PP41A-1332:
Carbon isotopes of plant biomarkers record past changes in the carbon cycle, but separating signal from noise is key to reducing uncertainties
Thursday, 18 December 2014
Aaron F Diefendorf, University of Cincinnati, Department of Geology, Cincinnati, OH, United States, Katherine H Freeman, The Pennsylvania State University, Department of Geosciences, University Park, PA, United States, Scott L Wing, Smithsonian Instituition, Department of Paleobiology, Washington, DC, United States and Ellen D Currano, University of Wyoming, Department of Botany, Laramie, WY, United States
Abstract:
The carbon isotopic composition of plant biomarkers (δ13C) can provide unique insights into the past carbon cycle perturbations and associated climate change, however local records are influenced by ecological processes, local climate, as well as changes in the carbon isotope composition of the atmosphere. To examine the sources and amounts of geographic variation, we focused on long-term changes in the carbon cycle. We combined modern calibrations, δ13C of biomarkers in sediment, and Monte Carlo analyses to measure and predict the fractionation of carbon isotopes by plants (Δleaf) and to estimate error. We used data from multiple sites of different ages, in the western U.S. For each age and location, Δleaf was calculated from the δ13C of plant biomarkers and atmospheric δ13C values inferred from marine carbonates. Δleaf values calculated from n-alkanes and triterpenoids (angiosperm biomarkers) were found to be the same at each site. Δleaf calculated from diterpenoids (conifer biomarkers) was 2‰ lower. This is consistent with differences in Δleaf between living angiosperms and conifers. Predicted Δleaf values, from modern calibrations and paleoclimate data, were consistently offset (0.7‰) from measured values indicating that modern calibrations are useful for reconciling past changes in plant fractionation and that vegetation and precipitation, like modern plants, were the key controls on Δleaf in ancient vegetation. However, uncertainties in the measured and predicted Δleaf values were very large (>2‰, 1σ). A one-at-a-time sensitivity analysis indicates that ‘biological noise’ in modern calibrations explains most of this uncertainty. If the full extent of this biological noise were transferred to sediments, then extracting signal from noise would be challenging. However, we speculate that the process of deposition homogenizes variability at the leaf and tree level thereby reducing ‘biological noise’ observed in modern calibrations.
