Using imaging spectroscopy to predict above- and below-ground carbon fluxes: A case study in alpine meadows

Sandra M. Duran, Ecology & Evolutionary Biology, University of Arizona, Sandra M. Duran, Ecology & Evolutionary Biology, University of Minnesota, Nicola Falco, Climate & Ecosystem Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, Haruko Wainwright and Eoin Brodie, Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, Sergio Marconi, School of Natural Resources and Environment, University of Florida, Gainesville, FL, Amanda N. Henderson, Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ, Heidi Steltzer, Biology, Fort Lewis College, Durango, CO, Scott R. Saleska, Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ, Brian Enquist, The Santa Fe Institute, Santa Fe, NM

Background/Question/Methods
Developing models that predict carbon exchange, the balance between photosynthetic uptake and release of carbon dioxide by respiration from autotrophs and heterotrophs, is critical for the terrestrial carbon cycle. Alpine meadows are ecosystems where soils are carbon-rich, and show sharp transitions in ecosystem structure due to the high extreme climatic conditions that the biota experiences such as snowmelt date, temperature seasonality, and length of the growing season. Thus, identifying indicators of vegetation change in alpine meadows is essential to understand how community changes affect rates of carbon and water cycling. We quantified above- and below-ground carbon fluxes along an elevation gradient (2700-3500 m) in 38 plots (1.2 m2) in the East River Watershed in Colorado at the peak of the growing season. We measured reflectance of plant communities at leaf, canopy and from airborne imaging to assess the different detection methods to predict rates of carbon cycling. We then evaluate whether reflectance, spectral diversity indices and plant traits derived from reflectance are able to predict rates of net ecosystem exchange (NEE), gross primary productivity (GPP), ecosystem respiration (Reco), soil respiration (SoilR) and soil organic carbon (SOC).
Results/Conclusions
Overall we found that spectral diversity was not a good predictor of carbon fluxes. We found a strong relationship between plant traits such as leaf water content, leaf area and specific leaf area and NEE using all three detection methods, with models using airborne data as good as models using traits derived from reflectance at the canopy level. Similar results were found for GPP. However, rates of Reco and SoilR were better explained from environmental variation in soil temperature, air temperature and soil moisture, with poor relationships between reflectance, spectral diversity or plant traits. We did find a positive relationship between SOC and spectra reflectance, with the strongest relationship between SOC and airborne imaging, and reflectance explaining as much of the variation in SOC as the environmental variables. Combine these results indicate the potential of imaging spectroscopy to predict rates of carbon gain and below-ground processes such as SOC. Furthermore, these analysis and the strong relationships between carbon fluxes and airborne imaging highlight the potential for developing scaling-up models of NEE and carbon gain using hyperspectral satellite data and improve predictions on how alpine communities will respond to global changes.