Changing atmospheric CO2 concentration was the primary driver of early Cenozoic climate

Journal name:
Nature
Volume:
533,
Pages:
380–384
Date published:
DOI:
doi:10.1038/nature17423
Received
Accepted
Published online
Corrected online

The Early Eocene Climate Optimum (EECO, which occurred about 51 to 53 million years ago)1, was the warmest interval of the past 65 million years, with mean annual surface air temperature over ten degrees Celsius warmer than during the pre-industrial period2, 3, 4. Subsequent global cooling in the middle and late Eocene epoch, especially at high latitudes, eventually led to continental ice sheet development in Antarctica in the early Oligocene epoch (about 33.6 million years ago). However, existing estimates place atmospheric carbon dioxide (CO2) levels during the Eocene at 500–3,000 parts per million5, 6, 7, and in the absence of tighter constraints carbon–climate interactions over this interval remain uncertain. Here we use recent analytical and methodological developments8, 9, 10, 11 to generate a new high-fidelity record of CO2 concentrations using the boron isotope (δ11B) composition of well preserved planktonic foraminifera from the Tanzania Drilling Project, revising previous estimates6. Although species-level uncertainties make absolute values difficult to constrain, CO2 concentrations during the EECO were around 1,400 parts per million. The relative decline in CO2 concentration through the Eocene is more robustly constrained at about fifty per cent, with a further decline into the Oligocene12. Provided the latitudinal dependency of sea surface temperature change for a given climate forcing in the Eocene was similar to that of the late Quaternary period13, this CO2 decline was sufficient to drive the well documented high- and low-latitude cooling that occurred through the Eocene14. Once the change in global temperature between the pre-industrial period and the Eocene caused by the action of all known slow feedbacks (apart from those associated with the carbon cycle) is removed2, 3, 4, both the EECO and the late Eocene exhibit an equilibrium climate sensitivity relative to the pre-industrial period of 2.1 to 4.6 degrees Celsius per CO2 doubling (66 per cent confidence), which is similar to the canonical range (1.5 to 4.5 degrees Celsius15), indicating that a large fraction of the warmth of the early Eocene greenhouse was driven by increased CO2 concentrations, and that climate sensitivity was relatively constant throughout this period.

At a glance

Figures

  1. Currently available Eocene atmospheric CO2 records and benthic foraminiferal δ18O values.
    Figure 1: Currently available Eocene atmospheric CO2 records and benthic foraminiferal δ18O values.

    a, CO2 reconstructions5, 6, 11, 12, 19, 20 with the 2 s.d. uncertainties. b, Benthic foraminiferal δ18O (ref. 1). The green line shows the five-point moving average of the grey line, with data age modified to account for the revised timing of the MECO30. Lower benthic δ18O values denote warmer bottom-water temperatures (or less ice volume, which is assumed to be minimal in the Eocene). Error bars on δ18O are smaller than the symbols (Methods).

  2. Eocene planktonic foraminiferal multi-species stable isotope arrays.
    Figure 2: Eocene planktonic foraminiferal multi-species stable isotope arrays.

    ae, Foraminiferal δ13C versus δ18O for each time slice. fj, Foraminiferal δ11B values versus δ18O inferred water depths for each time slice. h shows the anomalous behaviour of Globigerinatheka spp. (black diamonds) in both δ13C and δ11B (see Methods). The blue arrow in j shows the correction of T. cerroazulensis δ11B to shallower water values (Methods). Shaded areas in fj surround the species used in Fig. 3a. Errors on δ13C and δ18O are smaller than symbols. Error bars on δ11B are 2 s.d. based on long-term precision (Methods). VPDB, the Vienna Pee-Dee belemnite standard. For a full species list see Supplementary Table 1.

  3. New atmospheric CO2 reconstructions from shallow planktonic foraminiferal δ11B.
    Figure 3: New atmospheric CO2 reconstructions from shallow planktonic foraminiferal δ11B.

    a, δ11B of shallower foraminifera (symbols as in blue-shaded areas of Fig. 2) and asterisks represent the average of those for each time slice. Blue squares and orange circles represent the warmest species used in b. Ref. 12 data are offset (Supplementary Table 2, Methods). b, Atmospheric CO2 assuming δ11Bc = δ11Bborate using the warmest species of each time slice. The red line is based on ref. 11. c, Benthic foraminiferal δ18O (ref. 1) (as in Fig. 1b). Age error bars (b) are based on the ages of the nearest datums, δ11B errors are as in Fig. 2, and the CO2 uncertainty is based on the 95% confidence of Monte Carlo error propagation (Methods).

  4. CO2 as a driver of latitudinal cooling in the Eocene, and ECS analyses of the EECO and late Eocene time slices.
    Figure 4: CO2 as a driver of latitudinal cooling in the Eocene, and ECS analyses of the EECO and late Eocene time slices.

    a, Evolving relationship between SST14 for high and low latitudes and the CO2 forcing of each of our time slices relative to the EECO, with linear regression fits and coefficients of determination (R2). b, Apparent latitudinal SST sensitivity for the last 520 kyr (ref. 13). The dashed line is the second-order polynomial through the SST sensitivity data (grey crosses) of ref. 13. Red shading shows the SST sensitivity estimates averaged as a low-latitude mean, and blue lines show the high-latitude mean. c, Reconstructed (lines) and estimated (symbols) SST relative to 53.2 Myr ago. Symbols are estimated using the respective CO2 reconstructions and the average low-latitude (red circles) and high-latitude (blue triangles) SST sensitivities of b. Bold lines show the reconstructed long-term mean SST estimates using the TEX86 proxy14 at high latitudes (blue >55°) and low latitudes (red <30°) relative to the SST ~53 Myr ago. Error bars represent full propagation of errors at 95% confidence. d, Range in mean surface temperature change ΔT for early and late Eocene, corrected to remove the changes due to slow feedbacks2, 4, 21, 29. e, Forcing compared to the pre-industrial period calculated using our CO2 reconstructions for the time slices 53.2 Myr ago (early Eocene) and 36.9 Myr ago (late Eocene). f, Probability density functions of ECS for the early and late Eocene compared to IPCC estimates. Shaded bands around lines throughout show 95% confidence intervals.

