Zhonglin Xu

Spatial ecosystem-service assessment increasingly requires explicit links among service provision, demand, spatial delivery, and beneficiaries. This challenge is acute in arid basins, where mountain and headwater systems support spatially concentrated agricultural, urban, industrial, and port-related water use. Taking the Ebinur Lake Basin, China, as a case study, we develop a supply–flow–demand–compensation (S–F–D–C) framework integrating water-yield and demand assessment, spatial service-flow modelling, nonlinear driver analysis, and equity-oriented eco-compensation. From 2000 to 2023, water supply decreased from 2.494 × 10⁹ to 2.143 × 10⁹ m³, whereas demand increased from 1.288 × 10⁹ to 1.668 × 10⁹ m³, reinforcing a persistent mismatch between western source areas and central-eastern demand centres. Modelled inter-county flows revealed a stable ”two-provider and three-beneficiary” structure after 2005, with Wenquan and Bole as the principal providers and Jinghe, Alashankou, and Shuanghe as the main beneficiaries. BRT and MGWR identified nonlinear response intervals and spatially heterogeneous controls on the supply–demand balance. Flow-informed compensation and equity adjustment consistently ranked Wenquan highest in receiving priority and Alashankou highest in payment responsibility. By connecting biophysical supply and demand with spatial service dependence, beneficiaries, and monetary responsibility, the framework provides complementary information for ecosystem-service accounting, spatial planning, and cross-jurisdictional environmental governance.

Climate extremes can alter the balance between vegetation carbon uptake and ecosystem respiration, yet their nonlinear effects on vegetation net ecosystem productivity (NEP) across China remain insufficiently understood. Here, we integrated vegetation NEP and 15 extreme climate indices from 2001 to 2024 and combined trend analysis, partial-correlation networks, multi-model benchmarking and an XGBoost–SHAP–PDP framework to identify conditional associations, model contributions, interactions, nonlinear turning points and spatial heterogeneity. China experienced intensified warm extremes, weakened cold extremes and enhanced precipitation extremes. National mean NEP increased significantly (slope = 0.0020 kg C m⁻² yr⁻², p = 0.0232), and 70.04% of vegetated areas showed increasing trends, despite strong interannual variability, local declines and persistent negative tails. PRCPTOT, SDII, R99p and TN90p were identified as key NEP-related indicators, and XGBoost achieved the best performance among 13 candidate models. SHAP analysis showed that PRCPTOT contributed most strongly, followed by SDII, TN90p, TXx and WSDI, with strong interactions concentrated in precipitation amount–frequency–intensity and daytime–nighttime heat-extreme combinations. PDPs revealed decreasing responses for PRCPTOT and TN90p, increasing–plateau or peak-shaped responses for SDII and R99p, and segmented non-monotonic responses for TXx and WSDI. Local SHAP values further showed regional reversals in contribution direction. These findings demonstrate that climate-extreme effects on China’s vegetation NEP are threshold-dependent, interactive and spatially heterogeneous, and cannot be captured by a single linear relationship.

Flow-regime shifts driven by hydroclimatic variability and human water withdrawals are reshaping aquatic ecosystems in arid inland basins. For terminal lakes, a key uncertainty is whether restoration targets can be translated into attainable ecological inflow thresholds under future demand rigidity and hydrological extremes. We developed a Random Forest (RF)–adaptive NSGA-III (NSGA-III-Ada) framework for the Ebinur Lake Basin (ELB), Xinjiang, China, coupling lake-area–inflow response, regional-sectoral demand projection, multi-objective allocation and ecological-threshold assessment. Using 2013–2024 lake-area and inflow records, we derived staged ecological inflow targets and evaluated 2030 and 2035 allocation trade-offs under normal, dry and extremely dry scenarios (P=50%, P=75% and P=95%). The 522 km² core restoration target required Qtar₁ = 5.01 × 10⁸ m³, whereas the 800 km² ideal target required Qtar₂ = 7.19 × 10⁸ m³. Total demand was projected to decline from 18.36 × 10⁸ m³ in 2024 to 16.99 × 10⁸ m³ in 2030 and stabilize at 17.09 × 10⁸ m³ in 2035, with agricultural crops accounting for 82.0% of demand. Drier scenarios compressed the Pareto space, reducing economic benefits and increasing water shortages.Technique for Order Preference by Similarity to Ideal Solution (TOPSIS)-selected compromise solutions achieved Qtar₁ but not Qtar₂, leaving deficits of 0.61–1.16 × 10⁸ m³. These results identify Qtar₁ as an operational ecological threshold and Qtar₂ as a long-term restoration boundary for adaptive terminal-lake management under flow-regime shifts.

