Abstract: Understanding vegetation sensitivity to water deficit is essential for assessing ecosystem vulnerability and adaptive capacity. Based on flux and meteorological data from 77 global sites, we developed a new approach that combines percentile and standard deviation methods to characterize precipitation (PPT) and soil water content (SWC) deficit conditions. Simultaneously, we applied the SWH model to simulate evapotranspiration (ET) processes, separating transpiration (T) from evaporation (E). Spatially explicit analysis revealed significant variations in vegetation sensitivity to PPT and SWC deficits (S_PPT and S_SWC) across ecosystem types, generally intensifying with increasing deficit severity. Notably, nearly half of the sites exhibited contrasting responses, with positive S_SWC but negative S_PPT. This divergence was particularly pronounced in forest ecosystems, likely due to precipitation legacy effects. Moreover, the study revealed the unexpected increase in gross primary productivity (GPP) under SWC deficit conditions at certain sites, which was mechanistically linked to increased T, T/ET, and water use efficiency. We proposed that vegetation exhibits growth inertia, whereby plants that thrive under favorable prior conditions can sustain higher soil water utilization rates and GPP, which in turn leads to soil moisture depletion. Specifically, vegetation actively regulates water use to maintain productivity through transpiration-mediated adjustments, challenging conventional views of passive drought responses. To sum up, these results collectively highlighted that SWC surpasses PPT in determining vegetation sensitivity to water deficit, and that comprehensive vegetation drought sensitivity assessments must explicitly consider the differential impacts of E and T on SWC dynamics.

Key words: vegetation, sensitivity, soil water content, precipitation, gross primary productivity