  5. Palaeogeography and δ18O-derived temperature against foraminiferal calcification depth.
    Extended Data Fig. 1: Palaeogeography and δ18O-derived temperature against foraminiferal calcification depth.

    a, Approximate palaeoposition of Tanzanian Drilling Project (TDP) sites studied here (map generated from www.odsn.de). b, Reconstructed temperature (TEocene) and relative depth of each foraminifera within each time slice. The pale blue line represents the output of a General Circulation Model simulation run with Eocene boundary conditions89, 90, whereas the other coloured lines show the General Circulation Model output offset to intersect with the warmest temperature at depth zero for each time slice. Note that depth assignments are approximate (see Methods). Also shown are the reconstructed temperatures from core-top (Holocene) planktonic foraminiferal δ18O (Supplementary Table 1), where G. ruber is assigned a depth of zero and the rest of the planktonic foraminifera are offset to reproduce the measured temperature profile of the GLOW15 cruise (light blue/green line)51, 66. For Holocene temperature reconstructions we used the modern site latitude of 9° S, and δ18Osw (using the Standard Mean Ocean Water (SMOW) standard) of 0‰.

  6. δ11B versus pH as a function of δ11Bsw.
    Extended Data Fig. 2: δ11B versus pH as a function of δ11Bsw.

    Increasing δ11Bsw, as indicated for each line in units of per mil, results in lower pH for the same δ11B. However, for the same δ11B range, the reconstructed pH range is larger for higher δ11Bsw (see brown-shaded regions).

  7. cGENIE estimates of calcite saturation in surface waters.
    Extended Data Fig. 3: cGENIE estimates of calcite saturation in surface waters.

    Comparison of calcite saturation Ωcalc for pre-industrial times (PI; blue) and 55 Myr ago (red) at different latitudes.

  8. Compilation of several CO2 records for the Eocene in comparison to this study.
    Extended Data Fig. 4: Compilation of several CO2 records for the Eocene in comparison to this study.

    Data are from refs 5, 11, 12, 18 and 19. The contribution of different parameters to the uncertainty on our CO2 reconstructions is colour-coded; sequentially from bottom to top, red is from δ11Bsw, blue is from the Ωcalc uncertainty, green is from the δ11Bc error, black is the 40 p.p.m. uncertainty in the event of disequilibrium with the atmosphere. Other parameters contribute <10% uncertainty to the CO2 calculations and are not shown. Note that the data from Pearson et al.12 (orange circles) are corrected as in Fig. 3, and there are two scenarios included for the Tanzania records of ref. 12 and this study: one with δ11Bc = δ11Bborate (closed blue and orange symbols) and the other applying T. sacculifer (open blue and orange symbols) corrections to the shallowest symbiotic planktonic foraminifera (Methods). The y axis is in log-scale. Error bars are representative of each proxy’s reconstruction uncertainty (typically at 95% confidence). For the δ11Β reconstructions in this study, the errors are based on Monte Carlo propagation of relevant errors (Methods).

  9. CO2 as a driver of latitudinal cooling in the Eocene, and ECS analyses of the EECO and late Eocene time slices.
    Extended Data Fig. 5: CO2 as a driver of latitudinal cooling in the Eocene, and ECS analyses of the EECO and late Eocene time slices.

    The case for the T. sacculifer calibration applied to shallowest planktonic foraminifera. a, Evolving relationship between SST14 for high (blue) and low (red) latitudes and the CO2 forcing of each of our time slices relative to the EECO. Linear regression fits and coefficients of determination (R2) are also shown, with the 95% confidence interval (shaded bands). b, Apparent latitudinal SST sensitivity for the last 520 kyr (ref. 13). The dashed line is a second-order polynomial through the SST sensitivity data (grey crosses) of ref. 13, and the grey lines show the 95% confidence interval. A red rectangle surrounds the SST sensitivity estimates averaged as a low-latitude mean, and the blue line indicates the high-latitude mean (see text and Methods). c, Reconstructed (lines) and estimated (symbols) SST relative to 53.2 Myr ago. Symbols show each of our time slices, calculated using the respective CO2 reconstructions and the average low- (red) and high- (blue) latitude SST sensitivities of b. Bold lines and shaded uncertainty band (at 95% confidence) show the reconstructed long-term mean SST estimates using the TEX86 proxy at high (blue colour >55°) and low (red colour <30°) latitudes14 relative to the SST ~53 Myr ago. Error bars represent full propagation of errors at 95% confidence14. d, Range in mean surface temperature change for early (green) and late (black) Eocene corrected for changes due to slow feedbacks2, 4, 21, 29. e, Forcing compared to the pre-industrial period, calculated using our CO2 reconstructions for the time slices 53.2 Myr ago (early Eocene) and 36.9 Myr ago (late Eocene) (see Methods). f, Probability density functions of ECS for the early (green) and late (black) Eocene compared to IPCC estimates (dashed lines show the 95% confidence interval (solid pink line)).

