Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Orbitally paced climate oscillations across the Oligocene/Miocene boundary


The late Oligocene and early Miocene periods, some 21 to 27 million years ago, have generally been viewed as times of moderate global warmth and ice-free conditions. Yet several lines of evidence suggest that this interval was punctuated by at least one, and possibly several, episodes of high-latitude cooling and continental glaciation1,2,3. Here, we present stable-isotope and per cent coarse-fraction data from an equatorial, western Atlantic deep-sea-sediment core that provide high-resolution records of the climate variability across the Oligocene/Miocene transition (22.5–25.7 million years ago). A strong 40-kyr periodicity in the oxygen isotope record is consistent with a high-latitude orbital (obliquity) control on ice-volume and temperature. Orbital influences are also apparent at precession and eccentricity frequencies, including a series of 400-kyr oscillations that culminate in distinct maxima at the Oligocene/Miocene boundary, about 23.7 million years ago. Covariance between the carbon and oxygen isotope records suggests that the oceanic carbon cycle may have contributed to global cooling during the 400-kyr cycles, particularly at the Oligocene/Miocene boundary.


The Oligocene/Miocene boundary represents a significant climate transition in Cenozoic Earth history; ocean temperatures were slowly increasing and continental ice-volume was apparently decreasing from the large accumulations of the middle–late Oligocene2,3,4. Oxygen isotope records indicate that this trend may have been interrupted by a series of brief glaciations that began in the earliest Miocene and continued up to the time of significant middle Miocene cooling2. The first and largest of these events, Mi-1, occurred at the Oligocene/Miocene boundary.

As with most other pre-Pliocene intervals, previous isotope records that span the Oligocene/Miocene boundary lack the sampling density needed to resolve palaeoenvironmental changes occurring on timescales of <105 years. Attempts to construct higher resolution records have been hampered either by the lack of continuous, undisturbed sedimentary sections, or by diagenetic overprinting. These deficiencies were recently resolved with the successful recovery of an expanded and hiatus-free section of upper Oligocene and lower Miocene nannofossil chalk in Hole 929A, on Ceara rise in the western equatorial Atlantic Ocean (ref. 5) (4360 m water depth). This section is characterized by prominent decimetre-scale oscillations in sediment colour and magnetic susceptibility with periods similar to those associated with orbital forcing6.

Using samples from Hole 929A, we assembled high-resolution (10-cm sampling) isotope and per cent coarse-fraction time series for the Oligocene/Miocene boundary (22.5–25.7 Myr ago). Initially, samples were collected from four cores (929A 34X–37X) spanning a 40-m interval across the boundary7. We subsequently increased the length of the record to 60 m by adding samples from two additional cores (929A 33X-38X) to create time series nearly 3.2 Myr long across the Oligocene/Miocene boundary. Core recovery was 100% over this interval. Site-to-site comparisons of magnetic susceptibility and colour reflectance data suggest inter-core gaps of <40 cm (20 kyr)5. The stable-isotope record, plotted in Fig. 1 along with per cent coarse fraction (% CF), is based on the carbon and oxygen isotope ratios of the benthic foraminifer, Cibicidoides mundulus. The age model is based on sedimentation rates derived from the Hole 929A biostratigraphy5 and the wavelength of the 40-kyr cycle in the magnetic susceptibility record6 (Table 1). Ages are cumulative, starting from an assigned age of 22.5 Myr for the uppermost level (core 929A-33X, 10 cm; 301.10 m below sea floor) of the sampled section. This age is based on an interpolation of two biostratigraphic age datums calibrated against the most recent Geomagnetic Polarity Time Scale (GPTS)8.

Figure 1: a, Benthic foraminifer stable-isotope and % coarse fraction (%CF) records for Hole 929A plotted versus age.

Each sample was freeze-dried, weighed, washed and wet sieved over a 63-µm screen. The >63 µm fraction was dried and weighed to calculate %CF. From five to ten specimens of benthic foraminifer taxa, Cibidoides mundulus, were then picked from the >150-µm fraction of each sample and analysed for carbon and oxygen isotopic composition. Stable-isotope analyses were carried out in the Stable Isotope Laboratory at the University of California, Santa Cruz using a common acid-bath device coupled to either a Fisons PRISM or an OPTIMA mass spectrometer12. All values are reported in the per mil (‰) notation relative to Pee Dee Belemnite standard. External precision based on duplicates and triplicates is better than ±0.09 for both and C and O isotopes. The %CF represents the mass of the >63-µm size fraction relative to the total mass of each sample. The heavy lines in the stable-isotope plots represent the curve fits (locally weighted (15%) means) used for detrending. The age model is based on sedimentation rates inferred from the wavelength of the 40-kyr cycle in magnetic susceptibility6. This age model proved to be superior to one based exclusively on biostratigraphic datums which, because of dissolution and low abundances of planktonic microfossils at this site, tended to be less reliable. Nevertheless, the differences between the two age models are relatively small (see Supplementary Information). b, Expanded view of the oxygen-isotope (solid line; left-hand ordinate) and %CF (dashed line;right-hand ordinate) records for the period 23.4–24.6 Myr ago. The arrows denote individual “glacial maxima” associated with the Mi-1 event.

