Oligocene climate dynamics

Authors

  • Bridget S. Wade,

    1. Grant Institute of Earth Sciences, School of Geosciences, University of Edinburgh, West Mains Road, Edinburgh, UK
    2. Now at School of Earth, Ocean and Planetary Sciences, Cardiff University, Cardiff, UK.
    Search for more papers by this author
  • Heiko Pälike

    1. Department of Geology and Geochemistry, Stockholm University, Stockholm, Sweden
    2. Now at Southampton Oceanography Centre, School of Ocean and Earth Science, Southampton, UK.
    Search for more papers by this author

Abstract

[1] A planktonic and benthic foraminiferal stable isotope stratigraphy of the Oligocene equatorial Pacific (Ocean Drilling Program, Site 1218) was generated at 6 kyr resolution between magnetochrons C9n and C11n.2n (∼26.4–30 Ma on a newly developed astronomically calibrated timescale). Our data allow a detailed examination of Oligocene paleoceanography, the evolution of the early cryosphere, and the influence of orbital forcing on glacioeustatic sea level variations. Spectral analysis reveals power and coherency for obliquity (40 kyr period) and eccentricity (∼110, 405 kyr) orbital bands, with an additional strong imprint of the eccentricity and 1.2 Myr obliquity amplitude cycle, driving ice sheet oscillations in the Southern Hemisphere. Planktonic and benthic foraminifera δ18O are used to constrain the magnitude and timing of major fluctuations in ice volume and global sea level change. Glacial episodes, related to obliquity and eccentricity variations, occurred at 29.16, 27.91, and 26.76 Ma, corresponding to glacioeustatic sea level fluctuations of 50–65 m. Alteration of high-latitude temperatures and Antarctic ice volume had a significant impact on the global carbon burial and equatorial productivity, as cyclic variations are also recorded in the carbon isotope signal of planktonic and benthic foraminifera, the water column carbon isotope gradient, and estimated percent carbonate of bulk sediment. We also investigate the implications of a close correspondence between oxygen and carbon isotope events and long-term amplitude envelope extrema in astronomical calculations during the Oligocene, and develop a new naming scheme for stable isotope events, on the basis of the 405 kyr eccentricity cycle count.

1. Introduction

1.1. Oligocene Climate and Unipolar Glaciation

[2] The Oligocene has previously been perceived as an interval of climatic stability, yet the oscillation of the Antarctic ice sheet through the Oligocene gave rise to glacioeusatic sea level variations of up to 100 m [Miller et al., 1987, 1991, 1998; Pekar and Miller, 1996; Pekar et al., 2000, 2001]. Documentation of how the climate is influenced by substantial ice volume changes on Antarctica is vital in understanding primary climate processes. However, very little is known about climatic dynamics during this interval of unipolar glaciation and how major fluctuations in Southern Hemisphere ice volume would have propagated through the ocean-atmosphere system.

[3] Here we present a detailed stable isotope record of both planktonic (Figure 1) and benthic foraminifera for a 3.6 Myr interval spanning the early and late Oligocene (26.4 to 30.0 Ma) from the tropical Pacific. These data supply a comprehensive chronology of oceanographic and climatic fluctuations and an isotopic reference curve for the Oligocene Pacific Ocean. Samples were analyzed at a 10 cm resolution giving a sampling frequency of 1 sample/6000 years. This is the highest resolution ever applied over this interval and thus provides unprecedented insights into temperature variability, the carbon cycle major ice volume changes, and their relationship to astronomical cycles.

Figure 1.

Oligocene planktonic foraminifera from Site 1218 (note changes in scale bar). (a) and (b) Recrystallized test surface of Dentoglobigerina globularis from sample 1218B, 15-H-3, 60–62 cm. (c) Globoquadrina venezuelana, umbilical view, sample 1218B, 15-H-3, 60–62 cm. (d) Globoquadrina venezuelana, umbilical view, sample 1218A, 14-H-4, 50–52 cm.

1.2. Site Description

[4] Ocean Drilling Program (ODP) Leg 199, “The Paleogene Equatorial Transect” drilled 8 sites in the tropical Pacific Ocean (Sites 1215–1222). Site 1218 (8°53.378″N, 135°22.00″W) is located in the eastern equatorial Pacific, in a water depth of 4826 m [Shipboard Scientific Party, 2002]. Additional information can be found in the supplementary material. The Oligocene sequence from Site 1218 has a clear magnetostratigraphy and abundant foraminifera through most of the Oligocene. Cyclic variations are evident in sediment color and magnetic susceptibility which reflect orbital variations of solar insolation [Shipboard Scientific Party, 2002]. These features, coupled with the high-resolution stable isotope data generated here, allow a detailed examination of the icehouse world of the Oligocene. During the Oligocene, Site 1218 was located at a paleolatitude of 3–4°N of the equator [Shipboard Scientific Party, 2002] and thus planktonic foraminiferal stable isotopes provide a record of tropical paleoceanography in the Paleogene Pacific Ocean. Sedimentation rates are high and average 15m Myr−1, which facilitates the development of an orbital-scale stable isotope and physical property stratigraphy.

2. Materials and Methods

[5] The methodology and sample processing are detailed in the supplementary material.