  10. The cGENIE estimates of air–seawater CO2 disequilibrium.
    Extended Data Fig. 6: The cGENIE estimates of air–seawater CO2 disequilibrium.

    The colour scale shows the difference between pCO2 in air and pCO2 in sea water. The model uses Eocene boundary conditions and positive values mean that seawater is a source of CO2 (in parts per million); the star shows the palaeo-location of Tanzania.

  11. Comparison of Eocene and modern planktonic foraminiferal δ11B and δ13C with δ18O.
    Extended Data Fig. 7: Comparison of Eocene and modern planktonic foraminiferal δ11B and δ13C with δ18O.

    a and b show analyses from the time slices 53.2 Myr ago and 40.3 Myr ago, respectively (as in Fig. 2 and Supplementary Table 1). Cibicidoides species (Cibs) are shaded in blue. c, Core-top (Holocene) offshore Tanzania foraminiferal measurements. Seawater δ11Bborate and δ18O was calculated from temperature, alkalinity and dissolved inorganic carbon measurements (from GLODAP cruises 18 and 23, stations 17742 and 23037, and 53.96° E to 7.04° S and 52.37° E to 6.33° S, respectively), correcting for anthropogenic carbon input. The black line in c represents seawater-derived δ18O and δ11Bborate data (Methods). The symbols for the time slices 53.2 and 40.3 Myr ago are as in Fig. 2. Note the change in scale for the x axis between the Eocene and Holocene panels. Errors in δ11B represent 2 s.d. of long-term precision (Methods).

  12. Reconstructed pH using different combinations of published symbiont-bearing and non-symbiotic foraminiferal vital-effect calibrations for the two most complete with depth time slices.
    Extended Data Fig. 8: Reconstructed pH using different combinations of published symbiont-bearing and non-symbiotic foraminiferal vital-effect calibrations for the two most complete with depth time slices.

    a shows the case where no vital effect corrections were applied for comparison. The vital-effect corrections for deeper asymbiotic planktonic foraminifera are based on either Globigerina bulloides26 (b and d) and Neogloboquadrina pachyderma65 (c and e) calibrations. For shallow symbiont-bearing planktonics we used calibrations specific to modern shallow, symbiont-bearing Trilobatus sacculifer22 (b and c) and Orbulina universa62 (d and e) (as recalculated by ref. 64 and offset by −3‰ to account for analytical differences between negative thermal ionization mass spectrometry and MC-ICPMS instrumentation8, 9). For comparison we also show the pH reconstructions for the case where we assume δ11Bc = δ11Bborate.

  13. The effect of seawater composition on boron isotope calibrations in foraminifera.
    Extended Data Fig. 9: The effect of seawater composition on boron isotope calibrations in foraminifera.

    The example of T. sacculifer calibration22 and the ‘borate’ calibration (assuming δ11Bc = δ11Bborate) for modern and Eocene seawater compositions; see Methods and Supplementary Table 2.

  14. The cGENIE output of seawater pH versus δ13C for the top ~300 m of the ocean.
    Extended Data Fig. 10: The cGENIE output of seawater pH versus δ13C for the top ~300 m of the ocean.

    The scenarios explored are from offshore Tanzania (triangles), Walvis Ridge in the South Atlantic Ocean (crosses), and the global ocean (circles) at three different atmospheric CO2 concentrations (modern, 3× pre-industrial and 16× pre-industrial, PI). An additional scenario showing 3× pre-industrial CO2 but considering the temperature effect on remineralization (indicated as ‘Tremin’) is also shown as blue squares and triangles.

Change history

Corrected online 19 May 2016
The present address for author E.H.J. was amended.

