Nonlinear dynamics and driving mechanisms of precipitation and soil water use efficiency in the Three-North Region of China

The Three-North Shelterbelt Program region is an important ecological security barrier in the arid and semi-arid ecologically fragile zone of northern China, where the carbon-water coupling mechanisms under water limitation urgently need to be quantitatively identified. Using the Three-North Shelterbelt Program region as the study area, this study constructed precipitation use efficiency (PUE) and soil water use efficiency (SWUE) based on remote-sensing productivity products and reanalysis hydrometeorological data. We further compared two CEDAR-GPP scenarios with different representations of CO 2 -related processes to characterize the spatiotemporal variations and driving mechanisms of PUE and SWUE during 1982–2018. CatBoost-SHAP was further employed to reveal the nonlinear effects of key driving factors and the modulation effects of aridity gradients, and a partial least squares structural equation model was used to construct a cascading pathway framework linking geography, climate, vegetation, and human activities, so as to comprehensively analyze the underlying mechanisms. The results showed that PUE and SWUE under the scenarios with and without the CO 2 fertilization effect generally exhibited increasing trends at both monthly and annual scales, accompanied by significant interannual fluctuations. Spatially, the trend patterns under the two scenarios were generally consistent, with differences mainly reflected in the intensity of local changes. PUE was mainly controlled by evapotranspiration and showed significant threshold and saturation characteristics in the low evapotranspiration range; SWUE was mainly controlled by leaf area index and jointly modulated by factors such as radiation, air temperature, and latitude, while exhibiting diminishing marginal effects at high leaf area index levels, and aridity significantly regulated the strength and direction of the contributions of key driving factors. Topography constituted the main indirect pathway by influencing climate and regulating vegetation, while human activities exerted direct effects on both indicators as well as vegetation-mediated indirect effects, leading to differences in the composition of the pathways of PUE and SWUE. These results provide a new analytical framework for ecological restoration and carbon–water coupling mechanisms in the Three-North Shelterbelt Program region.