3. Results

3.1. Benthic Foraminifera Oxygen Isotopes

[6] It was found in previous studies that stable isotope measurements from modern benthic foraminifera are not in equilibrium with the surrounding seawater [Duplessy et al., 1970]. Thus it is necessary to use species-specific adjustment factors in order to make data comparable between different benthic species. We adjusted Cibicidoides δ18O values to seawater equilibrium by adding +0.64‰ [Shackleton and Opdyke, 1973]. No correction was applied to δ13C values. In the following benthic oxygen isotope description, and associated figures, we refer to the adjusted oxygen isotope values.

[7] Benthic foraminifera δ18O values range between 1.51 and 3.10‰ (Figure 2). The median value of benthic δ18O values is 2.43‰, with a standard deviation of 0.27‰. These values are similar to previous benthic oxygen isotope measurements from the central equatorial Pacific (DSDP 77, 574). From the early to the late part of the record we observe a general increase toward heavier δ18O values: a linear regression shows an increase of ∼0.4‰ throughout the study interval. High-amplitude variations of δ18O values (∼0.75–1.2‰) generally occur on short eccentricity timescales (80–130 kyr), shorter-term variability is approximately in the 0.2–0.5‰ range. In addition to the longer-term increase in δ18O values, we also observe three distinct time periods during which isotope values display heavier than average values: around ∼26.5–26.9 Ma, 27.6–28.0 Ma, and 28.8–29.2 on our timescale (Figure 2). Each one of these periods of heavy isotope values consists of one or two individual events, separated by 100–200 kyr.

Figure 2.

Benthic oxygen and carbon isotope measurements from Site 1218 plotted against astronomical age. Also shown are calculated obliquity and an “ETP” (eccentricity, tilt, and climatic precession) mix, derived from Laskar et al. [2004b]. Bandpass filters for eccentricity and stable isotope measurements were calculated for the ∼405 kyr long eccentricity component, and are the basis for a new event naming scheme, as annotated. Compilation data are from Zachos et al. [2001b] and were age adjusted to our new timescale by mapping from the magnetostratigraphy of Cande and Kent [1995] to the magnetochrons determined for Site 1218 [Shipboard Scientific Party, 2002]. Shown for reference are selected nannofosssil datums observed at Site 1218. Brackets span intervals that correlate to previously observed glacial (“Oi”) events. The position of Oi-2 is predicted on the basis of the astronomical pattern.

3.2. Benthic Foraminifera Carbon Isotopes

[8] The range of carbon isotope values obtained here is comparable to previous, lower-resolution measurements from the central equatorial Pacific (DSDP 77, 574). The median value of benthic δ13C values is 0.45‰, with a standard deviation of 0.25‰. The total range of values spans −0.18 to 1.09‰. Similar to the oxygen isotope values, the benthic δ13C record also displays a trend toward heavier values from old to young; the linear component across the record yields ∼0.47‰. The majority of this shift occurs above 165 rmcd, or younger than ∼28.5 Ma (Figure 2). On <100 kyr timescales, the amplitude variation of δ13C is in the 0.2–0.4‰ range, while larger variability in the ∼0.5–1.0‰ range occurs on eccentricity timescales. Similar to previous studies [Paul et al., 2000; Zachos et al., 2001b], the benthic carbon isotope record shows a very strong response on a ∼405 kyr timescale, as judged by an initial match of the magnetostratigraphy to the timescale of Cande and Kent [1995].

3.3. Planktonic Foraminifera Oxygen Isotopes

[9] Planktonic foraminifera oxygen isotope values record high and low frequency variability and fluctuate between −0.88 and 0.91‰ (Figure 3). Toward the younger end of the record there is an increase in δ18O values with a mean of −0.35‰ at 29.4 Ma, increasing to 0.30‰ at 26.5 Ma. Short frequency (<50 kyr) changes are generally minor (0.5‰), with higher-amplitude variations of 1.2‰ occurring on 80 to 120 kyr timescales. An enrichment in δ18O is recorded at 28.4 Ma, where oxygen isotope values increase by 1.39‰ over 500 kyr from −0.59‰ at 28.44 Ma to 0.80‰ at 27.92 Ma. Planktonic foraminiferal oxygen isotope values then become lighter again to −0.48‰, before a short-term increase in δ18O at 27.54 Ma, where oxygen isotope values change from −0.48‰ to 0.50‰ over 170 kyr. Increases in planktonic foraminiferal oxygen isotope values for 70 to 160 kyr are recorded at 29.16, 27.91 and 26.76 Ma. Oscillations in δ18O (1‰) with a period of ∼110 kyr are particularly evident from 29.0 Ma to 28.5 Ma. The vertical oxygen isotope difference between planktonic and benthic foraminifera (Δδ18O) was calculated and is also shown in Figure 3. The gradient varies between 1.64 and 3.28‰. From 29.5 to 26.4 Ma mean Δδ18O decreases slightly from 2.7‰ to 2.3‰.

Figure 3.

Benthic, planktonic, and bulk oxygen isotope measurements from Site 1218, as well as benthic-planktonic oxygen isotope gradient. Benthic isotope values are adjusted by adding 0.64‰. The shaded area marks an interval with a strong shift of planktonic isotope values that is not represented in the benthic foraminiferal record.