References

  1. Zachos, J. C., Dickens, G. R. & Zeebe, R. E. An early Cenozoic perspective on greenhouse warming and carbon-cycle dynamics. Nature 451, 279283 (2008)
  2. Caballero, R. & Huber, M. State-dependent climate sensitivity in past warm climates and its implications for future climate projections. Proc. Natl Acad. Sci. USA (2013)
  3. Lunt, D. J. et al. CO2-driven ocean circulation changes as an amplifier of Paleocene-Eocene thermal maximum hydrate destabilization. Geology 38, 875878 (2010)
  4. Loptson, C. A., Lunt, D. J. & Francis, J. E. Investigating vegetation-climate feedbacks during the early Eocene. Clim. Past 10, 419436 (2014)
  5. Beerling, D. J. & Royer, D. L. Convergent Cenozoic CO2 history. Nature Geosci. 4, 418420 (2011)
  6. Pearson, P. N. & Palmer, M. R. Atmospheric carbon dioxide concentrations over the past 60 million years. Nature 406, 695699 (2000)
  7. Pearson, P. N. & Palmer, M. R. Middle Eocene seawater pH and atmospheric carbon dioxide concentrations. Science 284, 18241826 (1999)
  8. Foster, G. L. et al. Interlaboratory comparison of boron isotope analyses of boric acid, seawater and marine CaCO3 by MC-ICPMS and NTIMS. Chem. Geol. 358, 114 (2013)
  9. Foster, G. L. Seawater pH, pCO2 and [CO32−] variations in the Caribbean Sea over the last 130 kyr: a boron isotope and B/Ca study of planktic foraminifera. Earth Planet. Sci. Lett. 271, 254266 (2008)
  10. Klochko, K. et al. Experimental measurement of boron isotope fractionation in seawater. Earth Planet. Sci. Lett. 248, 276285 (2006)
  11. Jagniecki, E. A., Lowenstein, T. K., Jenkins, D. M. & Demicco, R. V. Eocene atmospheric CO2 from the nahcolite proxy. Geology 43, 10751078 (2015)
  12. Pearson, P. N., Foster, G. L. & Wade, B. S. Atmospheric carbon dioxide through the Eocene–Oligocene climate transition. Nature 461, 11101113 (2009)
  13. Rohling, E. J. et al. Sea surface and high-latitude temperature sensitivity to radiative forcing of climate over several glacial cycles. J. Clim. 25, 16351656 (2012)
  14. Inglis, G. N. et al. Descent toward the Icehouse: Eocene sea surface cooling inferred from GDGT distributions. Paleoceanography 30, 10001020 (2015)
  15. Intergovernmental Panel on Climate Change (IPCC) Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) 11535 (2013)
  16. Crowell, J. C. & Frakes, L. A. Phanerozoic glaciation and the causes of ice ages. Am. J. Sci. 268, 193224 (1970)
  17. Raymo, M. E. & Ruddiman, W. F. Tectonic forcing of late Cenozoic climate. Nature 359, 117122 (1992)
  18. Pagani, M. et al. The role of carbon dioxide during the onset of Antarctic glaciation. Science 334, 12611264 (2011)
  19. Zhang, Y. G. et al. A 40-million-year history of atmospheric CO2. Phil. Trans. R. Soc. A 371, http://dx.doi.org/10.1098/rsta.2013.0096 (2013)
  20. Franks, P. J. et al. New constraints on atmospheric CO2 concentration for the Phanerozoic. Geophys. Res. Lett. 41, http://dx.doi.org/10.1002/2014gl060457 (2014)
  21. Lunt, D. J. et al. Warm climates of the past—a lesson for the future? Phil. Trans. R. Soc. A 371, http://dx.doi.org/10.1098/rsta.2013.0146 (2013)
  22. Foster, G. L., Lear, C. H. & Rae, J. W. B. The evolution of pCO2, ice volume and climate during the middle Miocene. Earth Planet. Sci. Lett. 341–344, 243254 (2012)
  23. Hönisch, B. et al. The geological record of ocean acidification. Science 335, 10581063 (2012)
  24. Demicco, R. V., Lowenstein, T. K. & Hardie, L. A. Atmospheric pCO2 since 60 Ma from records of seawater pH, calcium, and primary carbonate mineralogy. Geology 31, 793796 (2003)
  25. John, E. H. et al. Warm ocean processes and carbon cycling in the Eocene. Phil. Trans. R. Soc. A 371, http://dx.doi.org/10.1098/rsta.2013.0099 (2013)
  26. Martínez-Boti, M. A. et al. Plio-Pleistocene climate sensitivity evaluated using high-resolution CO2 records. Nature 518, 4954 (2015)
  27. Ridgwell, A. & Schmidt, D. N. Past constraints on the vulnerability of marine calcifiers to massive carbon dioxide release. Nature Geosci. 3, 196200 (2010)
  28. Byrne, B. & Goldblatt, C. Radiative forcing at high concentrations of well-mixed greenhouse gases. Geophys. Res. Lett. 41, 152160 (2014)
  29. PALAEOSENS Project Members. Making sense of palaeoclimate sensitivity. Nature 491, 683–691 (2012); erratum 494, 130 (2013)
  30. Bohaty, S. M., Zachos, J. C., Florindo, F. & Delaney, M. L. Coupled greenhouse warming and deep-sea acidification in the middle Eocene. Paleoceanography 24, PA2207 (2009)
  31. Nicholas, C. J. et al. Stratigraphy and sedimentology of the Upper Cretaceous to Paleogene Kilwa Group, southern coastal Tanzania. J. Afr. Earth Sci. 45, 431466 (2006)
  32. Pearson, P. N. et al. Further Paleogene and Cretaceous sediment cores from the Kilwa area of coastal Tanzania: Tanzania Drilling Project Sites 6-10. J. Afr. Earth Sci. 45, 279317 (2006)
  33. Kent, P. E., Hunt, J. A. & Johnstone, D. W. The Geology and Geophysics of Coastal Tanzania. Geophys. Paper No. 6, 1101 (Her Majesty’s Stationery Office, Inst. Geol. Sci, 1971)
  34. Bown, P. R. et al. A Paleogene calcareous microfossil Konservat-Lagerstätte from the Kilwa Group of coastal Tanzania. Geol. Soc. Am. Bull. 120, 312 (2008)
  35. Pearson, P. N. et al. Stable warm tropical climate through the Eocene Epoch. Geology 35, 211214 (2007)
