Marine Geology 217 (2005) 323–338 www.elsevier.com/locate/margeo Biological overprint of the geological carbon cycle Miriam E. Katza,T, James D. Wrighta, Kenneth G. Millera, Benjamin S. Cramerb, Katja Fennelc, Paul G. Falkowskia,c aDepartment of Geological Sciences, Rutgers University, 610 Taylor Road, Piscataway, NJ 08854, United States bInstitute of Geology and Paleontology, Tohoku University, Aoba, Aramaki, Sendai 980-8578, Japan cInstitute of Marine and Coastal Sciences, Rutgers University, 71 Dudley Road, New Brunswick, NJ 08901-8521, United States Received 21 May 2004; received in revised form 11 August 2004; accepted 2 March 2005 Abstract The oxidation of Earth’s atmosphere is coupled to the net sequestration of organic matter, which is related to the relative fractions of organic carbon ( forg) and carbonate ( fcarb) buried in marine sediments. These fractions can be inferred from carbon 13 13 13 isotope data. We present bulk sediment d C records of carbonate (d Ccarb) and organic carbon (d Corg) with a compilation of evolutionary trajectories of major eucaryotic phytoplankton for the past 205 million years. Our analysis indicates that changes in phytoplankton community structure, coupled with the opening of the Atlantic Ocean basin and global sea-level rise, increased the efficiency of organic carbon burial beginning in the Early Jurassic; in turn, this organic carbon burial increased the oxidation state of Earth’s surface while drawing down atmospheric CO2 levels (assuming no substantial negative feedbacks). The net oxidation and CO2 drawdown appear to be related to the opening phase of the current Wilson cycle, where the newly formed passive plate margins store organic matter for hundreds of millions of years. This process should reverse during the closing phase of the Wilson cycle, when the continents reassemble and the Atlantic Ocean basin closes. The associated oxidation and storage of organic matter have contributed to the long-term depletion of CO2, which was a key factor that selected C4 photosynthetic pathways in marine and terrestrial ecosystems in the latter part of the Cenozoic; these pathways increasingly 13 13 influenced d Corg, and ultimately contributed to the reversal of the long-term trend in d Ccarb. D 2005 Elsevier B.V. All rights reserved. Keywords: carbon isotopes; organic carbon burial; oxidation state; Wilson cycle; phytoplankton 1. Introduction The d13C signature of the ocean’s mobile carbon 13 T Corresponding author. Tel.: +1 732 445 3445; fax: +1 732 445 reservoir is controlled by the d C signatures and 3374. fluxes of carbon sources and sinks on timescales longer E-mail address: [email protected] (M.E. Katz). than the residence time of carbon in the oceans (~180 0025-3227/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.margeo.2004.08.005 324 M.E. Katz et al. / Marine Geology 217 (2005) 323–338 kyr). The relationship between the sources (input) and 2. Methods sinks (output) of the geological carbon cycle is typically quantified using the standard equation that 2.1. Site and sample selection provides the backbone of carbon isotope models (e.g., 13 Berner and Kothavala, 2001; Kump and Arthur, 1999): We measured d Ccarb on bulk sediment samples from Jurassic and Cretaceous sections (Figs. 1 and 13 13 13 13 2; Appendix A, web-archived at http://mychronos. f 4d C þ f 4d C ¼ f 4d C þ f 4d C w w v v carb carb org org chronos.org/~miriamkatz/20040728/). Bulk sediment ð1Þ samples were analyzed because they best characterize the inorganic carbon outflow from the ocean/atmos- where f=fraction, w=weathering, v=volcanic/hydro- phere/biosphere, and provide the average d13C of the thermal, carb=carbonate, and org=organic carbon. total carbonate produced and preserved in the marine Carbon is supplied to the ocean through outgassing system (Shackleton, 1987). This is the signal that is from hydrothermal/volcanic activity and from erosion needed for the purposes of this paper; therefore, we of continental rocks, while carbon is removed from the use bulk sediment isotope records as a proxy for the ocean through deposition of marine sediments. Over average d13C of the inorganic output, allowing us to time, these sediments integrate large kinetic fractiona- monitor long-term changes in the global carbon cycle tions from photosynthetic reduction of CO2 to organic through time (e.g., Shackleton, 1987). Although the matter with small thermodynamic fractionations from mobile carbon reservoir in the deep ocean is ion exchange reactions in carbonate precipitation. The substantially larger (36700Â1015 g today) than the relative fractions of carbonate and organic carbon carbon reservoirs in the surface ocean and atmosphere buried in marine sediments are inferred from d13C (670Â1015 g and 720Â1015 g, respectively) (e.g., records of