Dryland grasslands face elevated risks of rapid threshold crossing under a regime of warming, precipitation redistribution, and intensified interannual hydrothermal variability. Using the Ebinur Lake Basin (ELB) as a case, we developed an integrated structure × function assessment—linking land-use/cover change (LUCC) transitions with functional indicators of net primary productivity (NPP), net ecosystem production (NEP), soil conservation (SC), and grass supply (GS)—and coupled it with Bayesian-optimized XGBoost, SHAP, and partial dependence plots (PDPs) at a 30 m pixel scale to identify dominant drivers and ecological thresholds, subsequently translating them into governance zones. From 2003 to 2023, overall grassland status was dominated by degradation (20,160.62 km²; 69.42%), with restoration at 8878.85 km² (30.57%) and stability at 2.79 km² (0.01%). NPP/NEP followed a rise–decline–recovery trajectory, while SC exhibited marked bipolarity. Precipitation and temperature emerged as primary drivers (interaction X3 × X4 = 0.0621), whose effects, together with topography and accessibility, shaped a spatial paradigm of piedmont sensitive–oasis sluggish–lakeshore vulnerable. Key thresholds included an annual precipitation recovery threshold of ~200 mm and an optimal window of 272–429 mm; a road-density divide near ~0.06 km km⁻²; and sustainable grazing windows of ~2.2–4.2 and ~4.65–5.61 livestock units (LU) km⁻². These thresholds underpinned four management units—Priority Control (52.53%), Monitoring and Alert (21.53%), Natural Recovery (20.40%), and Optimized Maintenance (5.55%)—organized within a “two belts–four zones–one axis” spatial framework, closing the loop from threshold detection to adaptive governance. The approach provides a replicable paradigm for climate-adaptive management and ecological risk mitigation of dryland grasslands under warming.

Ecosystems in arid regions are facing increasing anthropogenic and environmental pressures, raising global concern about their sustainable development. Landscape ecological risk (LER) and ecological carrying capacity (ECC) are essential assessment frameworks for optimizing resource allocation and ensuring ecological security. In this study, the spatial and temporal evolution characteristics of LER and ECC in the Tianshan Mountains, a typical arid zone in Central Asia, was systematically assessed between 2000 and 2020. The PLUS model was employed to project ecological zoning patterns for 2030 and 2040, while XGBoost-SHAP was used to quantify the influence of 16 driving factors. LER was dominated by low to moderate-low risk zones (89%), while moderate-high to high risk zones rose from 0.43% to 2.82%. ECC showed a central-high and peripheral-low pattern, with zones of moderately high and high carrying capacity covering 48% of the area and increasing by 2.56%. Grasslands remained the dominant land type (48%) and expanded by 12,054.94 km ² . Cropland (39.99%) and construction land (105.85%) also increased. In contrast, water and unused land declined significantly. Ecological conservation zones remained dominant, the cultivation zone expanded sharply (+452.77%), and the restoration zone contracted (−70.84%). Projections suggest continued dominance of the conservation zone, while the restoration zone may experience strong fluctuations. Land use, temperature, precipitation, GDP, and population density were identified as key drivers of LER and ECC， with different effects in the four zones. These findings can be used to augment conservation management strategies in semi-arid grasslands.