  36. van Dongen, B. E. et al. Well preserved Palaeogene and Cretaceous biomarkers from the Kilwa area, Tanzania. Org. Geochem. 37, 539557 (2006)
  37. Takahashi, T. et al. Climatological mean and decadal change in surface ocean pCO2, and net sea-air CO2 flux over the global oceans. Deep Sea Res. II 56, 554577 (2009)
  38. Barker, S., Greaves, M. & Elderfield, H. A study of cleaning procedures used for foraminiferal Mg/Ca paleothermometry. Geochem. Geophys. Geosyst. 4, 8407 (2003)
  39. Yu, J., Elderfield, H., Greaves, M. & Day, J. Preferential dissolution of benthic foraminiferal calcite during laboratory reductive cleaning. Geochem. Geophys. Geosys. 8, Q06016 (2007)
  40. Rae, J. W. B., Foster, G. L., Schmidt, D. N. & Elliott, T. Boron isotopes and B/Ca in benthic foraminifera: proxies for the deep ocean carbonate system. Earth Planet. Sci. Lett. 302, 403413 (2011)
  41. Edgar, K. M., Anagnostou, E., Pearson, P. N. & Foster, G. L. Assessing the impact of diagenesis on δ11B, δ13C, δ18O, Sr/Ca and B/Ca values in fossil planktic foraminiferal calcite. Geochim. Cosmochim. Acta 166, 189209 (2015)
  42. Kim, S.-T. & O’Neil, J. R. Equilibrium and non-equilibrium oxygen isotope effects in synthetic carbonates. Geochim. Cosmochim. Acta 61, 34613475 (1997)
  43. Cramer, B. S., Miller, K. G., Barrett, P. J. & Wright, J. D. Late Cretaceous–Neogene trends in deep ocean temperature and continental ice volume: reconciling records of benthic foraminiferal geochemistry (δ18O and Mg/Ca) with sea level history. J. Geophys. Res. 116, C12023 (2011)
  44. Zachos, J. C., Stott, L. D. & Lohmann, K. C. Evolution of early Cenozoic marine temperatures. Paleoceanography 9, 353387 (1994)
  45. Pearson, P. N. in Reconstructing Earth's Deep Time Climate—The State of the Art in 2012 Vol. 18 (eds Ivany, L. C. & Huber, B. T.) 138 (The Paleontological Society, 2012)
  46. LeGrande, A. N. & Schmidt, G. A. Global gridded data set of the oxygen isotopic composition in seawater. Geophys. Res. Lett. 33, L12604 (2006)
  47. Sexton, P. F., Wilson, P. A. & Pearson, P. N. Palaeoecology of late middle Eocene planktic foraminifera and evolutionary implications. Mar. Micropaleontol. 60, 116 (2006)
  48. Pearson, P. N., Shackleton, N. J. & Hall, M. A. Stable isotope paleoecology of middle Eocene planktonic foraminifera and multi-species isotope stratigraphy, DSDP Site 523, South Atlantic. J. Foraminiferal Res. 23, 123140 (1993)
  49. Pearson, P. N. et al. Warm tropical sea surface temperatures in the Late Cretaceous and Eocene epochs. Nature 413, 481487 (2001)
  50. Edgar, K. M. et al. Symbiont ‘bleaching’ in planktic foraminifera during the Middle Eocene Climatic Optimum. Geology 41, 1518 (2013)
  51. Birch, H. et al. Planktonic foraminifera stable isotopes and water column structure: disentangling ecological signals. Mar. Micropaleontol. 101, 127145 (2013)
  52. Fairbanks, R. G. & Wiere, P. H. & Bé, A. W. H. Vertical distribution and isotopic composition of living planktonic foraminifera in the western North Atlantic. Science 207, 6163 (1980)
  53. Spero, H. & Williams, D. F. Extracting environmental information from planktonic foraminiferal δ13C data. Nature 335, 717719 (1988)
  54. Berger, W. H., Killingley, J. S. & Vincent, E. Stable isotopes in deep-sea carbonates: Box Core ERDC-92, West Equatorial Pacific. Oceanol. Acta 1, 203216 (1978)
  55. Seki, O. et al. Alkenone and boron-based Pliocene pCO2 records. Earth Planet. Sci. Lett. 292, 201211 (2010)
  56. Hönisch, B. et al. Atmospheric carbon dioxide concentration across the Mid-Pleistocene transition. Science 324, 15511554 (2009)
  57. Sanyal, A. & Bijma, J. A comparative study of the northwest Africa and eastern equatorial Pacific upwelling zones as sources of CO2 during glacial periods based on boron isotope paleo-pH estimation. Paleoceanography 14, 753759 (1999)
  58. Lemarchand, D., Gaillardet, J., Lewin, E. & Allegre, C. J. The influence of rivers on marine boron isotopes and implications for reconstructing past ocean pH. Nature 408, 951954 (2000)
  59. Foster, G. L., Pogge von Strandmann, P. A. E. & Rae, J. W. B. Boron and magnesium isotopic composition of seawater. Geochem. Geophys. Geosyst. 11, Q08015 (2010)
  60. Rink, S., Kühl, M., Bijma, J. & Spero, H. J. Microsensor studies of photosynthesis and respiration in the symbiotic foraminifer Orbulina universa. Mar. Biol. 131, 583595 (1998)
  61. Zeebe, R. E., Wolf-Gladrow, D. A., Bijma, J. & Hönisch, B. Vital effects in foraminifera do not compromise the use of δ11B as a paleo-pH indicator: evidence from modelling. Paleoceanography 18, 1043 (2003)
  62. Sanyal, A. et al. Oceanic pH control on the boron isotopic composition of foraminifera: evidence from culture experiments. Paleoceanography 11, 513517 (1996)
  63. Sanyal, A., Bijma, J., Spero, H. & Lea, D. W. Empirical relationship between pH and the boron isotopic composition of Globigerinoides sacculifer: implications for the boron isotope paleo-pH proxy. Paleoceanography 16, 515519 (2001)
  64. Henehan, M. J. et al. Calibration of the boron isotope proxy in the planktonic foraminifera Globigerinoides ruber for use in palaeo-CO2 reconstruction. Earth Planet. Sci. Lett. 364, 111122 (2013)