Table 1 Age model parameters

The oxygen-isotope time series shows changes on several different scales, including pervasive high- and low-frequency (104–105 yr) oscillations and a brief positive δ18O excursion of 1.2‰ just above the Oligocene/Miocene boundary (Fig. 1a). The high-frequency oscillations have amplitudes of between 0.5 and 0.6‰; the lower-frequency oscillations have slightly larger amplitudes. The 1.2‰ excursion initiated 24.0 Myr ago and peaked at 23.7 Myr. The magnitude and timing suggest it is equivalent to the Mi-1 event. In detail, it consisted of a series of higher-frequency cycles of varying amplitude superimposed on a longer-term increase (Fig. 1b). The carbon isotope record also shows variability on several scales, including high- and low-frequency oscillations, and a long-term rise. The high-frequency cycles are of small amplitude, generally <0.4‰. The lower-frequency cycles are of significantly larger amplitude, in excess of 0.8‰, and appear to covary with oxygen isotopes over most of the record. The long term rise began 25.0 Myr agoand peaked at 23.65 Myr with mean δ13C values increasing by >0.7‰.

The coarse fraction (>63-µm) of these sediments consists largely of planktonic foraminifer tests. The %CF record is dominated by high-frequency variability (104 yr) with values varying between 0.1 and 12%. A low-frequency pattern is present as well; there are three intervals, each 100–200 kyr long, with low mean concentrations and values of %CF rarely exceeding 1%. These three intervals are centred at roughly 22.8, 24.1 and 25.1 Myr ago.

To identify the main periods of variance in the isotope and %CF records, as well as internal phase relationships (coherency) between variables, spectral and cross-spectral analyses were carried out using a program based on the Blackman–Tukey method9 (for details see legends of Figs 2 and 3). High concentrations of variance in the δ18O spectrum were identified at frequency components that correspond to the Milankovitch periods of 19, 23, 40, 100 and 413 kyr (0.284, 0.235, 0.130, 0.054 and 0.013 cycles per 5.4 kyr; Fig. 2). The δ13C spectra shows concentration of variance at the 19-, 41-, 100- and 413-kyr periods, whereas the %CF spectra is dominated by the 40-kyr period.

Figure 2: Power spectra (variance versus frequency) of three variables, the benthic δ18O and δ13C, and %CF values.

The left-hand ordinate shows the spectral power of each record. Bandwidth (0.0133) and confidence intervals (80%) are shown. The time-series analysis was carried out using a numerical approach based on the Blackman–Tukey method (SPECTX; Oregon State Univ.). Following SPECMAP convention, the oxygen isotope record was multiplied by −1 so that interglacials (less ice and warmer temperatures) are represented by higher values. After converting to age, the data set was resampled at an equal time interval of 5.4 kyr (n = 578). The number of lags (m) was set to 100. To resolve the lower-frequency components of the oxygen and carbon isotope time series, locally weighted means (Fig. 1a) were fitted to the records and subtracted to remove the longer-term trends from each time series. Residual long-term trends were then removed using a linear detrend. This pre-processing substantially improved the degree to which variance at the low-frequency (400 and 100 kyr) end of the spectrum could be resolved. The vertical bars show the frequencies of the primary orbital periods.

Figure 3: Top panels, coherence spectra of the δ13C (panel a) and %CF (panel b) time series relative to δ18O (solid lines).

Following SPECMAP convention, all δ18O values were multiplied by −1 so that phase relationships are relative to glacial minima rather than maxima. The isotope time series were detrended by subtracting a locally weighted mean, and then resampled at a constant 5.4-kyr time interval. Variance density and coherency (solid line with plus signs) were calculated based on 100 lags of the autocovariance function. The coherencies are plotted on a hyperbolic arctangent scale along the y-axis so that the confidence intervals for the coherency estimates are a constant length. The horizontal line is the 80% confidence interval testing the non-zero hypothesis for the estimates. Bottom panels, the phase angles (in degrees) and associated 80% confidence intervals (vertical lines).