To overcome the limited ability of the UNCCD “three‑indicator” set to detect ecosystem‑service change and management thresholds in arid grasslands, we develop an explicit–implicit–supply, five‑dimension framework and apply it to the Ebinur Lake Basin (2000–2020). Explicit shifts are mapped with a LULC transition atlas (code = A × 10 + B) and a cumulative trend index (Dex). Pixel‑wise linear slopes of NPP, NEP, SC and GS are tested with Mann–Kendall; an “OOAO + LfL” rule yields the LDN status. A Bayesian‑optimised XGBoost classifier, explained with SHAP, quantifies the marginal and interactive impacts of 15 natural and anthropogenic factors. Grassland change is dominated by degradation—69.42 % (20160.62 km ² ) versus 30.57 % recovery (8878.85 km ² ) and 0.01 % stability (2.79 km ² ); explicit net loss totals 1786.77 km ² and the centroid migrates ~11.1 km. Implicit functions improve overall: NPP rises across 77.58 % (22529.90 km ² ) and NEP across 76.35 % (22174.16 km ² ); SC is bipolar (45.40 % gain, 42.57 % loss); GS increases over 64.48 %. In improving pixels, 78.7 % show synchronous enhancement of all four implicit metrics, whereas 56.3 % of degraded pixels exhibit decline in only one metric; SHAP ranks human‑activity intensity (HAI = 0.706) above precipitation (0.438), elevation (0.297), temperature (0.243) and road density (0.217). Optimal windows emerge at 260–280 mm rainfall, 2000–2500 m elevation and HAI 2000–4000, while road density > 1 km km ^-2 exerts growing negative effects. The coupled control of human pressure and hydro- topography governs degradation–recovery patterns. The five‑dimension index set, combined with explainable machine learning, markedly enhances the accuracy and actionability of LDN assessments, informing avoid–reduce–reverse zoning in China’s Three‑North shelterbelt and grazing- to- grass initiatives.

High-altitude lakes are sensitive indicators of climate change, reflecting the hydrological impacts of global warming in alpine regions. This study investigates the long-term dynamics of the water level and surface area of Lake Karakul on the eastern Pamir Plateau from 1991 to 2020 using integrated satellite altimetry data from ERS-1/2, ICESat, and CryoSat-2. A multi-source fusion approach was applied to generate a continuous time series, overcoming the temporal limitations of individual missions. The results show a significant upward trend in both water level and area, with an average lake level rise of 8 cm per year and a surface area increase of approximately 13.2 km² per decade. The two variables exhibit a strong positive correlation (r = 0.84), and the Mann–Kendall test confirms the significance of the trends at the 95% confidence level. The satellite-derived water levels show high reliability, with an RMSE of 0.15 m when compared to reference data. These changes are primarily attributed to increased glacial meltwater inflow, driven by regional warming and accelerated glacier retreat, with glacier area shrinking by over 10% from 1978 to 2001 in the eastern Pamir. This study highlights the value of integrating multi-sensor satellite data for monitoring inland waters and provides critical insights into the climatic drivers of hydrological change in high-altitude endorheic basins.

Fly ash and bentonite, two abundant industrial by-products in China, offer promising potential for ecological restoration in arid regions. In this study, we conducted a factorial pot experiment with 24 treatment combinations (6 fly ash levels: F0–F5 = 0–100%; 4 bentonite levels: P0–P3 = 0–70 g kg^-1) to evaluate their synergistic effects on two xerophytic shrubs, Lycium ruthenicum and Ephedra intermedia. A total of 600 seeds per species were sown (5 seeds × 5 replicates × 24 treatments), and plant performance was assessed based on germination rate, survival rate, plant height, dry mass, and root morphological traits. Results showed species-specific and dosage-dependent responses. For L. ruthenicum, the highest germination rate (76%) was observed under 80% fly ash without bentonite (P0F4), but biomass and root development declined under higher fly ash or bentonite-only treatments (e.g., dry mass in P1F2: 10.40% ± 6.02%). In contrast, E. intermedia showed broader tolerance, with peak germination (88%) under moderate bentonite and 60%–80% fly ash (P1F3/P1F4), and optimal growth under P3F5 (root surface area: 10.18 ± 3.09 cm²; dry mass: 33.76% ± 26.53%). Two-way ANOVA revealed significant main effects and interactions between fly ash and bentonite on key traits (e.g., p < 0.001 for germination and survival in E. intermedia). Moderate fly ash addition (F2–F3, i.e., 40%–60%) with bentonite levels of 30–50 g kg^-1 effectively enhanced soil water retention and promoted early-stage growth, especially in E. intermedia. This study highlights the importance of species-specific amendment strategies and demonstrates the ecological benefits of valorizing industrial by-products for arid-land revegetation.

