Future projections suggest an increase in drought globally with climate change. Current vegetation models typically regulate the plant photosynthetic response to soil moisture stress through an empirical function, rather than a mechanistic response where plant water potentials respond to changes in soil water. This representation of soil moisture stress may introduce significant uncertainty into projections for the terrestrial carbon cycle. We examined the use of the soil moisture limitation function in historical and future emissions scenarios in nine Earth system models. We found that soil moisture-limited productivity across models represented a large and uncertain component of the simulated carbon cycle, comparable to 3-286% of current global productivity. Approximately 40-80% of the intermodel variability was due to the functional form of the limitation equation alone. Our results highlight the importance of implementing mechanistic water limitation schemes in models and illuminate several avenues for improving projections of the land carbon sink.
USDA National Institute of Food and Agriculture, Agricultural and Food Research Initiative Competitive Programme, Ecosystem Services and Agro-ecosystem Management, grant no. 2017-05521, National Science Foundation grant 1714972, US Department of Energy, Office of Science, Office of Biological and Environmental Research, Terrestrial Ecosystem Science (TES) Program award DE-SC0014363 , National Science Foundation Award 1151102 , and USDA National Institute of Food and Agriculture Postdoctoral Research Fellowship Grant No. 2017-07164