Hydrologic implications of vegetation response to elevated CO2 in climate projections

Climate model projections using offline aridity and/or drought indices predict substantial terrestrial drying over the twenty-first century1,2,3,4,5,6,7,8,9,10,11. However, these same models also predict an increased runoff12,13,14,15. This contradiction has been linked to an absence of vegetation responses to an elevated atmospheric CO2 concentration [CO2] in offline impact models12,14,16,17. Here we report a close and consistent relationship between changes in surface resistance (rs) and [CO2] across 16 CMIP5 models. Attributing evapotranspiration changes under non-water-limited conditions shows that an increase in evapotranspiration caused by a warming-induced vapour pressure deficit increase18 is almost entirely offset by a decrease in evapotranspiration caused by increased rs driven by rising [CO2]. This indicates that climate models do not actually project increased vegetation water use under an elevated [CO2], which counters the perception that ‘warming leads to drying’ in many previous studies1,2,3,4,5,6,7,8,9,10,11. Moreover, we show that the hydrologic information in CMIP5 models can be satisfactorily recovered using an offline hydrologic model that incorporates the [CO2] effect on rs in calculating potential evapotranspiration (EP). This offers an effective, physically-based yet relatively simple way to account for the vegetation response to elevated [CO2] in offline impact models