<p>Understanding how topography, vegetation, and rainfall affect sediment connectivity patterns is key for watershed management, but determining their dominant roles and interactions remains challenging. This study modified the connectivity index (IC) model by dynamically weighting vegetation and runoff variables to develop three indices: structural connectivity index (IC_S), functional connectivity index (IC_F), and composite connectivity index (IC_C). Using the Geographical Detector Model (GDM), we quantified the explanatory power of environmental factors over the spatial heterogeneity and interaction strengths of IC_C within the Zhongtianshe Watershed. Results indicate that IC_S and IC_C hotspots occur primarily on steep slopes, displaying gully or patchy spatial patterns. IC_C correlated strongly with watershed outlet sediment yield at monthly (<i>r</i> = 0.77) and seasonal scales (<i>r</i> = 0.74). Topography consistently served as the primary intrinsic driver. Vegetation (summer) and cumulative surface runoff depth (winter) significantly explained IC_C heterogeneity (<i>q</i> = 0.299 and 0.441, respectively). The spatial integration of relevant factors can enhance our understanding of watershed connectivity. The interaction effects among vegetation, rainfall, and topography exhibit stronger explanatory power for spatial heterogeneity than any single factor alone. While runoff depth and IC_C showed a positive correlation (<i>r</i> = 0.61) during the 2018–2019 seasonal transition, statistical significance was not achieved. The composite index effectively integrates structural and functional elements, providing an improved characterization of sediment yield processes and dominant drivers. Ultimately, this study provides a theoretical framework for understanding the spatial heterogeneity of soil erosion in small watersheds and offers scientific support for optimizing watershed management strategies.</p>