  65. Yu, J., Thornalley, D. J. R., Rae, J. W. B. & McCave, N. I. Calibration and application of B/Ca, Cd/Ca, and δ11B in Neogloboquadrina pachyderma (sinistral) to constrain CO2 uptake in the subpolar North Atlantic during the last deglaciation. Paleoceanography 28, 237252 (2013)
  66. Kroon, D. & the Shipboard Scientific Party. Tropical Temperature History during Paleogene Global Warming (GLOW) Events. Netherlands Institute for Sea Research (NIOZ) Site Survey Cruise Report (RV Pelagia cruise number 64PE303) 1151 (NIOZ, 2010)
  67. McCorkle, D., Corliss, B. & Farnham, C. Vertical distributions and stable isotopic compositions of live (stained) benthic foraminifera from the North Carolina and California continental margins. Deep Sea Res. I 44, 9831024 (1997)
  68. Penman, D. E. et al. Rapid and sustained surface ocean acidification during the Paleocene-Eocene Thermal Maximum. Paleoceanography 29, 2014PA002621 (2014)
  69. Hönisch, B. & Hemming, N. G. Ground-truthing the boron isotope-paleo-pH proxy in planktonic foraminifera shells: partial dissolution and shell size effects. Paleoceanography 19, PA4010 (2004)
  70. Martínez-Boti, M. A. et al. Boron isotope evidence for oceanic carbon dioxide leakage during the last deglaciation. Nature 518, 219222 (2015)
  71. Hönisch, B. et al. The influence of symbiont photosynthesis on the boron isotopic composition of foraminifera shells. Mar. Micropaleontol. 49, 8796 (2003)
  72. Raitzsch, M. & Hönisch, B. Cenozoic boron isotope variations in benthic foraminifers. Geology 41, 591594 (2013)
  73. John, E. H., Wilson, D. J., Ridgwell, A. & Pearson, P. N. Temperature-dependent remineralization and carbon cycling in the warm Eocene oceans. Palaeogeogr. Palaeoclimatol. Palaeoecol. 413, 158166 (2014)
  74. Goyet, C., Healy, R. J. & Ryan, J. P. Global distribution of total inorganic carbon and total alkalinity below the deepest winter mixed layer depths. Report ORNL/CDIAC-127, NDP-076 (Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, US Department of Energy, 2000)
  75. Key, R. M. et al. A global ocean carbon climatology: results from the Global Data Analysis Project (GLODAP). Glob. Biogeochem. Cycles 18, GB4031 (2004)
  76. Battino, R., Rettich, T. R. & Tominaga, T. The solubility of oxygen and ozone in liquids. J. Phys. Chem. Ref. Data 12, 163178 (1983)
  77. Jenkins, W. J. & Goldman, J. C. Seasonal oxygen cycling and primary production in the Sargasso Sea. J. Mar. Res. 43, 465491 (1985)
  78. Sarmiento, J. L. & Gruber, N. Ocean Biogeochemical Dynamics 1528 (Princeton Univ. Press, 2006)
  79. Hull, P. M., Osborn, K. J., Norris, R. D. & Robinson, B. H. Seasonality and depth distribution of a mesopelagic foraminifer, Hastigerinella digitata, in Monterey Bay, California. Limnol. Oceanogr. 56, 562576 (2011)
  80. Pälike, H. et al. A Cenozoic record of the equatorial Pacific carbonate compensation depth. Nature 488, 609614 (2012)
  81. Ridgwell, A. & Zeebe, R. E. The role of the global carbonate cycle in the regulation and evolution of the Earth system. Earth Planet. Sci. Lett. 234, 299315 (2005)
  82. Tyrrell, T. & Zeebe, R. E. History of carbonate ion concentration over the last 100 million years. Geochim. Cosmochim. Acta 68, 35213530 (2004)
  83. Coggon, R. M. et al. Reconstructing past seawater Mg/Ca and Sr/Ca from mid-ocean ridge flank calcium carbonate veins. Science 327, 11141117 (2010)
  84. Horita, J., Zimmermann, H. & Holland, H. D. Chemical evolution of seawater during the Phanerozoic: implications for the record of marine evaporites. Geochim. Cosmochim. Acta 66, 37333756 (2002)
  85. Dickson, A. J. Fossil echinoderms as monitor of the Mg/Ca ratio of Phanerozoic oceans. Science 298, 12221224 (2002)
  86. Lowenstein, T. K. et al. Oscillations in the Phanerozoic seawater chemistry: evidence from fluid inclusions. Science 294, 10861088 (2001)
  87. Lear, C. H., Elderfield, H. & Wilson, P. A. Cenozoic deep-sea temperatures and global ice volumes from Mg/Ca in benthic foraminiferal calcite. Science 287, 269272 (2000)
  88. Wade, B. S. & Pearson, P. N. Planktonic foraminiferal turnover, diversity fluctuations and geochemical signals across the Eocene/Oligocene boundary in Tanzania. Mar. Micropaleontol. 68, 244255 (2008)
  89. Huber, M. & Caballero, R. The early Eocene equable climate problem revisited. Clim. Past 7, 603633 (2011)
  90. Tindall, J. et al. Modelling the oxygen isotope distribution of ancient seawater using a coupled ocean-atmosphere GCM: implications for reconstructing early Eocene climate. Earth Planet. Sci. Lett. 292, 265273 (2010)
  91. Wade, B. S., Pearson, P. N., Berggren, W. A. & Pälike, H. Review and revision of Cenozoic tropical planktonic foraminiferal biostratigraphy and calibration to the geomagnetic polarity and astronomical time scale. Earth Sci. Rev. 104, 111142 (2011)
  92. Ben-Yaakov, S. & Goldhaber, M. B. The influence of sea water composition on the apparent constants of the carbonate system. Deep-Sea Res. 20, 8799 (1973)
  93. Millero, F. J. & Pierrot, D. A chemical equilibrium model for natural waters. Aquat. Geochem. 4, 153199 (1998)
  94. Hershey, J. P., Fernandez, M., Milne, P. J. & Millero, F. J. The ionization of boric acid in NaCl, Na-Ca-Cl and Na-Mg-Cl solutions at 25°C. Geochim. Cosmochim. Acta 50, 143148 (1986)

