aggregation; and (3) Variance partitioning analysis identified Landscape Pattern as the dominant unique contributor to SCS improvement (16.53 %), outperforming Ecological Engineering (12.17 %), Human Activities (7.36 %), and Climate (3.68 %); (4) SEM further confirmed that SCS enhancement was primarily driven by increased landscape aggregation (standardized path coefficient: 0.307, p < 0.01), followed by reduced fragmentation and improved connectivity. Notably, the GGP exerted its effects on SCS indirectly—by promoting landscape aggregation and vegetation dynamics—rather than through direct vegetation area expansion alone. These findings demonstrate that optimizing landscape aggregation is a critical, yet often overlooked, strategy for enhancing soil conservation, underscoring the need for restoration policies to move beyond area-based targets toward strategic, pattern-oriented landscape planning that maximizes conservation outcomes.

ETHNOPHARMACOLOGICAL RELEVANCE:Rubus suavissimus S.Lee (RS), a traditional ethnomedicine Guangxi, China, has long been used to manage diabetes and its complications. Existing studies have demonstrated the antidiabetic activity of RS and its complications, but the pharmacological material basis and molecular mechanism of its efficacy have not been clarified. AIM OF THE STUDY:This study aimed to elucidate the active constituents and specific molecular mechanisms underlying the therapeutic effects of RS in diabetic kidney disease (DKD). MATERIALS AND METHODS:Ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) was employed to identify major active components within RS polyphenol extracts. Core signaling pathways and key active components were screened using network pharmacology analysis and enrichment methods. Subsequently, an in vitro MPC-5 podocyte cell model was established under high-glucose and high-lipid conditions. Data-independent acquisition (DIA) mass spectrometry was used to analyze differentially expressed proteins in the cellular secretome. Mitochondrial ultrastructure was assessed using transmission electron microscopy (TEM). Key protein expression changes were validated by Western blotting. RESULTS:Network pharmacology screening identified myricetin (Myr) as the compound exhibiting the highest binding affinity to PIK3R1, suggesting its role as a key active component in the anti-DKD effects of RS polyphenols. In vitro, high-glucose and high-lipid exposed MPC-5 cells exhibited pronounced mitochondrial swelling and cristae disruption. Myr treatment significantly preserved mitochondrial morphology and induced the formation of double-membrane autophagic vesicles encapsulating damaged mitochondria, indicative of activated mitophagy. Proteomic analysis corroborated these findings. This study demonstrates for the first time that Myr, a principal active component of RS polyphenols, exerts its therapeutic potential in DKD by inhibiting PIK3R1. This inhibition promotes XBP1 expression, indirectly activating both the PI3K/Akt and PINK1/Parkin pathways, ultimately enhancing autophagic flux. CONCLUSIONS:Myr effectively activated autophagy and mitophagy by targeting the PI3K/Akt and PINK1/Parkin signaling pathways, facilitating the removal of dysfunctional mitochondria and mitigating cellular damage in DKD models. These findings provide a mechanistic foundation for the use of RS-derived polyphenols in chronic kidney disease management and highlight Myr's potential as a natural therapeutic agent for DKD.

Understanding factors that influence spontaneous plant species similarity across urban sites provides insights into species exchange processes in urban environments. We surveyed spontaneous plants in 30 urban sites in Nanjing, China, to investigate how seed capture opportunity, establishment resistance, and movement resistance affect species similarity between paired sites. Using various gravity models, we found that establishment resistance, measured by differences in land cover and building density between sites, explained 35.52 % of species similarity variation. Seed capture opportunity, calculated from paired sites' forest areas, explained 25.70 % of the variation. While Euclidean distance and land cover-based movement resistance showed no significant correlation with species similarity, building density-based movement resistance (measured within 50-meter buffers) explained 20.52 % of the variation. A combined model incorporating these factors achieved an R2 of 47.93 %. Analysis of dispersal modes revealed that wind-dispersed plants showed the highest inter-site similarity, followed by animal-dispersed plants, with unassisted dispersal plants showing the lowest similarity. Establishment resistance strongly influenced unassisted dispersal plants, while wind- and animal-dispersed plants responded to combinations of all three factors. These findings highlight how establishment resistance, forest coverage, and building density patterns shape spontaneous plant distribution across urban landscapes.

Natural ecosystems and water infrastructure (such as reservoirs) jointly exert an influence water flow by means of retaining, regulating, storing, and releasing water, thereby enhancing the availability of water resources to satisfy human demands. Previous research has predominantly concentrated on the role of natural ecosystems in water provision services; however, studies that integrate the contributions of both natural ecosystems and infrastructure to quantify their respective impacts on water provisioning services remain scarce. Here we utilize the SWAT hydrological model to simulate the spatiotemporal dynamics of water provisioning services in the Qinling-Danjiang watershed—an area prone to seasonal water shortages. The study delineates the supply and beneficiary areas, quantifying the relative contributions of natural ecosystems and infrastructure to watershed water provisioning services and their ecosystem service values were respectively evaluated by delineating the supply and beneficiary areas of water provisioning services and using scenario analyses. The annual water provision in the Danjiang watershed was 2.394 × 103 million m3, with significant variation across watershed and months. The total water demand from stakeholders was 1.122 × 103 million m3, with agricultural irrigation being the largest consumer, and 52.81% of the area experiencing a supply deficit. Under the baseline scenario, the value of water provisioning services was 14.602 billion CNY. In a scenario without reservoir infrastructure, water provision of natural ecosystems increased by approximately 27% to 3.039 × 103 million m3 (about 18.538 billion CNY), but exacerbated seasonal imbalances. Conversely, in a scenario without natural ecosystems, the water provision of reservoir infrastructure dropped by over 90% to 193 million m3 (about 1.179 billion CNY), which was insufficient to meet regional demands. This study provides a novel perspective for understanding the interactions between natural ecosystems and infrastructure in water provisioning services and offers a new approach to distinguish their relative contribution in water provisioning services, which is of great significance for accounting nature’s contribution to people.