OOS 43-10
Combining long-term observation and manipulation with modeling to predict carbon cycle feedback to climate in terrestrial ecosystems

Wednesday, August 12, 2015: 11:10 AM
329, Baltimore Convention Center
John Harte, Energy and Resources Group, University of California, Berkeley, CA
Scott R. Saleska, Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ
Charlotte Levy, Ecology and Evolutionary Biology, Cornell University, Ithaca, NY
Background/Question/Methods

Ecosystem responses to climate change can exert positive or negative feedbacks to climate, mediated in part by slow-moving factors such as induced shifts in vegetation community composition.  Long-term experimental manipulations and observations of ecological trends in time and space can be used to predict such ecosystem responses, but the  realism of the former method, and the capacity of the latter method to yield mechanistic understanding, have been questioned.  Moreover, neither method may be providing insight at relevant temporal scales.  Here, using a combination of 23 years of experimental and observational data, and robust modeling, we answer the following questions: Can we causally attribute already observed decadal-scale ecological trends to a changing ambient climate? Can we predict both short and long-term future ecosystem responses to global warming? Can we predict the sign and magnitude of future ecologically-mediated carbon cycle feedback to the climate?  

Results/Conclusions

We document multi-decadal trends in ambient ecosystem plots toward earlier snowmelt, increasing dominance of woody plants relative to forbs, and diminished soil carbon.  Over the same time period, and with respect to all these variables, experimentally heated plots respond qualitatively similarly, but at a faster rate.  The findings demonstrate the realism of an experimental manipulation, allow attribution of a climate cause to observed ambient ecosystem trends, and demonstrate how a combination of long-term study of ambient and experimentally heated plots allows identification of driving mechanisms and thus realistic predictions of the conditions under which ecosystems are likely to become carbon sources or sinks over a range of future timescales.