3.4. Planktonic Carbon Isotope Results

[10] Pronounced, cyclic variations are recorded in the δ13C values from Globoquadrina venezuelana of the Oligocene at Site 1218, between 0.32 and 1.67‰ (Figure 4). Variations in δ13C are recorded in 405 and 100 kyr eccentricity cycles, which are particularly apparent toward the upper part of the record. Maxima in planktonic foraminifera carbon isotopes (1.5‰) are recorded on 1.2 Myr timescales at 29.16, 27.91 and 26.76 Ma. In general, trends in the planktonic foraminifera δ13C record match those from the benthic foraminiferal δ13C record, but with a higher degree of variability.

Figure 4.

Benthic, planktonic and bulk carbon isotope measurements from Site 1218, as well as benthic-planktonic carbon gradient.

[11] The difference of δ13C between planktonic G. venezuelana and benthic C. grimsdalei is small (Figure 4), and ranges from a minimum of 0‰ to a maximum of 1‰. The increased gradient in δ13C reoccurs on a 405 kyr cycle, and is mainly due to enrichment of δ13C in planktonic foraminifera.

3.5. Stable Isotope Measurements From Bulk/Fine Fraction Samples

[12] Fine fraction bulk analyses reveal changes in δ18O between −0.56 and 0.39‰ (Figure 3). Although the record of bulk carbonate is short (526 kyr), the δ18O signal is cyclic at the short-term (∼100 kyr) eccentricity period. The bulk δ18O measurements are distinctly lighter than planktonic foraminiferal values, particularly between 27.39 and 27.12 Ma, yet also record an increase of δ18O values toward the younger end of the interval measured.

[13] High-amplitude, cyclic changes are recorded in the fine fraction bulk δ13C data, between 1.33 and 1.94‰ (Figure 4). The δ13C bulk values are generally 0.7‰ heavier than those recorded by the planktonic foraminifera G. venezuelana, but also show cyclic shifts on the short eccentricity scale (∼100 kyr). Amplitude variations of individual cycles are strongly coherent with those of the planktonic record. Similar to the bulk δ18O measurements, the bulk δ13C values also increase toward the younger end of the measured interval. An interesting feature is the smoothness of the bulk δ13C record for timescales shorter than eccentricity. All stable isotopic data from foraminifera and bulk carbonate measurements are tabulated in a data report [Wade and Pälike, 2004].

3.6. Timescale Generation

[14] ODP Site 1218 provided a high-resolution (∼1–2 cm/kyr) biogenic sediment record from the late Paleocene to the early Miocene. These sediments were found to contain an uninterrupted set of geomagnetic chrons, as well as a detailed record of calcareous and siliceous biostratigraphic datum points [Shipboard Scientific Party, 2002]. Shipboard lithological proxy measurements and shore-based X-ray fluorescence scanning revealed clearly recognizable cycles that can be attributed to climatic change, driven by Milankovitch style orbital variations of the Earth. Discovering drill sites with a well-defined magnetostratigraphic and biostratigraphic record that also show clear lithological cycles is rare and valuable, and opened the opportunity to develop a detailed stratigraphic intersite correlation (H. Pälike et al., Integrated stratigraphic correlation and improved composite depth scales for ODP Sites 1218 and 1219, submitted to Ocean Drilling Program Scientific Results, 2004), as well as providing the data to refine and extend the astronomical age calibration for the Oligocene. Lithological proxy parameters that were obtained from depth-adjusted and stacked records from ODP Sites 1218 and 1219, XRF elemental concentration measurements and stable isotope measurements were used to refine our astronomical timescale.

[15] Carbonate content (%CaCO3), estimated from XRF and MST data, varies from 80 to 93% during the studied interval. Both, oxygen and carbon fine fraction measurements, also correlate very highly with lithological indicators of CaCO3 content. These records show an exceptionally strong imprint of both short (∼110 kyr) and long (∼405 kyr) eccentricity, in addition to a weaker imprint of both obliquity (∼40 kyr) and climatic precession (∼21 kyr). The imprint of several amplitude modulations in the astronomically driven lithological and stable isotope records provide a strong relative age control. The basis for our age model was a new astronomical calculation [Laskar et al., 2004a, 2004b], that was also used in the timescale generation. An additional study [Pälike et al., 2004] showed that this new astronomical solution provides a better fit with geological data than previous solutions, at least back to ∼30 Ma.

[16] Our data show a much stronger relative spectral power at both eccentricity periods (∼100 and ∼405 kyr) than insolation calculations. Thus we created an artificial “ETP” (eccentricity, tilt, precession) mix with an enhanced eccentricity component to make our data visually more comparable with the target curve. The age model is primarily based on the eccentricity signal in the data, and was constructed such that light stable isotope values correspond to eccentricity maxima, consistent with previous studies [Shackleton et al., 1999a; Zachos et al., 2001b], and guided by obliquity amplitude modulation patterns.

4. Discussion

[17] The orbitally adjusted timescale and detailed stable isotope data set for both planktonic and benthic foraminifera for Site 1218 provide a comprehensive analysis of Oligocene oceanographic development, particularly in regards to temperature, paleoproductivity and the evolution of the early Cenozoic cryosphere.