Download references

Author information

  1. Present addresses: School of Geography, Earth Science and Environment, University of the South Pacific, Suva, Fiji (E.H.J.); School of Geography, Earth and Environmental Sciences, University of Birmingham, Birmingham B15 2TT, UK (K.M.E.).

    • Eleanor H. John &
    • Kirsty M. Edgar

Affiliations

  1. Ocean and Earth Science, National Oceanography Centre Southampton, University of Southampton Waterfront Campus, Southampton SO14 3ZH, UK

    • Eleni Anagnostou &
    • Gavin L. Foster
  2. School of Earth and Ocean Sciences, Cardiff University, Park Place, Cardiff CF10 3AT, UK

    • Eleanor H. John,
    • Kirsty M. Edgar &
    • Paul N. Pearson
  3. School of Earth Sciences, Bristol University, Bristol BS8 1RJ, UK

    • Kirsty M. Edgar
  4. School of Geographical Sciences, Bristol University, Bristol BS8 1SS, UK

    • Andy Ridgwell &
    • Daniel J. Lunt
  5. Department of Earth Sciences, University of California, Riverside, California 92521, USA

    • Andy Ridgwell &
    • Daniel J. Lunt
  6. Organic Geochemistry Unit, School of Chemistry, University of Bristol, Cantock’s Close, Bristol BS8 1TS, UK

    • Gordon N. Inglis &
    • Richard D. Pancost
  7. Cabot Institute, University of Bristol, Bristol BS8 1UJ, UK

    • Gordon N. Inglis &
    • Richard D. Pancost

Contributions

E.A. conducted all boron isotope and trace element analyses, performed calculations, and drafted the manuscript. E.H.J. and K.M.E. prepared foraminifer samples and conducted the stable isotope analysis. P.N.P. led the fieldwork, performed the taxonomy and prepared foraminifer samples. A.R. provided cGENIE model results. P.N.P. and G.L.F. designed the study and all authors discussed the results and contributed to the final text.

Competing financial interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to:

Author details

Extended data figures and tables

Extended Data Figures

  1. Extended Data Figure 1: Palaeogeography and δ18O-derived temperature against foraminiferal calcification depth. (147 KB)

    a, Approximate palaeoposition of Tanzanian Drilling Project (TDP) sites studied here (map generated from www.odsn.de). b, Reconstructed temperature (TEocene) and relative depth of each foraminifera within each time slice. The pale blue line represents the output of a General Circulation Model simulation run with Eocene boundary conditions89, 90, whereas the other coloured lines show the General Circulation Model output offset to intersect with the warmest temperature at depth zero for each time slice. Note that depth assignments are approximate (see Methods). Also shown are the reconstructed temperatures from core-top (Holocene) planktonic foraminiferal δ18O (Supplementary Table 1), where G. ruber is assigned a depth of zero and the rest of the planktonic foraminifera are offset to reproduce the measured temperature profile of the GLOW15 cruise (light blue/green line)51, 66. For Holocene temperature reconstructions we used the modern site latitude of 9° S, and δ18Osw (using the Standard Mean Ocean Water (SMOW) standard) of 0‰.

  2. Extended Data Figure 2: δ11B versus pH as a function of δ11Bsw. (129 KB)

    Increasing δ11Bsw, as indicated for each line in units of per mil, results in lower pH for the same δ11B. However, for the same δ11B range, the reconstructed pH range is larger for higher δ11Bsw (see brown-shaded regions).

  3. Extended Data Figure 3: cGENIE estimates of calcite saturation in surface waters. (129 KB)

    Comparison of calcite saturation Ωcalc for pre-industrial times (PI; blue) and 55 Myr ago (red) at different latitudes.

  4. Extended Data Figure 4: Compilation of several CO2 records for the Eocene in comparison to this study. (206 KB)

    Data are from refs 5, 11, 12, 18 and 19. The contribution of different parameters to the uncertainty on our CO2 reconstructions is colour-coded; sequentially from bottom to top, red is from δ11Bsw, blue is from the Ωcalc uncertainty, green is from the δ11Bc error, black is the 40 p.p.m. uncertainty in the event of disequilibrium with the atmosphere. Other parameters contribute <10% uncertainty to the CO2 calculations and are not shown. Note that the data from Pearson et al.12 (orange circles) are corrected as in Fig. 3, and there are two scenarios included for the Tanzania records of ref. 12 and this study: one with δ11Bc = δ11Bborate (closed blue and orange symbols) and the other applying T. sacculifer (open blue and orange symbols) corrections to the shallowest symbiotic planktonic foraminifera (Methods). The y axis is in log-scale. Error bars are representative of each proxy’s reconstruction uncertainty (typically at 95% confidence). For the δ11Β reconstructions in this study, the errors are based on Monte Carlo propagation of relevant errors (Methods).

