Increasing Vegetation Sensitivity to Precipitation Patterns in China’s Drylands

Global warming is intensifying precipitation variability, yet how vegetation sensitivity to different precipitation patterns changes through time remains unclear. Here, we quantified vegetation sensitivity to total precipitation (SITpr), precipitation frequency (SIFrq), and consecutive dry days (SICDD) across China’s drylands from 1982 to 2020 using satellite-derived leaf area index (LAI) and high-resolution precipitation data. Temporal changes in these sensitivity indices were assessed using a moving-window framework, and their environmental associations were examined using random forest models combined with Shapley Additive Explanations (RF-SHAP). Our results showed that all sensitivity indices increased significantly over the past four decades and exhibited non-stationary temporal changes, with breakpoints detected in 2000 and 2007 for SITpr, and in 1997, 2004, and 2011 for SICDD. SITpr and SIFrq decreased along the aridity gradient, whereas SICDD was highest in relatively moist subhumid areas. Vegetation greening was associated with intensified sensitivity trends, particularly for SICDD, whereas SITpr and SIFrq exhibited nonlinear responses along greening gradients. RF-SHAP analysis identified distinct environmental associations among the three sensitivity indices. SITpr was primarily associated with potential evaporation and temperature, whereas SIFrq and SICDD were more closely associated with atmospheric water demand and soil conditions. These findings suggest that continued greening in drylands may coincide with increasing sensitivity to prolonged dry spells, highlighting the need to consider precipitation timing and dry spells when evaluating dryland ecosystem vulnerability.