4.1. Comparison With Previous Deep-Sea Records

[18] In Figure 2, the benthic isotope results are plotted with previous results from ODP and DSDP sites (data compiled by Zachos et al. [2001a], adjusted to our timescale). Benthic foraminifera carbon isotope results are consistent with previous records from both Atlantic, Pacific and Southern Ocean sites. The close correspondence in δ13C values recorded at a number of sites from different oceans and different latitudes indicates that the Oligocene oceans were homogeneous and the δ13C gradient between the oceanic basins was absent or small during the Oligocene, and was also observed by Billups et al. [2002]. The δ13C variations, on at least eccentricity frequencies or lower, are thus reflecting a global signal rather than a regional record.

[19] The δ18O results of benthic foraminifera at Site 1218 are comparable in value to data elsewhere, except ODP Sites 689, 690 and 748 in the compilation of Zachos et al. [2001a], which are generally 0.3‰ heavier, indicative of cooler and/or more saline deep water around Antarctica.

4.2. Paleoecology of Globoquadrina Venezuelana

[20] The paleobiology of Oligocene planktonic foraminifera is poorly constrained. Information on the depth habitat of extinct planktonic foraminifera can be attained through multispecies stable isotope analyses (see Pearson et al. [1993] for discussion). While it is now established which species occupied the mixed layer through much of the Cenozoic, very few studies have been performed on Oligocene planktonic foraminifera. Groups such as paragloborotaliids have been documented as surface [van Eijden and Ganssen, 1995], thermocline (B. Wade, unpublished data, 2004) and subthermocline dwellers [Douglas and Savin, 1978]. The same problem exists for Globoquadrina venezuelana, where oxygen isotope values suggest an intermediate dwelling by Biolzi [1983], but is documented as deep dwelling form in the Miocene [Gasperi and Kennett, 1993]. Multispecies analysis (B. Wade, unpublished data, 2004) and comparison with bulk isotope data suggest that G. venezuelana was an upper thermocline dweller. The resulting isotope analyses are thus not considered to reflect those of the mixed layer and are probably heavier in δ18O and lighter in δ13C than those of surface dwelling planktonic foraminifera.

[21] We observe a clear step (1.39‰) in the oxygen isotope values of G. venezuelana at 27.4 Ma (Figure 3). However, this change is not associated with any of the Oi events and is not reflected in the benthic foraminiferal record, suggesting that the heavy δ18O in planktonic foraminifera are not related to a major increase in ice volume. Bulk oxygen isotope results show a general decrease in δ18O but do not match the oxygen isotope record of G. venezuelana. Comparison with the percent carbonate, color and physical properties values over this interval also reveal no comparable alteration. Unlike the Oi events the increased δ18O values are associated with a decrease in δ13C from 1.4 to 0.5‰. The nature of this event is therefore different to the forcing mechanisms elsewhere in the record. This shift may reflect changes in surface water circulation related to flow through the Central American gateways or a slightly cooler water mass as the site moves north of the equator. Alternatively the decrease in both δ18O and δ13C may indicate a change in the depth habitat of G. venezuelana from the upper thermocline, as suggested by multispecies data (Wade, unpublished data, 2004) to a deeper dwelling form, similar to the Miocene [Gasperi and Kennett, 1993].

4.3. Diagenesis and Isotope Fractionation Effects in Deep-Sea Carbonate

[22] Recrystallization of foraminiferal specimens is evident in the scanning electron microscope (SEM) images of planktonic foraminifera at Site 1218 (Figure 1). The SEM images (of unsonicated specimens) reveal partial recrystallization of test walls, attached secondary calcite, and some infilling of pores and apertures. Specimens chosen for isotopic analysis were devoid of calcite infilling and sonication was applied to remove attached calcite.

[23] Carbonate diagenetic processes are likely to have altered the primary δ18O signal and increased the oxygen isotope values of planktonic foraminifera [Schrag et al., 1992, 1995]. This may cause paleotemperatures reconstructed from tropical planktonic foraminifera to be underestimated [Pearson et al., 2001]. There is no means of constraining this diagenetic offset at present, but it nevertheless must be small relative to the signal amplitude. The preservation of distinctive stable isotope signals, particularly in the carbon isotope record, the cyclic nature of the data, and the large (2.5‰) gradient in δ18O between G. venezuelana and benthic foraminifera isotope values suggests that diagenesis did not destroy the original signal. Additional multispecies analysis (not shown) reveals clear offsets in δ18O between species, which are not expected to be retained if the tests were completely recrystallized.

