Ecosystem service potential, flow, demand and their spatial associations: a comparison of the nutrient retention service between a human- and a nature-dominated watershed

Nutrient regulation is an important ecosystem regulating service in watersheds. However, systematic investigations of the spatial associations between the potential, flow, and demand of the nutrient regulation service are still lacking. Therefore, we performed a case study comparing the total phosphorus (TP) retention in the Dianchi Lake (DL) watershed (human-dominated) with that in the Lower Reach of the Zi River (LRZR) watershed (nature-dominated). We used four indicators—TP retention potential, TP retention, TP load, and TP export—to represent the potential, flow, demand, and flow–demand budget of the TP retention service, respectively. We estimated the TP retention and export using the InVEST tool, mapped the four TP indicators and calculated their correlations, and estimated the contributions of different ecosystem types and terrain ranges to TP retention and export. We determined the following: (1) the incongruity between the spatial distribution of the TP retention potential and the other three TP indicators was smaller in the LRZR watershed than in the DL watershed; (2) the TP retention potentials generally increased—while the other three TP indicators decreased—with increases in the elevation gradient in the DL watershed and the slope gradients in both study areas; and (3) paddy fields exhibited the highest TP retention intensity and residential areas exhibited the highest TP export intensity among the major ecosystem types in both study areas. Moreover, the TP retention intensities of dryland crops and residential areas in the DL watershed were much higher than they were in the LRZR watershed. Our findings imply that the flow of the nutrient retention service is influenced more by the service demand than by the service potential and that it is influenced by both landscape composition and pattern. Because of the limitations and uncertainties in the modeling outputs, our results should be carefully used in other studies or in decision-making.