  5. Extended Data Figure 5: CO2 as a driver of latitudinal cooling in the Eocene, and ECS analyses of the EECO and late Eocene time slices. (264 KB)

    The case for the T. sacculifer calibration applied to shallowest planktonic foraminifera. a, Evolving relationship between SST14 for high (blue) and low (red) latitudes and the CO2 forcing of each of our time slices relative to the EECO. Linear regression fits and coefficients of determination (R2) are also shown, with the 95% confidence interval (shaded bands). b, Apparent latitudinal SST sensitivity for the last 520 kyr (ref. 13). The dashed line is a second-order polynomial through the SST sensitivity data (grey crosses) of ref. 13, and the grey lines show the 95% confidence interval. A red rectangle surrounds the SST sensitivity estimates averaged as a low-latitude mean, and the blue line indicates the high-latitude mean (see text and Methods). c, Reconstructed (lines) and estimated (symbols) SST relative to 53.2 Myr ago. Symbols show each of our time slices, calculated using the respective CO2 reconstructions and the average low- (red) and high- (blue) latitude SST sensitivities of b. Bold lines and shaded uncertainty band (at 95% confidence) show the reconstructed long-term mean SST estimates using the TEX86 proxy at high (blue colour >55°) and low (red colour <30°) latitudes14 relative to the SST ~53 Myr ago. Error bars represent full propagation of errors at 95% confidence14. d, Range in mean surface temperature change for early (green) and late (black) Eocene corrected for changes due to slow feedbacks2, 4, 21, 29. e, Forcing compared to the pre-industrial period, calculated using our CO2 reconstructions for the time slices 53.2 Myr ago (early Eocene) and 36.9 Myr ago (late Eocene) (see Methods). f, Probability density functions of ECS for the early (green) and late (black) Eocene compared to IPCC estimates (dashed lines show the 95% confidence interval (solid pink line)).

  6. Extended Data Figure 6: The cGENIE estimates of air–seawater CO2 disequilibrium. (270 KB)

    The colour scale shows the difference between pCO2 in air and pCO2 in sea water. The model uses Eocene boundary conditions and positive values mean that seawater is a source of CO2 (in parts per million); the star shows the palaeo-location of Tanzania.

  7. Extended Data Figure 7: Comparison of Eocene and modern planktonic foraminiferal δ11B and δ13C with δ18O. (166 KB)

    a and b show analyses from the time slices 53.2 Myr ago and 40.3 Myr ago, respectively (as in Fig. 2 and Supplementary Table 1). Cibicidoides species (Cibs) are shaded in blue. c, Core-top (Holocene) offshore Tanzania foraminiferal measurements. Seawater δ11Bborate and δ18O was calculated from temperature, alkalinity and dissolved inorganic carbon measurements (from GLODAP cruises 18 and 23, stations 17742 and 23037, and 53.96° E to 7.04° S and 52.37° E to 6.33° S, respectively), correcting for anthropogenic carbon input. The black line in c represents seawater-derived δ18O and δ11Bborate data (Methods). The symbols for the time slices 53.2 and 40.3 Myr ago are as in Fig. 2. Note the change in scale for the x axis between the Eocene and Holocene panels. Errors in δ11B represent 2 s.d. of long-term precision (Methods).

  8. Extended Data Figure 8: Reconstructed pH using different combinations of published symbiont-bearing and non-symbiotic foraminiferal vital-effect calibrations for the two most complete with depth time slices. (118 KB)

    a shows the case where no vital effect corrections were applied for comparison. The vital-effect corrections for deeper asymbiotic planktonic foraminifera are based on either Globigerina bulloides26 (b and d) and Neogloboquadrina pachyderma65 (c and e) calibrations. For shallow symbiont-bearing planktonics we used calibrations specific to modern shallow, symbiont-bearing Trilobatus sacculifer22 (b and c) and Orbulina universa62 (d and e) (as recalculated by ref. 64 and offset by −3‰ to account for analytical differences between negative thermal ionization mass spectrometry and MC-ICPMS instrumentation8, 9). For comparison we also show the pH reconstructions for the case where we assume δ11Bc = δ11Bborate.

  9. Extended Data Figure 9: The effect of seawater composition on boron isotope calibrations in foraminifera. (117 KB)

    The example of T. sacculifer calibration22 and the ‘borate’ calibration (assuming δ11Bc = δ11Bborate) for modern and Eocene seawater compositions; see Methods and Supplementary Table 2.

  10. Extended Data Figure 10: The cGENIE output of seawater pH versus δ13C for the top ~300 m of the ocean. (206 KB)

    The scenarios explored are from offshore Tanzania (triangles), Walvis Ridge in the South Atlantic Ocean (crosses), and the global ocean (circles) at three different atmospheric CO2 concentrations (modern, 3× pre-industrial and 16× pre-industrial, PI). An additional scenario showing 3× pre-industrial CO2 but considering the temperature effect on remineralization (indicated as ‘Tremin’) is also shown as blue squares and triangles.

Supplementary information

Excel files

  1. Supplementary Tables (33 KB)

    This file contains Supplementary Tables 1-2.

Additional data