[24] To assess the preservation of carbonate, nannofossil assemblages were examined from intervals of both high and low %CaCO3 and δ13C. These showed no obvious preservational differences (J. Backman and I. Raffi, personal communication, 2004), suggesting that the changes in δ13C are not of secondary, diagenetic origin. Samples were dominated by the nannofossil Cyclicargolithus floridanus, however, one key component of the assemblages in high carbonate and heavy δ13C intervals were the increased size of nannoplankton and the abundance of sphenoliths. Sphenoliths were previously documented as k-selected species, being adapted to a warm water, oligotrophic and stable environment [Haq and Lohmann, 1976; Haq, 1980; Lohmann and Carlson, 1981; Perch-Nielsen, 1985; Aubry, 1992]. However, their paleoecology has recently been questioned by Gibbs et al. [2004] who noted a covariance between the abundance of sphenoliths and productivity variations during the Pliocene. B. Wade and P. Bown (Calcareous nannofossils in extreme environments: The Messinian salinity event, submitted to Palaeogeography, Palaeoclimatology, Palaeoecology, 2004) also noted a high abundance and almost monospecific assemblages of sphenoliths in diatom rich samples from the Messinian, Cyprus. This would suggest that contrary to previous reports, sphenoliths were capable of inhabiting a wide range of environments, and enhanced their abundance during high productivity intervals. This finding is particularly important as it suggests the shifts in planktonic foraminiferal and bulk isotope δ13C are a direct response to productivity changes in surface waters, and not a secondary effect of diagenetic alteration. The corresponding cyclic changes in percent carbonate and increases in δ13C suggest an enhanced production of biogenic material and equatorial Pacific productivity fluxes that were directly related to oscillations in solar insolation.

4.4. Orbital Climate Forcing During the Oligocene

[25] The recognition of isotope variability that might be associated with orbital climate forcing relies on high-resolution records from the same site. Even though moderate-resolution records exist for the Oligocene as part of data compilations [Zachos et al., 2001a], stratigraphic correlation uncertainties, different depositional settings, water depths and latitudes between sites that are part of these compilations make the identification of orbitally influenced variability difficult.

[26] The high-resolution isotope records from Site 1218 have enabled us to document the effect of orbital forcing on the Oligocene climate. In order to evaluate the relative strength of potential astronomical forcing in the stable isotope data of this study, and to estimate phase relationships between the time series, we performed cross-spectral analyses between the benthic and planktonic δ18O and δ13C measurements that were presented in Figures 2, 3 and 4. These spectra, coherency estimates and phase plots are shown in Figure 5.

Figure 5.

Blackman-Tukey cross-spectral analysis of benthic and planktonic stable isotopes, against astronomical tuning target ETP. Both astronomical and stable isotope series were re-sampled at equal intervals corresponding to the median time step of ∼6 kyr. Cross-spectra, confidence intervals (C.I.), and bandwidths (BW) were calculated with 140 lags after linear detrending. Phase estimates are plotted only where coherency values are significant and spectral power is present. The eccentricity and obliquity frequency bands are marked by bars. Stable isotope values were flipped prior to analysis. Spectra were calculated with AnalySeries [Paillard et al., 1996].

[27] The confidence intervals in Figure 5 solely depend on the sample spacing, and the spectral analysis method used, and do not depend on the choice of a particular noise model. The interpretive choice of a noise null-model determines to what level of significance spectral peaks cannot be explained by noise. Paleoclimatological data often exhibit a red-noise pattern, with higher power at lower frequencies, and autoregressive models are used to calculate a best fitting noise continuum. We provide auxiliary material that demonstrates the significance estimates for benthic δ13C spectral peaks by assuming a red-noise background. However, statistical tests [Schulz and Mudelsee, 2002] performed on our spectra reveal that a red-noise model is not appropriate to describe the spectra, which do not show the indicative straight-line shape on log-log frequency versus spectral power plots. Instead, our noise-spectra are best matched with an exponential fit (supplementary material).

[28] In our data, spectral power is strong with significant coherence for the 40 kyr obliquity band in the benthic δ13C data. Benthic isotope data and planktonic δ13C all show a strong response around the short (∼100 kyr) eccentricity cycle, and all four stable isotope series show strong peaks with significant coherence in the long eccentricity (∼405 kyr) band. It is possible that an even higher sample resolution might increase the spectral power in the climatic precession band (∼19–23 kyr).

[29] Power in the orbital bands of eccentricity (406, 130–94 kyr) and obliquity (40 kyr) is indicated by spectral analysis of the isotopic records. Cross-spectral analysis reveals significant coherency between the stable isotope records and orbital forcing and indicates that the Oligocene climate was sensitive to orbital changes in solar insolation.

[30] The benthic stable isotope time series show a strong and coherent response at the ∼100 and ∼405 kyr short and long eccentricity periods. This eccentricity imprint is also reflected in the planktonic isotope series, but with a less clear response to short eccentricity in the planktonic δ18O series. The pronounced ∼100 and ∼405 kyr periodicity within the isotope records confirms the importance of eccentricity oscillations in pre-Pleistocene climate.

[31] Surprisingly, the benthic carbon isotope record shows a stronger response at the ∼40 kyr obliquity period than the benthic oxygen isotope record, for which the obliquity peak is offset toward slightly lower frequencies, and for which filters display a less stable phase relationship, thus reducing its power. Phase estimates are only meaningful where spectral power (and coherency) values are high.

[32] The phase relationships at the eccentricity periods are as expected for the benthic stable isotopes, but the response of planktonic δ18O is 180 degrees out of phase with the other three isotope records. The individual short eccentricity cycles are not completely symmetric, thus some difference in phases can be expected. A lag of planktonic oxygen isotopes with respect to the planktonic carbon record at the obliquity frequency is also suggested. We cannot make any confident predictions about phase relationships at the climatic precession frequencies yet, but anticipate that an increased sample resolution could enhance our observations.

