The interactions between hydrological and ecological processes are key issues to improve our predictions of ecosystem responses to increasing droughts. However, predicting the dynamics and the impacts of vegetation water stress remains challenging because of complex ecohydrological feedbacks. The ecohydrological optimality approach proposes that functional adjustments within plant communities may buffer the increase in vegetation water stress despite local water shortage. This study aimed to test whether vegetation water stress may be invariant across contrasting plant communities, reflecting possible optimality processes. We addressed the following question: does a lower soil water storage capacity under the same climate generate greater vegetation water stress over time? We hypothesized that vegetation water stress would be buffered around a low and constant level through the adjustment of vegetation biomass productivity net primary productivity (NPP), evapotranspiration (ET) and/or water-use efficiency (WUE) in relation with local soil water storage capacity. We monitored 12 native plant communities distributed along a gradient of soil water storage capacity (ranging from 20 mm to 120 mm) during five successive years. Net primary productivity, ET, WUE as well as soil water dynamics were assessed and modelled for each plant community throughout the 5 years of study. Vegetation water stress was determined for each plant community as the deviation of between actual ET and their maximum ETm rate achieved under non-limiting conditions. We found that NPP and ET were together proportionally related to local soil water storage capacity across the 5 years of study while WUE did not differ between plant communities. Vegetation water stress was found quite similar for all plant communities whatever the soil water storage capacity. These results suggested that vegetation water stress was strongly buffered by the community-level plant growth rates and total water use along th