Short-Term Fluctuations of Ecosystem Services Beneath Long-Term Trends in the Pinglu Canal Basin in China

The Pinglu Canal, the first river-to-sea canal project since the founding of the People’s Republic of China, reshapes basin-scale ecosystem services by altering land use/land cover, landscape patterns, and soil–water processes. This study assesses spatiotemporal variations in net primary production (NPP), soil conservation, water yield, and nitrogen/phosphorus output in the Pinglu Canal Basin from 2000 to 2024. Hotspot analysis, the interannual fluctuation index, Random Forest and Shapley additive explanations, climate–NPP residual analysis, and Geodetector were integrated. The results indicate that: (1) Over the past 25 years, ecosystem services have generally improved, with NPP increasing significantly and nitrogen/phosphorus outputs declining; northern hilly and low-mountain forests formed stable service supply areas, whereas water yield was more prominent in southern plains and river valleys. Water yield showed the strongest interannual fluctuation, while soil conservation remained relatively stable. (2) Hotspot stability was mainly regulated by population density, normalized difference vegetation index (NDVI), elevation, and precipitation, with evident nonlinear threshold effects. When the NDVI reaches about 0.63, the NPP and nutrient retention capacity are significantly enhanced, indicating that vegetation coverage and community structure can exert their ecological regulation functions effectively after reaching a certain level. (3) During construction, negative NPP disturbances expanded along the canal. The Geodetector results showed that while the explanatory power of population density was enhanced during the construction stage of the Pinglu Canal, precipitation, NDVI, and topographic/hydrothermal conditions remain foundational constraints for ecosystem services spatial differentiation. Furthermore, interactions, such as population–NDVI and population–precipitation, generally manifest as bivariate enhancement. This study supports ecological monitoring, risk warning, and zoned restoration for large linear infrastructure projects.