[33] The short eccentricity cycle, present in lithological as well as stable isotope data, is amplitude modulated with a period of ∼2.4 Myr. This amplitude modulation pattern is consistent with that of astronomical target curves (Figure 6), and together with the clear ∼405 kyr eccentricity cycles (Figures 2 and 5) supports our age model.

Figure 6.

Comparison of ∼2.4 Myr eccentricity modulation of the short ∼100 kyr eccentricity cycle of two astronomical models [Laskar et al., 1993, 2004b] and benthic stable isotope time series.

[34] High-resolution studies from the late Oligocene/early Miocene have shown a pervasive imprint of orbital frequencies in various climate proxies and stable isotope measurements [Shackleton et al., 1999a, 2000; Paul et al., 2000; Zachos et al., 2001b; Billups et al., 2002, 2004]. These studies, from both high and low latitude sites, have demonstrated several characteristic features of orbital forcing in the Oligocene climate. In particular, a strong imprint of the ∼405 kyr long eccentricity cycle has been found to be consistently associated with dominant variations of δ13C records. During the Oligocene, these long eccentricity cycles, as recorded in the isotope records, appear to be synchronized in phase for records from different latitudes [Billups et al., 2004], and without any significant ocean-to-ocean gradient, at least during the late Oligocene and early Miocene [Billups et al., 2002]. In addition, the ∼405 kyr long eccentricity cycle is also being considered as the most stable of the Earth's long-term astronomical frequencies [Laskar, 1999].

[35] Owing to its global nature, matching long eccentricity cycles with those found in stable isotope measurements provides an ideal cyclostratigraphic correlation tool. What makes the long eccentricity cycle imprint particularly interesting is the suggestion that geological records on this timescale could potentially also show a relationship with eustatic sea level change and sequence boundaries [Gale et al., 2002; Hardenbol, 2003].

4.5. Obliquity Nodes and Glacial Forcing

[36] Previous studies from the Pliocene/Pleistocene implicated periods of obliquity maxima, modulated with a period of ∼1.2 Myr, with times of enhanced glaciations [Lourens and Hilgen, 1997]. In addition to the ∼100 and ∼405 kyr short and long eccentricity cycles demonstrated in Oligocene data sets, we also find a strong direct imprint of the ∼1.2 Myr obliquity cycle (Figure 2). In astronomical calculations, this cycle is present mainly through its amplitude modulation of the ∼40 kyr obliquity cycle.

[37] In contrast to the Pliocene/Pleistocene, studies from the Miocene and Oligocene [Zachos et al., 2001b; Turco et al., 2001] found the opposite: periods of low obliquity amplitude variations correspond to glacial events and periods. Our data confirm this view: we find a ∼1.2 Myr amplitude modulation of the recorded obliquity signal in the benthic δ13C data. We also find a direct imprint of the 1.2 Myr cycle in our benthic oxygen isotope data, where intervals with particularly heavy isotopes correspond to minima of the 1.2 Myr obliquity amplitude cycle, and generally to minima in the ∼405 and ∼100 kyr eccentricity cycles. The direct imprint of the 1.2 Myr cycle most likely arises through an enhanced sensitivity of the cryosphere to low variations in seasonality. Zachos et al. [2001b] suggested that it is the prolonged absence of particularly warm (Southern Hemisphere) summers rather than the occurrence of particularly cool ones that is the significant astronomical factor for inhibiting summer ice melt and thus triggering ice sheet initiation. On the basis of our data, we can support this hypothesis, and postulate that glaciations during the Oligocene are primarily driven by the eccentricity cycles, but with enhanced probability of glaciations during intervals of low obliquity amplitude variations. The 1.2 Myr glacial cycles suggest that ice sheets on Antarctica were dynamic and sensitive to orbital variations of solar insolation. This pattern is also observed in additional studies from Site 1218 for lower parts of the Oligocene [Coxall et al., 2004] which suggests that this relationship between obliquity minima and glaciations has been consistent for at least 15 Myr between the earliest Oligocene and the Miocene.

4.6. A new Paleogene Cyclostratigraphic Naming Scheme, in Relation to Glacial Isotope Events

[38] Previous studies designated named glaciation “events” in the Paleogene, based on intervals where particularly heavy oxygen isotope ratios were observed from several sites, interpreted to represent global changes in ice volume [Miller et al., 1991; Pekar and Miller, 1996; Zachos et al., 1996; Pekar et al., 2002]. In general, these events were numbered consecutively from old to young for each epoch (“Oi-x” for the Oligocene, “Mi-x” for the Miocene etc., where x represents a number/letter code).

[39] Several problems exist with these naming schemes. First, different authors designated the same event name to different events: Oi-1a and Oi-1b are defined differently in the work of Miller et al. [1991] and Zachos et al. [1996], respectively. Second, the named events were based on relatively low-resolution records from a limited number of sites. Thus as more and higher-resolution records became available, a larger number of heavy oxygen isotope ratio time intervals were recognized, not only resulting in the augmentation of existing event names by additional letter codes, but also additional “unnamed” events [Pekar and Miller, 1996]. Third, higher-resolution records, as the one presented in this study, show that some of the previously defined glaciation events in the Oligocene instead comprise several oxygen isotope maxima, implying that the previous lower-resolution records can result in the correlation of separate events that were close, but not identical. Finally, the lack of detailed additional magnetostratigraphic control meant that it was not always obvious where these events were correctly placed stratigraphically. For these reasons, we propose here a new naming scheme that relates astronomical (chronological) information with magnetostratigraphy, in order to provide a more consistent framework for identifying and correlating global isotope events during the Paleogene.