At the 400- and 100-kyr periods, −δ18O (δ18O is multiplied by −1 following SPECMAP convention) and δ13C show a high-degree coherency (k = 0.91 and 0.89, respectively), with −δ18O (glacial minima) nearly 180° out of phase with δ13C (Fig. 3a). These variables are also coherent (k = 0.72) and nearly in phase at the 40-kyr period. A high degree of coherency (k = 0.85) also occurs between %CF and δ18O at the 40-kyr period with −δ18O lagging %CF by 40° (4.4 ± 1.7 kyr; Fig. 3b).

The strong response in the oxygen isotope record at the obliquity (40-kyr) period is significant. Because the effects of obliquity on insolation are strongest at high latitudes10, oscillations with a 40-kyr period suggest a high-latitude climate control (ice-volume and/or temperature) on δ18O across the Oligocene/Miocene boundary. This finding, along with evidence of similar periodicities in earliest Oligocene and early middle Miocene oxygen isotope records11,12, as well as in Plio-Pleistocene records, suggests that a strong response to obliquity forcing has been a common feature of the global climate signal for at least the past 34 Myr.

These obliquity-paced climate oscillations were closely linked to deep-water circulation patterns and chemistry. At the 40-kyr period, %CF appears to be tightly coupled to δ18O over most of this record (Fig. 3a). Three factors can influence %CF: surface production, dilution by terrigeneous debris, and dissolution13. At Hole 929A, a general association of low %CF with increased fragmentation and low carbonate concentration (as inferred from magnetic susceptibility and colour reflectance5) indicates that dissolution was the controlling factor. This coupling of δ18O and carbonate dissolution is similar to patterns observed in Quaternary sediments14,15 and can be explained by increased flux of more corrosive bottom water into the equatorial Atlantic during glacial periods. The presence of a south-to-north decrease in deep-water δ13C during the early Miocene16 suggests a Southern Ocean source for this more corrosive water.

The Hole 929 δ18O record resolves two critical features of the Mi-1 glacial maximum. The first is the structure of the Mi-1 excursion; it actually consisted of a series of higher-frequency (40-, 100-kyr) oscillations superimposed on a long-term increase (Fig. 1b). The direct response at the 100-kyr cycle appears to be slightly stronger just after the Mi-1 event, although further work is needed to resolve the detailed evolution of the climate spectra over this and other intervals. The second important feature is the magnitude of Mi-1; a 1.2‰ increase requires an accumulation of continental ice equivalent in volume to the present-day East Antarctic Ice Sheet, or 5–6 °C cooling of bottom waters, or, more likely, some combination of ice-volume/temperature change. The presence of Upper Oligocene and Lower Miocene glacial marine sediments in the Ross Sea1,17, as well as new evidence that links a prominent eustatic lowering18 to Mi-119 suggests that part of the δ18O increase was due to ice-sheet expansion.

Perhaps the most intriguing features of the isotope spectra are the high-amplitude 400-kyr oscillations, particularly in δ13C. Variance in climate near this period, as well as at the 100-kyr period, is thought to originate from an asymmetrical response mechanism that preferentially introduces variance from the warmer portions of the eccentricity modulated precession cycle20. Although rarely recorded in Pliocene/Pleistocene20,21, 400-kyr oscillations commonly arise in lower-resolution Oligocene and Miocene benthic foraminiferal δ13C time series12,22,23,24. Such a strong expression of this period in the carbon isotope record indicates a nonlinear relation between the carbon cycle and eccentricity forcing. At Hole 929A, δ18O and δ13C are closely coupled at the 400-kyr period (lag 24 ± 20 kyr) as well as at 100-kyr period, with δ18O leading δ13C. Covariance between oxygen and carbon in early Oligocene and middle Miocene records has previously been attributed to the effects of climate change on ocean/atmosphere circulation, ocean productivity, and organic carbon23,24,25. Investigators speculate that as high-latitude temperatures decrease and ice sheets grow, oceanic and atmospheric circulation intensify, increasing oceanic turnover, productivity and organic-carbon fluxes. In turn, burial rates of organic carbon increase thereby driving mean ocean δ13C towards higher values26. The process apparently reverses as climate warms and/or nutrient levels are reduced. This coupling may have helped amplify the response at the lower frequencies.