Populus euphratica Oliv, a representative species in arid ecosystems, serves vital ecological functions, including windbreak, sand stabilization and carbon sequestration. Investigating its nutrient resorption mechanisms is important for elucidating plant adaptation and growth strategies in nutrient-deficient environments. This study selected thirty sampling sites in Xinjiang across different locations and collected Populus euphratica leaves. Correlation analysis, regression analysis and redundancy analysis (RDA) were employed to assess the characteristics of nutrient resorption in Populus euphratica and their driving factors. We determined the leaf nitrogen (N) and phosphorus (P) concentrations in plants and then calculated the N:P ratio, nitrogen resorption efficiency (NRE) and phosphorus resorption efficiency (PRE). The results of the correlation analysis indicated that the NRE was not significantly correlated with the N and P contents or the N:P ratio in green leaves but was significantly related to those in senescent leaves. In contrast, the PRE was negatively correlated with the P content in the senescent leaves. The NRE:PRE ratio was positively correlated with the N in the green leaves and the P in the senescent leaves but negatively related to the N:P ratio in the senescent leaves. Nutrient resorption efficiency appeared to change with varying environments. Regression analysis revealed that both the NRE:PRE ratio and the NRE were positively correlated with latitude and mean annual precipitation (MAP) but negatively correlated with altitude and mean annual temperature (MAT). However, the PRE showed a negative correlation with latitude and MAP and positive correlations with altitude and MAT. Among these factors, the contribution of the MAP was the greatest, accounting for 85.1% and thus dominating the nutrient resorption processes in Populus euphratica. This study has enhanced the understanding of nutrient resorption conservation strategy and contributes to revealing the adaptation mechanisms of Populus euphratica under stress conditions such as drought and high temperatures.

As global forest areas decline and face increased risk from extreme events, optimizing forest types for long-term stability becomes crucial. However, empirical evidence for the effects of mixing methods on carbon and nitrogen dynamics in forest ecosystems remains limited. This study investigates five forest types in Southern China: the Tsuga longibracteata W.C.Cheng pure forests, the Tsuga longibracteata–hardwood mixed forests, the Tsuga longibracteata–Phyllostachys edulis (Carr.) J.Houz. mixed forests, the Tsuga longibracteata–Rhododendron simiarum Hance mixed forests, and the Tsuga longibracteata–hardwood–Rhododendron simiarum mixed forests (the tree species are all dominant community species). We examined one monoculture and four mixed forests, categorized into pure tree forests and tree–shrub mixed forests, and categorized by tree species richness levels of 1, 2, and 3. We measured carbon (C) and nitrogen (N) content, along with the C:N, of coarse woody debris (CWD) at various decay stages and in the adjacent topsoil (0–10 cm) to analyze decomposition rates and their effects on soil nutrients. Our results indicate that the C content and density of CWD differed significantly among forest types (p < 0.001). The Tsuga longibracteata–Phyllostachys edulis mixed forest exhibited the highest C and N content in CWD, but the lowest in adjacent topsoil, alongside the fastest decomposition rate. Soil C content and the C:N ratio showed highly significant differences among forest types (p < 0.001), and N content showed a significant difference (p < 0.05). Optimal outcomes occurred at a species richness level of 2, as excessive or insufficient species richness can diminish decomposition rates. The ecological benefits of tree–shrub mixed forests surpassed pure tree forests. Overall, these findings suggest that mixed forests do not always provide greater ecological advantages than pure forests, and that improper mixing can deplete soil.