Intensifying compound dryness amplifies soil moisture control on water-use efficiency and vapor pressure deficit control on productivity across global terrestrial ecosystems
Abstract Dryness in soil and atmosphere individually imposes substantial stress on vegetation. However, their compound impact on ecosystems remains poorly understood. This work addresses how increasing compound land–atmosphere dryness modulates the relative roles of soil moisture (SM) and vapor pressure deficit (VPD) in regulating vegetation carbon (GPP: gross primary productivity) and water (WUE: water use efficiency) fluxes globally. We find that the likelihood of CLAD is highest over deciduous forests in dry regions, shrublands in humid regions, and evergreen forests in very-humid regions. Both GPP and WUE exhibit stronger negative anomalies in dry deciduous forests. In very-humid regions, negative GPP anomalies are more pronounced over shrubland–grasslands, and similarly for WUE in evergreen forests, highlighting biome-specific responses. We further observe a weakening of SM-VPD coupling with increasing dryness in wetter regions, predominantly over forests, whereas the opposite pattern emerges in non-forest ecosystems in dry regions. Using non-linear partial correlation analysis, we find that SM consistently exerts a stronger control on vegetation fluxes than VPD across all dryness-levels. However, the control of SM and VPD differs with increasing compound dryness. As compound dryness intensifies, the influence of SM on WUE amplifies 2.5-fold, whereas the impact of VPD on GPP rises 1.7-fold. Overall, GPP response to compound dryness is stronger in shallow-rooted ecosystems in humid regions, whereas the same for WUE is in deep-rooted trees in dry regions. The contrasts between SM and VPD impacts are more prominent over dry deciduous forests and less pronounced in grasslands in very humid regions. These findings demonstrate that the hierarchy between SM and VPD in controlling ecosystem carbon and water fluxes is not static but is intensity-dependent and biome-stratified. This work provides crucial benchmarks for improving compound stress representations in land-surface and terrestrial biosphere models, which could help to reduce uncertainty while assessing land-carbon sink due to increasing aridity.