Soil Hydrological Functions and Threshold Effects Under Different Modes of Vegetation Restoration in a Reclaimed Coal Mine of the Loess Plateau

Traditional evaluations of ecological restoration in mining lands have long been dominated by above ground vegetation metrics, such as coverage and community diversity. Yet in arid and semi-arid mining regions, soil hydrology recovery is key, as soil organic carbon (SOC) and bulk density (BD) are critical factors affecting soil hydrology. However, their thresholds for water holding capacity and infiltration remain unclear. Therefore, this study determined the SOC and BD thresholds and evaluated the hydrological trends across them. The study was in a Loess Plateau coal reclamation area. Five restoration types and one reference forest were selected, and soil samples were collected from the 0–10 cm layer. Redundancy analysis (RDA), partial least squares structural equation modeling (PLS–SEM), and the threshold regression model were used for analysis. Results show the following: (1) Vegetation indirectly controls soil hydrology via soil structure and chemical properties (RDA: 76.04%). BD limited water holding capacity (path coefficient: −0.908), while chemical properties promoted infiltration (path coefficient: 0.397). (2) Soil hydrological recovery exhibits non-linear threshold responses. The SOC thresholds for water holding capacity and infiltration were 9.52 g/kg and 5.93 g/kg, respectively, while the BD thresholds were 0.93 g/cm3 and 1.03 g/cm3. The reclamation of soil hydrological functions in mining lands follows a two-stage mechanism. First, control by carbon accumulation, then follow by structural dominance. During the early stage, soil organic carbon (SOC) buildup quickly boosts hydrological performance by promoting aggregate formation. However, once the system nears its functional threshold, simply adding more carbon gives limited returns. At this stage, breaking through physical constraints becomes vital to overcoming the bottleneck. Above-ground metrics alone cannot capture below-ground ecosystem functional trajectories. Consequently, a dynamic strategy centered on SOC and BD is proposed.