Disentangling the dominance of atmospheric and soil water stress on vegetation productivity in global drylands

• Intensified decline in vegetation productivity concurs with the acceleration of VPD increase and SM decrease during 1982–2015. • Similar thresholds are captured in the sensitivity of vegetation productivity to atmospheric and soil water stress concerning multiyear average growing-season atmospheric water demands. • The importance of atmospheric water stress on vegetation productivity increased across global drylands in the context of global warming. Atmospheric water demands and soil moisture are crucial components of vegetation water stress, especially for dryland ecosystems where water availability is a severe constraint for their sustainable development. Although the effects of water stress on vegetation productivity and the underlying ecological mechanism have been recognized extensively, the relative contribution of atmospheric and soil water stress changes on vegetation productivity are still in debate, and their temporal dynamics remain unclear. To fill this knowledge gap, here we collected remote sensing meteorological, root-zone soil moisture and vegetation productivity proxies (represented by kernel Normalized Difference Vegetation Index, kNDVI and Nirv-GPP) during the period 1982–2015 to quantify the sensitivity of vegetation productivity to atmospheric water stress (represented by vapor pressure deficit (VPD)), soil water stress (represented by root-zone soil moisture (SM)) and their interaction (represented by SM × VPD) across global drylands, based on a series of dedicated factorial experiments within random forest (RF) framework. The results showed that soil water stress exerted predominant influence on vegetation carbon uptake spatially. Moreover, a rising sensitivity of vegetation productivity to SM and a declining sensitivity to VPD were widely captured. We also found that atmospheric water stress dominated the temporal change in the sensitivity of vegetation productivity to their interactive effect, indicating a weakening importance of soil water stress. Our research highlights the increasing importance of atmospheric water stress and enhances our understanding of how vegetation carbon and water cycles respond to climate change in dryland ecosystems.