[40] The original studies that defined and correlated the “Oi-x” glaciation events [Miller et al., 1991; Pekar and Miller, 1996; Miller et al., 1998] linked these to glacioeustatic sea level changes as observed from seismic sequence stratigraphy. Additional studies suggested that third-order sequences might be intricately related to ∼405 kyr eccentricity cycles [Gale et al., 2002; Hardenbol, 2003]. In our records, as well as other records from the late Oligocene/early Miocene [Paul et al., 2000; Zachos et al., 2001b; Billups et al., 2004], the ∼405 kyr eccentricity cycle is very well expressed in stable isotope records, particularly in δ13C records. As this frequency component of Earth's eccentricity cycle is considered relatively stable over geological time [Laskar, 1999], it has been proposed previously as the basis for chronostratigraphic correlation and timescale designations [Shackleton et al., 1999b].

[41] For different astronomical calculations [Laskar et al., 1993, 2004b] individual ∼405 kyr cycle ages differ by less than 200 kyr throughout the Cenozoic, thus a cycle count number for a given absolute age is unique to within one cycle. Our new scheme is to define events by 405 kyr cycle numbers as follows. The cycle count number is defined by ∼405 kyr eccentricity minima from Laskar et al. [2004b]. Cycle count numbers increase going back in time, starting with count number 1 for the cycle comprising the most recent minimum, ∼10 kyr before present. We define each cycle number to end at the younger zero crossing following each ∼405 kyr eccentricity minimum. In addition to the cycle count, which is a formal part of our designation, we extend this code for practical purposes to also include a subscripted code for the geological epoch together with the magnetochron closest to the ∼405 kyr eccentricity minimum (following the style of Cande and Kent [1995], but excluding subchrons, e.g., C6n instead of C6Cn.1n, and appending rn or nr where the cycle falls close to a reversal). This coding scheme is illustrated in Figure 2. We propose that one could extend this scheme to include ∼100 kyr eccentricity cycles (and maybe also shorter obliquity/climatic precession cycles) within the ∼405kyr eccentricity cycle framework.

4.7. “Oi” Stable Isotope Events

[42] Previous Oligocene glacial events have been defined and described by Shackleton [1986], Miller et al. [1987, 1991], Wright et al. [1992], Pekar and Miller [1996], and Pekar et al. [2002]. Sequence stratigraphic studies and continental margin sequences of the Oligocene [Browning et al., 1996; Miller et al., 1996; Pekar et al., 2000; Kominz and Pekar, 2001] correspond to a number of notable enrichments (>0.5‰) in Oligocene benthic foraminiferal δ18O records [Pekar et al., 2002]. These intervals have been interpreted as significant increases in Antarctic ice volume and support the oscillation of ice volume and sea level throughout the Oligocene Miller et al. [1987, 1991, 1998]; Pekar and Miller [1996]; Pekar et al. [2000, 2001].

[43] The lower resolution in previous studies has prevented the timing and magnitude of glacial (Oi) events being accurately defined. In our records, increases in δ18O are evident in both the planktonic (0.75 to 1.01‰) and benthic (0.77 to 0.93‰) record at 29.16, 27.91 and 26.76 Ma (Figure 3). We interpret these events to correspond to the Oi-2b (here assigned eccentricity cycle 67Ol), Oi-2a (70Ol) and Oi “unnamed” or Oi-2* (73Ol), as defined by Miller et al. [1991] and Pekar and Miller [1996], but also recognize additional events close in time. The timing, amplitudes and durations of the various Oi events at Site 1218 are summarized in Table 1. Events are typified by covarying enrichment in planktonic and benthic δ18O values (>0.7‰), heavy adjusted benthic oxygen isotope values (>2.9‰), and durations from ∼70 to 240 kyr.

Table 1. Isotope Events Summary
Code/Chronδ18O EventAge, MaBiozonesMaximum, ‰Amplitude, ‰Duration, kyrIce Volume, ‰aEustatic Change, mb
ForamNannoδ18Obenδ18Oplδ18Oben%δ18Opl
67Ol-C9nOi-2b26.76P21bCP19b3.000.730.860.85160–2360.6559
70Ol-C10nOi-2a27.91P21bCP19a2.900.800.771.0170–1300.5853
73Ol-C10rnOi-2*29.16P21aCP19a2.920.170.930.75106–1360.7064
76Ol-C11rOi-230.35P20CP18       

