Scale Effects of Water-Related Ecosystem Service Interactions and Their Driving Mechanisms in the Qinling–Daba Mountains, China: Implications for Ecological Management

Accurately understanding the scale dependence of complex interactions among ecosystem services and their driving mechanisms is of great significance for regional ecosystem management. However, existing research has insufficiently addressed the scale sensitivity of ecosystem services interactions and their underlying driving mechanisms, particularly in ecologically critical mountainous regions. This study focused on the Qinling–Daba Mountains in Shaanxi Province and examined three key water-related ecosystem services. The InVEST model was employed to assess water yield, water purification, and soil conservation services across three spatial scales—5 km × 5 km grid, sub-watershed, and county—over the period 2002–2022. Pearson correlation analysis and geographically weighted regression were applied to quantify trade-off and synergies relationships at each scale, and the Geodetector method was used to systematically analyze their multi-scale driving mechanisms. The main findings were as follows: (1) All three ecosystem services exhibited significant spatiotemporal heterogeneity across scales. Water yield showed an overall increasing trend, rising from 230.25 mm in 2002 to 333.64 mm in 2022; water purification declined persistently, decreasing from 0.38 kg/ha to 0.30 kg/ha; and soil conservation increased substantially, from 78.59 t/ha to 99.85 t/ha, with the spatial patterns of each service varying markedly across scales. (2) Ecosystem services interactions demonstrated strong temporal stability but significant spatial scale dependence. The synergies between water yield and water purification weakened monotonically with increasing scale, whereas both the intensity of synergies between water yield and soil conservation and the trade-offs intensity between water purification and soil conservation strengthened progressively with coarsening scale. (3) Both the number and the explanatory power of significant driving factors and their interactions increased systematically with coarsening scale. Land use factors (FLP, CLP) were dominant at grid scale, while the importance of climatic factors (PRE) and socioeconomic factors (GDP, POP) increased progressively from the sub-watershed to the county scale. Interaction types underwent three-stage scale-dependent transitions: two-factor enhancement and nonlinear enhancement coexisted at the grid scale; nonlinear enhancement became overwhelmingly dominant at the sub-watershed scale; and enhancement and weakening types coexisted at the county scale, producing a heterogeneous interaction structure. These findings reveal that the direction, intensity, and driving mechanisms of ecosystem services interactions are inherently scale-dependent. These conclusions methodologically demonstrate the necessity of multi-scale analytical frameworks and provide important scientific foundations for developing scale-matched, spatially differentiated ecological management strategies.