Vegetation Restoration–Sediment Connectivity Coupling in an Ecologically Fragile Transitional Landscape: Insights for Watershed Management

Vegetation restoration serves as a crucial strategy for regulating sediment connectivity and enhancing carbon sequestration in ecologically fragile watersheds. However, accurately quantifying the spatial relationship between vegetation restoration and sediment connectivity presents a critical challenge for sustainable watershed management, particularly in assessing the efficacy of ecological engineering in targeting areas at high risk of erosion.

This study utilizes the Huangshui River Basin—a representative hydro-geomorphological transition zone between the Qinghai-Tibet Plateau and the Loess Plateau of China as a case study. It employs multi-source remote sensing data spanning from 2001 to 2023, the Index of Connectivity (IC), and interpretable machine learning techniques (XGBoost-SHAP) to explore the interconnections between carbon dynamics and hydro-geomorphological processes.

Over the past two decades, the basin has exhibited a significant "greening" trend, with Net Primary Productivity (NPP) increasing at an average rate of 2.83 gC·m^-2a^-1. Utilizing the Geographical Detector and XGBoost-SHAP model, we identified temperature as the primary non-linear driving factor (q=0.68). Land-use change decomposition reveals that growth improvements in stable vegetation contributed over 85% of the net NPP increment, while ecological engineering projects contributed a net increment of 0.14 TgC through the conversion of marginal croplands. Incorporating sediment connectivity index, the study further revealed a non-linear interaction between geomorphology and vegetation, wherein carbon sink gains across different vegetation types initially increased and subsequently decreased with escalating erosion risk (IC). Distinct restoration thresholds were identified for cropland (IC peak at -5.9), grassland (-6.1), and shrubland (-5.6). Forests demonstrate remarkable adaptability to environments characterized by high connectivity, sustaining elevated levels of productivity even in geomorphologically unstable regions. The spatial synergistic patterns further reveal that areas characterized by "high risk-high restoration" are predominantly concentrated in the fragmented gullies and steep slopes located in the central and eastern parts of the basin.

This research confirms that anthropogenic restoration measures have been effectively targeted the key source areas of the watershed. In contrast, areas characterized by "low risk-high restoration" areas are widely distributed across the western alpine meadows, reflecting climate-driven natural recovery processes. These findings provide crucial insights for spatially differentiated governance, suggesting that future strategies should prioritize maintaining the stability of high-connectivity forests while addressing the anthropogenic challenges faced by urban agglomerations in valley regions.