The interplay between ecosystem productivity and climate water deficit (CWD) is a key indicator of ecological sensitivity to climate change. This study uses Bayesian linear regression to assess the impact of CWD variations on vegetation productivity, measured by the Normalized Difference Vegetation Index (NDVI), in southwestern China. The results show that 81.58% of the region exhibits negative sensitivity, where NDVI decreases with increasing CWD, particularly in karst peak-cluster depression areas. Conversely, mid-to-high mountain regions in the northwest show positive sensitivity, with vegetation productivity increasing under drought conditions. Lithological types significantly influence ecological sensitivity, with pure dolomite areas showing the highest sensitivity and a strong negative association between ecological sensitivity and vegetation productivity increases (p < .05). Pure limestone regions demonstrate greater adaptability. Additionally, the interaction between regolith thickness and soil depth modulates sensitivity, with sensitivity decreasing significantly when soil thickness exceeds 0.57 m. These findings provide insights into the mechanisms driving ecological sensitivity and their implications for vegetation resilience to climate change.