Vegetation dynamics along drydowns under shifting drought regimes

Climate warming is reshaping drought regimes and their impacts on terrestrial vegetation, yet most large-scale studies still describe drought–vegetation relationships using long-term mean states or trend metrics that integrate over many processes and do not reveal how ecosystems reorganize during individual drydowns. Here, we adopt an event-scale perspective by explicitly tracking vegetation responses along discrete drydown events. We identify droughts as periods of increasing cumulative water deficit (CWD), defined as the running imbalance between precipitation and actual evapotranspiration, and we quantify hydroclimatic forcing using the severity of the most extreme drydown in each year, expressed as the annual maximum absolute CWD (CWDmax, mm). Over 2000–2023, significant CWDmax trends occur in 16.51%of vegetated grid cells, corresponding to 18.71% of total vegetated land area (area-weighted). Significant increases in CWDmax account for 7.39% of vegetated grid cells (7.88% of vegetated land area) across much of the Northern Hemisphere, the Sahel and the Amazon, while significant decreases account for 9.12% of grid cells (10.58% of land area). Positive CWDmax trends indicate that the most severe annual drydowns are reaching larger absolute deficits over time, consistent with intensification of hydroclimatic water stress, whereas negative trends indicate a weakening of extreme deficits; with typical magnitudes of 24.5 mm per decade, and 50% of significant trends falling between 14.1 and 37.9 mm per decade (IQR). To characterise vegetation responses at the event scale, we track satellite-based surface greenness (Enhanced Vegetation Index, EVI) along each year’s most severe drydown and fit smooth EVI–CWD trajectories to locate productivity peaks and subsequent critical losses. We define EVIpeak​ as the fitted maximum greenness and its associated deficit (i.e., CWDcritical) along the event trajectory and EVIcritical as the greenness level at a standardized loss threshold (90% of EVIpeak). Across climates, the fractional decline from peak to critical states is relatively conserved (~10–24%), yet the cumulative deficit required to reach that decline spans a five-fold range (~40–200 mm), highlighting strong hydroclimatic modulation of event-scale greenness loss. We summarise long-term changes in these within-event thresholds into five threshold pathways : Stable (no trend), Greening and Browning (co-trending EVIpeak and EVIcritical), and two decoupled modes: Overshoot (EVIpeak↑, EVIcritical↓) and Compensatory (EVIpeak↓, EVIcritical↑).  While ~80% of vegetated land area shows no detectable change (Stable), a latitudinal band (~50–65°N) exhibits a marked increase in non-stationary pathways, with Overshoot, Compensatory, and Browning over-represented; within 60–65°N, Browning reaches ~21.9%. Across regions where drought severity is intensifying in absolute terms (positive CWDmax trends), mid- to high-latitude systems more often shift toward Overshoot or Browning, whereas many dryland systems remain largely Greening. By uniting trends in absolute drought severity with within-event productivity thresholds, the framework provides state-dependent indicators of ecosystem pathways, highlighting where event-scale buffering appears stationary and where threshold dynamics indicate increasing vulnerability to greenness loss.