Spatial Patterns and Influencing Factors of Forest Net Ecosystem Productivity in the Middle and Upper Reaches of the Ganjiang River Basin

Net ecosystem productivity (NEP) reflects the net carbon balance of forest ecosystems and is widely used to evaluate their carbon sink capacity. For the Ganjiang River Basin, identifying where forest NEP is high or low and explaining its controlling factors can support more targeted carbon sink management. However, under complex environmental conditions, the nonlinear responses of NEP and the differences among vegetation types are still not fully clear. In this study, forest NEP in the middle and upper reaches of the Ganjiang River Basin was estimated for 2023. An XGBoost–SHAP framework was then used to examine the effects of climatic, topographic, and stand structural factors and to identify possible threshold responses. The results showed that forest NEP had clear spatial differences. High NEP values were mainly distributed in peripheral areas, whereas low values were concentrated in the central region. The spatial distribution also showed significant positive autocorrelation. At the regional scale, elevation (DEM), mean annual temperature (TEMP), and vapor pressure deficit (VPD) were the dominant factors affecting NEP. However, the main drivers varied among different vegetation types. The SHAP results further indicated that several factors had nonlinear threshold effects. Precipitation showed an inhibitory effect within 1400–1680 mm, and VPD showed a similar negative response within 0.48–0.54 kPa. These results help explain the formation of regional forest carbon sinks and provide a reference for forest-type-specific ecological management.