水位波动与土地利用变化对香溪河流域及其消落带湿地景观生态系统服务的影响OACHSSCD

[Objective]Large reservoirs reshape river-reservoir-riparian systems through regulated water-level fluctuations,while land-use change modifies vegetation cover,soil properties,pollutant transport,carbon pools,and habitat condition.Although both processes can affect ecosystem services,their relative explanatory roles have rarely been compared within a unified analytical framework,particularly between an entire watershed and its directly inundated drawdown zone.Because the drawdown zone is alternately inundated and exposed under reverse-seasonal regulation,it may respond more strongly than the basin-wide land-use pattern.Accordingly,this study asks which factor group provides greater explanatory power for each ecological indicator and whether that relationship changes with spatial scale.Focusing on the Xiangxi River Basin,a major tributary close to the Three Gorges Dam,this study examines how water-level fluctuations and land-use change are associated with changes in multiple ecosystem services.It further evaluates whether the 145−175 m drawdown zone functions as a sensitive wetland landscape unit and translates the findings into implications for ecological restoration and sustainable landscape design. [Methods]Daily water-level,inflow,and outflow records for the Three Gorges Reservoir from 2003 to 2020 were used to characterize mean water level,water-level range,cumulative fluctuation intensity,duration within the 145−175 m elevation range,days above selected elevation thresholds,and inflow−outflow imbalance.A 30 m digital elevation model was used to delineate the drawdown zone and its elevation subzones.The zone was divided into[145,155),[155,165),[165,170),and[170,175]m bands to represent elevation-dependent inundation gradients.Land-use maps for 2000,2005,2010,2015,and 2020 were analyzed at the whole-basin,non-drawdown,and drawdown-zone scales.Five ecosystem services were represented by six indicators derived from the integrated valuation of ecosystem services and tradeoffs(InVEST)modules and associated spatial analyses.Annual water yield represented potential water provision;soil retention represented erosion regulation;nitrogen and phosphorus exports were treated as negative pressure indicators and reverse-standardized to represent the benefit direction of nutrient retention and water purification;carbon storage represented climate regulation;and habitat quality was evaluated separately as a proxy for biodiversity condition.Zonal statistics and land-use transition analysis were combined with Spearman's rank correlation and grey relational analysis.Random forest regression and SHapley Additive exPlanations(SHAP)were then used to compare the relative explanatory contributions of whole-basin land-use change,drawdown-zone land-use change,and water-level fluctuation indicators.Because only four land-use transition intervals were available,the statistical and machine-learning results were interpreted as exploratory associations and model-based explanatory contributions rather than causal effects. [Results]The 145−175 m drawdown zone covered 13.36 km2,representing 0.42％of the valid basin area.The[145,155)and[170,175]m bands were the largest,covering 4.37 and 4.25 km2,respectively,indicating an uneven elevation structure.The hydrological regime changed markedly after 2005.The available records for 2003−2005 contained no days with reservoir water levels within the 145−175 m range,whereas the corresponding durations reached 1,428 days in 2006−2010,1,821 days in 2011−2015,and 1,827 days in 2016−2020.Cumulative water-level fluctuation intensity increased from 169.95 to 576.34 m,indicating the establishment of a recurring annual impoundment-drawdown regime.Land-use change was more concentrated in the drawdown zone than across the basin.Between 2000 and 2020,open-water area within the zone increased from 0.71 to 5.27 km2,while cropland,forest land,and built-up land decreased by 1.92,1.50,and 1.13 km2,respectively.Ecosystem responses differed among indicators.Mean annual water yield in the drawdown zone decreased from 598.61 to 556.58 mm·a−1 but recovered after 2015.Soil retention declined before 2015 and recovered to 234.56 t·hm−1·a−1 in 2020.Nitrogen and phosphorus export pressures decreased by 17.05％and 15.34％,respectively,although both remained higher than in non-drawdown areas.Carbon storage declined by 35.44％,whereas habitat quality increased slightly.The latter should not be interpreted as evidence of overall biodiversity improvement because stable terrestrial riparian habitats were simultaneously lost.Random forest and SHAP results showed that water-level fluctuation indicators accounted for 97.01％and 96.18％of the model-based explanation for water yield and soil retention at the whole-basin scale and 80.90％and 86.51％within the drawdown zone.Land-use change provided the larger explanatory contribution for nitrogen and phosphorus export pressures,carbon storage,and habitat quality,ranging from 62.63％to 71.30％at the whole-basin scale and from 42.33％to 54.85％within the drawdown zone. [Conclusion]Water-level fluctuations and land-use change were associated with different dimensions of ecosystem response.Water-level indicators provided the greatest explanatory contributions for potential water provision and erosion regulation,whereas land-use change was more influential in explaining nutrient-export pressure,carbon storage,and habitat condition.This differentiated pattern cautions against interpreting all indicator changes as a uniform improvement or decline.The drawdown zone therefore represents a key interface through which reservoir operation and land-cover transformation jointly affect the wetland landscape.Sustainable management should treat it as a distinct planning unit and coordinate elevation-specific vegetation restoration,riparian buffer construction,control of built-up land,non-point-source pollution reduction,and long-term monitoring of water-level processes.This dual-scale framework provides a reproducible basis for adaptive restoration and sustainable landscape design in tributary wetlands affected by large reservoirs.