Pattern reconstruction and threshold mutual feedback of soil erosion sensitivity and risk in the black soil region of Northeast China: A 40-year analysis based on RUSLE and GeoDetector

Soil erosion poses a serious threat to global black soil security, yet long-term threshold feedback between erosion sensitivity and risk remains unclear. Clarifying spatiotemporal erosion patterns, sensitivity dynamics, and multi-factor drivers is critical for targeted prevention strategies. To address this gap, we quantified erosion intensity, sensitivity, and risk across the black soil region of Northeast China (BSNC) from 1980 to 2020 by integrating the Revised Universal Soil Loss Equation (RUSLE) with the Geographical Detector (GeoDetector). The results reveal that the high-risk centre of soil erosion has shifted from the Changbai Mountains to agricultural plains including the Song-Liao and flood plain. A positive feedback loop between erosion intensity and sensitivity keeps plain areas persistently locked at a mild-to-moderate intensity level. Critical soil erosion sensitivity thresholds show prominent hierarchical spatial divergence: the western steppe requires only 65% vegetation cover to curb erosion, whereas eastern agricultural plains require coverage above 82%. The steppe also exhibits an erosion-triggering rainfall erosivity threshold of only 1040 MJ·mm·hm −2 ·h −1 ·a −1 , far below mountain zones. The Song-Liao plain exhibits a bimodal soil erosion susceptibility (S kd ) distribution, with both S kd I and S kd V areas exceeding 30% of the total land. GeoDetector results indicate that the interaction between soil erodibility and land use reaches 0.4979, far higher than any single factor, highlighting that synergistic amplification of multiple factors drives erosion risk accumulation in the region. The nonlinear vegetation thresholds, positive feedback loops, and gentle-slope erosion mechanisms revealed here offer a transferable framework for other intensively farmed gentle-slope regions.