Close coupling of climate (as expressed by δ18O) and the oceanic carbon cycle may also explain the unusual scale of the Mi-1 event. If the low-frequency carbon-isotope cycles reflect periodic changes in the rate of organic-carbon burial as postulated above, it is likely that the atmospheric partial pressure of CO2 ( p CO 2 ) was oscillating as well, and generating a positive feedback. We speculate that the variations in p CO 2 during the Oligocene were within a range where the feedback on the global radiative balance was comparatively small until the earliest Miocene (Mi-1) when p CO 2 had dropped below some threshold. The process of reaching this threshold was probably gradual, as suggested by the progressive step by step increase in carbon isotope ratios before Mi-1, a trend which indicates significant oceanic carbon-cycle change during the latest Oligocene.

The Hole 929A isotope times series presented here provide critical new insight into the character of climate variability across the Oligocene/Miocene transition. The persistence of a strong 40-kyr period in the δ18O record is consistent with the presence of Antarctic ice sheets during this period of moderate global warmth. Documentation of 100- and 400-kyr periods provides strong support for a palaeoceanograhic response to the longer-period Milankovitch cycles. Covariance between δ13C and δ18O at these eccentricity periods hints at a strong coupling between climate and the ocean carbon cycle. Tuning of the Hole 929 isotope timeseries to orbital spectra (N. J. Shackleton, personal communication) will ultimately produce a standard reference section that will provide a degree of chronostratigraphic control for the lateOligocene to early Miocene that approaches that of the Quaternary.


  1. 1

    Leckie, R. M. & Webb, P. N. Late Oligocene-early Miocene glacial record of the Ross Sea, Antarctica: evidence from DSDP Site 270. Geology 11, 578–582 (1983).

    ADS  Article  Google Scholar 

  2. 2

    Miller, K. G., Wright, J. D. & Fairbanks, R. G. Unlocking the ice house: Oligocene-Miocene oxygen isotopes, eustacy and margin erosion. J. Geophys. Res. 96, 6829–6848 (1991).

    ADS  Article  Google Scholar 

  3. 3

    Wright, J. D. & Miller, K. G. Miocene stable isotope stratigraphy, Site 747, Kerguelen Plateau. Proc. ODP Sci. Res. 120, 855–866 (1992).

    CAS  Google Scholar 

  4. 4

    Stott, L. D., Kennett, J. P., Shackleton, N. J. & Corfield, R. M. The evolution of Antarctic surface waters during the Paleogene, inferences from the stable isotopic composition of planktonic foraminifera. Proc. ODP Sci. Res. 113, 849–864 (1990).

    Google Scholar 

  5. 5

    Curry, W. al. Init. Rep. ODP A 154, (1995).

    Google Scholar 

  6. 6

    Weedon, G. P., Shackleton, N. J. & Pearson, P. N. The Oligocene time scale and cyclostratigraphy on the Ceara Rise, Western Equatorial Atlantic. Proc. ODP Sci. Res.(in the press).

  7. 7

    Flower, B. P., Zachos, J. C. & Paul, H. Milankovitch-scale variability recorded near the Oligocene/Miocene Boundary: Hole 929A. Proc. ODP Sci. Res.(in the press).

  8. 8

    Berggren, W. al. in Geochronology, Time Scales and Global Stratigraphic Correlation(eds Berggren, W. A., Kent, D. V., Aubry, M.-P. & Hardenbol, J.) 129–212 (Spec. Publ. 54, Soc. Econ. Paleontol. Mineral., Tulsa, OK, (1995)).

    Google Scholar 

  9. 9

    Jenkins, G. M. & Watts, D. G. Spectral Analysis and its Applications(Holden-Day, San Francisco, (1968)).

    Google Scholar 

  10. 10

    Berger, A. & Loutre, M. F. Insolation values for the climate of the last 10 m.y. Quat. Sci. Rev. 10, 297–317 (1991).

    ADS  Article  Google Scholar 

  11. 11

    Pisias, N. G., Shackleton, N. J. & Hall, M. A. Stable isotope and calcium carbonate records from hydraulic piston cored Hole 574A: high-resolution records from the middle Miocene. Init. Repts DSDP 85, 735–748 (1985).

    CAS  Google Scholar 

  12. 12

    Zachos, J. C., Quinn, T. M. & Salamy, K. Earliest Oligocene climate transition: Constraints from high resolution (104 yr) deep-sea foraminiferal δ18O and δ13C time series. Paleoceanography 11, 251–266 (1996).

