Intensified water stress driven by greenhouse gas-induced warming plays a pivotal role in regulating terrestrial vegetation growth across arid-to-humid transition zones, with significant implications for the global carbon cycle. However, the shifting sensitivity of the vegetation productivity to a warming climate remains poorly understood. Since the early 2000s, Northeast (NEA) has experienced pronounced reductions in gross primary production (GPP), primarily attributed to notable soil moisture (SM) decreases and water vapor deficit (VPD) increases. Our study explored the relative dominance of SM and VPD to GPP variations through random forest (RF) algorithm, and demonstrated distinct varied responses of vegetation growth across aridity gradients: vegetation growth in arid regions is predominantly influenced by SM, while VPD exerts a stronger influence in semi-arid to humid zones. Under warming and drying conditions, GPP sensitivity to water availability intensified, with vegetation in semi-arid to humid zones becoming increasingly vulnerable to VPD. As drylands expand and climate variability intensifies, understanding these sensitivities is essential for predicting ecosystem vulnerability and assessing vegetation responses to future climate scenarios.