MACROSTRATIGRAPHIC CONSTRAINTS ON THE GLOBAL CARBON CYCLE

Carbon isotopic measurements (δ¹³C) of marine carbonates are the most commonly used proxies for studying the ancient carbon cycle. Secular change in high fidelity δ¹³C time-series are typically interpreted as reflecting the relative importance of two modes of carbon burial: (1) organic matter and (2) carbonate minerals. Recent work, however, has challenged this paradigm by suggesting that authigenic carbonate, precipitated from isotopically depleted organic matter metabolized during early burial diagenesis, may be an important third sink for carbon. If this model is correct, our understanding of the carbon cycle and global redox budgets needs to be redressed at a fundamental level. However, testing this model is challenging because authigenic carbonate can be disseminated over a large volume of sediment. Here we use 23,813 rock units, distributed among 949 geographic regions in the North American component of the Macrostrat database, to constrain the potential magnitude of the authigenic carbonate burial effect. We do this by empirically calibrating the time-varying magnitudes of each of the three carbon sinks. Because sedimentary lithologies vary in their potential for burying phases of carbon, our analysis focuses on quantifying the North American burial flux of 55 individual sedimentary lithotypes. Model values for weight percent authigenic and organic carbon were assigned to each lithology based upon more than 15,000 Macrostrat-linked measurements derived from the USGS National Geochemical Database. Across a range of reasonable parameters for weight percent organic and authigenic carbonate in siliciclastic units, and their δ¹³C offsets from seawater, the predicted δ¹³C of seawater DIC reproduces long-term trends observed in composite δ¹³C time-series measured from marine carbonates for the past 900 Myr. Incorporating additional geochemical data into Macrostrat (e.g., unit-specific isotopic measurements and total organic/inorganic carbon abundance) is necessary to improve the accuracy and precision of model predictions.