<p>Rocky desertification governance significantly influences vegetation phenology in karst regions by improving the eco-environment and enhancing ecosystem functions. However, the spatiotemporal dynamics and underlying driving mechanisms of these governance-induced phenological shifts have not been systematically revealed due to the absence of a comprehensive analytical framework. To address the insufficient attention to governance–phenology linkages, this study developed a Governance-Phenological Induction Analysis Framework (GPIAF) and applied it to the Yunnan–Guizhou–Guangxi rocky desertification region from 2003 to 2020. Based on long-term MODIS time-series data, we extracted the rocky desertification indicator (RDI) and three key phenological parameters (SOS, EOS, LOS). Trend analysis and coupling coordination degree models were employed to characterize changing patterns and quantify governance-phenology relationships. Furthermore, gradient boosting machine algorithms combined with SHapley Additive exPlanations were used to identify dominant drivers. The results demonstrated that in effective phenology zones of rocky desertification, RDI decreased significantly in 56.28% of the regions, SOS advanced in 33.46%, EOS was delayed in 26.50%, and LOS extended in 23.41%. A high inductive effect (averaging 93.6%) was exhibited between rocky desertification governance and LOS, and this effect strengthened over time. Environmental factors and human activity indicators made comparable contributions to the inductive mechanism, with precipitation (negative) and nighttime light (positive) being the dominant drivers. The multidimensional analysis of GPIAF indicates that, in addition to climatic factors, rocky desertification governance also exerts significant influences on vegetation phenology, providing a scientific basis for optimizing governance strategies.</p>