    ADS  Article  Google Scholar 

  13. 13

    Wu, G., Herguera, J. C. & Berger, W. H. Differential dissolution, modification of late Pleistocene oxygen isotope records in the western equatorial Pacific. Paleoceanography 5, 581–594 (1990).

    ADS  Article  Google Scholar 

  14. 14

    Bassinot, F. al. Coarse fraction fluctuations in pelagic carbonate sediments from the tropical Indian Ocean: A 1500-kyr record of carbonate dissolution. Paleoceanography 9, 579–600 (1994).

    ADS  Article  Google Scholar 

  15. 15

    Curry, W. B., Duplessy, J. C., Labeyrie, L. D., Oppo, D. & Kallel, N. P. Changes in the distribution of δ13C of ΣCO2 between the last glaciation and the Holocene. Paleoceanography 3, 317–342 (1988).

    ADS  Article  Google Scholar 

  16. 16

    Wright, J. D. & Miller, K. G. in The Antarctic Paleoenvironment: A Perspective on Global Change(eds Kennett, J. P. & Warnke, D. A.) 141–165 (Antarctic Res. Ser. 56, Am. Geophys. Un., Washington DC, (1993)).

    Google Scholar 

  17. 17

    Barrett, P. J. (ed.) Antarctic Cenozoic History From CIROS-1 Drill Hole McMurdo Sound. NZ DSIR Bull. 245, 1–251 (1989).

    Google Scholar 

  18. 18

    Haq, B. U., Hardenbol, J. & Vail, P. R. Chronology of fluctuating sea levels since the Triassic. Science 235, 1156–1167 (1987).

    ADS  CAS  Article  Google Scholar 

  19. 19

    Miller, K. al. Drilling and dating New Jersey Oligocene-Miocene sequences; ice volume, global sea level, and Exxon records. Science 271, 1092–1095 (1996).

    ADS  CAS  Article  Google Scholar 

  20. 20

    Clemens, S. C. & Tiedemann, R. Eccentricity forcing of Pliocene–Early Pleistocene climate revealed in a marine oxygen isotope record. Nature 385, 801–804 (1997).

    ADS  CAS  Article  Google Scholar 

  21. 21

    Mix, A. al. Benthic Foraminifer stable isotope record from Site 849, 0–5 Ma: Local and global climate changes. Proc. ODP Sci. Res. 138, 371–412 (1995).

    Google Scholar 

  22. 22

    Diester-Haass, L. & Zahn, R. Eocene-Oligocene transition in the Southern Ocean: History of water mass circulation and biological productivity. Geology 24, 163–166 (1996).

    ADS  CAS  Article  Google Scholar 

  23. 23

    Woodruff, F. & Savin, S. M. Mid-Miocene isotope stratigraphy in the deep sea: high-resolution correlations, paleoclimatic cycles, and sediment preservation. Paleoceanography 6, 755–806 (1991).

    ADS  Article  Google Scholar 

  24. 24

    Flower, B. P. & Kennett, J. P. Middle Miocene deepwater paleoceanography in the southwest Pacific: Relations with east Antarctic ice sheet development. Paleoceanography 10, 1095–1113 (1995).

    ADS  Article  Google Scholar 

  25. 25

    Flower, B. P. & Kennett, J. P. Middle Miocene ocean/climate transition: High-resolution oxygen and carbon isotopic records from DSDP Site 588A, Southwest Pacific. Paleoceanography 8, 811–843 (1993).

    ADS  Article  Google Scholar 

  26. 26

    Shackleton, N. J. Marine Petroleum Source Rocks(eds Brooks, J. et al.) 423–434 (Spec. Publ. 26, Geol. Soc., London, (1987)).

    Google Scholar 

Download references


We thank S. Clemens, P. DeMenocal, P. Koch, N. Shackleton, J. Revenaugh and Q.Williams for comments and discussion; K. Vencil for technical assistance; and S. D'Hondt, L. Hinnov and K. Miller for reviews that significantly improved the manuscript. Samples for this project were provided by the Ocean Drilling Program. This work was supported by the USSSP and NSF.

Author information



Corresponding author

Correspondence to James C. Zachos.

Supplementary information

Supplementary Image

Coherency spectrum based on a linear regression fit to 6 upper Oligocene and lower Miocene microfossil datums (JPG 93 kb)

Supplementary Information

Information on the Hole 929A time seriesanalytical work and age model (DOC 20 kb)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Zachos, J., Flower, B. & Paul, H. Orbitally paced climate oscillations across the Oligocene/Miocene boundary. Nature 388, 567–570 (1997).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing