Nature 514, 213–217 (2014)

Feedback mechanisms between the terrestrial biosphere and the atmosphere — related to the climatically modulated uptake and release of carbon by vegetation and the soil — remain a key source of uncertainty in climate change projections. The residence time of carbon in terrestrial ecosystems is one of the factors that determine the nature of these feedbacks so misrepresentation in coupled climate/carbon-cycle models could bias projections of climate change.

Nuno Carvalhais from the Max Planck Institute for Biogeochemistry, Germany, and co-authors undertook a global assessment of ecosystem carbon turnover times. Their observationally based analysis provides spatially explicit estimates that take account of vegetation and soil organic carbon stocks and fluxes. They find a clear dependence of turnover time on temperature, in line with expectations. Surprisingly, they also find a similarly strong association with precipitation. Simulated ecosystem carbon turnover times varied widely but on average were found to underestimate the global carbon turnover time by 36%. Furthermore, the models generally did not reproduce the strength of the spatial relationship with precipitation. These findings suggest that climate/carbon-cycle feedbacks depend more strongly on hydrological changes than is currently represented in state-of-the-art climate models.