carbonates and organic matter in sedimen- (Falkowski and Raven, 1997), relatively little sedi- tary rocks (Hayes et al., 1999; Kump and Arthur, mentary carbonate is produced in deepwaters. There- 13 13 13 1999). Changes in marine d Ccarb and d Corg fore, a d C record generated from deepwater benthic through time serve as archives of changes in carbon foraminifera does not provide a record of the average sources and sinks (for detailed summaries of the d13C of the total sedimentary carbonate preserved in carbon cycle, see Hayes et al., 1999; Kump and the marine system. Rather, d13C analyses of benthic Arthur, 1999). foraminifera record dissolved inorganic carbon (DIC) In this study, we document and discuss long-term in deepwaters, and can be used to reconstruct deep- 13 trends (tens to hundreds of million years) in d Ccarb water circulation changes through time (e.g., Miller et 13 and d Corg records for the Jurassic–Cenozoic using al., 1987; Zachos et al., 2001); such a reconstruction is both new and published data. We interpret the long- not the goal of this paper. A productivity signal can be term trends based on GEOCARB III (Berner and extracted from benthic foraminiferal d13C only by Kothavala, 2001) model simulations that use these comparing it with planktonic foraminiferal d13C from 13 13 d Ccarb and d Corg data, with additional information phosphate-free surface waters. Even then, it is only a from comparisons with phytoplankton diversity proxy for the carbon to phosphorous ratio (C/P) records and geological proxies. Our goal is to (Broecker and Peng, 1982), which is proportional to investigate the geological and biological processes productivity only as a function of stability/mixing. that interacted through time to produce these records This would be impossible to do for the Jurassic– of the global carbon cycle. A series of sensitivity tests Cenozoic because planktonic foraminifera did not based on Eq. (1) allows us to place constraints on evolve until the Cretaceous and there is a lack of well- potential variations in the carbon sources and sinks. enough preserved foraminifera for a continuous Each of these sensitivity tests is designed to predict Jurassic–Cretaceous isotope record. 13 the maximum response to changing a single variable, Similarly, we do not use d Ccarb records generated and therefore does not take into account potential from specific organisms (e.g., belemnites, oysters, and feedbacks through time that may have muted this foraminifera), which reflect the different environ- response. ments where each of those organisms lived (e.g., M.E. Katz et al. / Marine Geology 217 (2005) 323–338 325 δ13C‰ -2 0 2 4 0 Quat. late Pl. early -180 -120 -60 0 60 120 180 late Site 525B 90 90 middle 60 Mochras 60 Site 528 534 402 Mioc. early Site 527 30 30 20 137 late Site 516F 0 0 Site 137 -30 -30 Oli. early 516 525-8 Site 402A -60 -60 late Site 534A 40 -90 -90 middle Mochras -180 -120 -60 0 60 120 180 Tertiary SSA filter present day early Eocene late -180 -120 -60 0 60 120 180 60 90 90 Pal. early 60 Mochras 60 Maastrichtian 534 402 30 30 137 Campanian 0 0 80 -30 525-8 -30 Santonian -60 516 -60 Late Coniacian Turonian -90 -90 -180 -120 -60 0 60 120 180 Cenomanian 100 50 Ma reconstruction Albian -180 -120 -60 0 60 120 180 90 90 Time (Ma) Aptian 60 Mochras 60 120 534 Cretaceous Barremian 30 402 30 Early 137 Hauterivian 0 0 -30 -30 Valanginian 516 525-8 -60 -60 140 Berriasian -90 -90 -180 -120 -60 0 60 120 180 Tithonian 100 Ma reconstruction Kimmeridgian Late Oxfordian 160 Callovian -180 -120 -60 0 60 120 180 Bathonian 90 90 60 60 Bajocian Mochras 30 534 30 Aalenian 180 0 0 Toarcian -30 -30 Jurassic -60 -60 Pliensbachian -90 -90 200 EarlySinemurian Middle -180 -120 -60 0 60 120 180 Hettangian 150 Ma reconstruction Fig. 1. Composite bulk sediment d13C record for the Jurassic through the Cenozoic (see Methods for site selection criteria). Mesozoic d13C data (this study) and Cenozoic d13C data (Shackleton and Hall, 1984) are primarily from open ocean Atlantic Deep Sea Drilling Project (DSDP) boreholes (see Methods). Data (Appendix A) are web-archived (http://www.mychronos.chronos.org/~miriamkatz/20040728/). Site locations are shown in a series of paleogeographic reconstructions at 50 myr intervals (http://www.odsn.de/odsn/index.html). We use least squares regression 13 13 13 (95% confidence interval) to determine the long-term trends in d Ccarb, where x=age and y=d Ccarb: (1) Dd Ccarb=À2.52x for 0–15 Ma: 13 y=(0.168F0.024)x+(0.049F0.17), R=0.89; (2) Dd Ccarb=1.1x for 16–205 Ma: y=(À0.006F0.001)x+(2.64F0.12), R=0.38. We note that 13 including the Lower Jurassic section (Mochras borehole data) in the linear regression produces a lower rate of increase in d Ccarb, which yields 13 a more conservative estimate of the magnitude of the long-term increase.