[44] The coeval shifts in benthic and planktonic δ18O records are thought to indicate expansion in Antarctic ice volume. The amplitude of δ18O variations at the Oi events within the planktonic foraminifera record is 0.75 to 1.00‰ and exceeds the magnitude recorded at Site 77 [Keigwin and Keller, 1984]. The interval of heavy oxygen isotope events are longer in the benthic foraminifera record than in the planktonic foraminifera record by approximately 30–70 kyr. This reflects substantial high-latitude cooling that preceded the ice volume increases, which are picked up by the benthic foraminiferal but not tropical planktonic foraminiferal record. The amplitude of δ18O change in benthic foraminifera at Oi-2b (67Ol) and Oi-2a (70Ol) are comparable to those recorded at sites elsewhere [Pekar et al., 2002]. However, we record a much greater change in benthic and planktonic δ18O at Oi-2* (73Ol) than in earlier studies. This arises because the event has previously been unresolved in low-resolution studies of the Oligocene. The covariance between planktonic and benthic foraminiferal oxygen isotope records indicates fluctuations in global ice volume and the sustained existence of oscillating ice sheets on Antarctica during the Oligocene.

[45] Increases in benthic foraminiferal δ18O are also evident at intervals throughout the record (at 28.89, 28.27, 27.71 and 26.53 Ma). These episodes are 45 to 78 kyr in duration and are characterized by shifts in benthic δ18O of 0.44 to 1.24‰ and heavy benthic oxygen isotope values (>2.90‰). They are associated with minor increases in planktonic foraminiferal δ18O and are thought to reflect small changes in Southern Hemisphere ice volume. For lower-resolution studies there is therefore the potential to misidentify the Oi events for these shifts in benthic δ18O. While the original studies defined the Oi events based entirely on δ18O values, the placement of the Oi events here is constrained through simultaneous increases in planktonic foraminiferal δ18O and maxima in benthic and planktonic foraminiferal δ13C.

4.8. Ice-Volume Changes During the Oligocene

[46] Fluctuations in global ice volume, eustatic sea level, and temperature change during glaciation events can be constrained by comparing the amplitude of δ18O variation between planktonic and benthic foraminifera. Oxygen isotope increases range from 0.77 to 0.93‰ and 0.75 to 1.01‰ in the benthic and planktonic foraminiferal records, respectively. While the amplitude of change in benthic foraminiferal δ18O is 0.93‰ at Oi-2* (73Ol), ice volume fluctuations cannot exceed the variation recorded in planktonic foraminifera δ18O (0.75‰), the rest of change has to be attributable to deep sea temperature changes and/or salinity variations and other factors. If a maximum of 75% of the benthic foraminiferal increases are ascribed to ice volume [Fairbanks, 1989], this would suggest a glacial effect of 0.58 to 0.70‰ during the Oi events. The magnitude of change recorded at Site 1218 suggests there were large variations in the amount of ice on Antarctica during the Oligocene.

4.9. Eustatic Sea Level Change in Relation to Oi Events

[47] Previous workers have suggested that third-order eustatic sea level changes, identified from seismic reflection profiles, are related to orbitally influenced climate variations, particularly the ∼405 kyr eccentricity cycle [Gale et al., 2002; Hardenbol, 2003] but also the ∼1.2 Myr obliquity amplitude modulation cycle [Lourens and Hilgen, 1997].

[48] On the basis of a the late Pleistocene calibration of 0.11‰ increase in δ18O per 10 m sea level fall [Fairbanks and Matthews, 1978], the eustatic fall at the Oi events is of the order of 50 to 65 m (Table 1). This is comparable with the eustatic change estimated by Pekar and Miller [1996] of 45 to 75 m during Oi-2b (67Ol), but has been further constrained here by the planktonic foraminiferal δ18O record. Using simultaneous changes between benthic and planktonic oxygen isotope values as an indicator of eustatic sea level changes, we calculated a relative sea level curve by multichannel singular spectral analysis with the software SSA Toolkit [Ghil et al., 2002]. We selected those spectral components that are common to both signals, and reconstructed a combined smoothed time series (Figure 7). Our records indicate that one can match third-order eustatic sea level variations with high resolution oxygen isotope records, and that major glaciation cycles are indeed driven by the confluence of eccentricity cycles and longer-term obliquity amplitude variations.

Figure 7.

Sea level variation estimates based on covariation of benthic and planktonic oxygen isotope measurements and in relation to sequence stratigraphic estimates of sea level variations. The eustatic sea level curve of Haq et al. [1987] was age adjusted to the timescale of Cande and Kent [1995]. A comparison of the median benthic δ18O values and derived temperature estimates [Erez and Luz, 1983] with Mg/Ca temperatures estimates from the same site [Lear et al., 2004] results in a seawater oxygen isotope ratio of approximately −0.62‰ compared to present.

4.10. Paleotemperature Estimates From Stable Isotope Measurements

[49] Several factors need to be taken into consideration in the calculation of paleotemperatures from foraminifera. This includes the ambient isotopic composition of seawater (δ18Osw) in which the foraminiferal calcite was precipitated (which is influenced by salinity and global ice volume), diagenetic modification [Schrag et al., 1992, 1995; Pearson et al., 2001], and the concentration of seawater carbonate ion [Spero et al., 1997; Zeebe et al., 1999; Russell and Spero, 2000]. These limitations have to be considered in the following discussion, and we regard paleotemperatures to be estimated values and not absolute. Owing to the uncertainty regarding diagenetic imprint of δ18O, salinity of the Oligocene tropical ocean and the unconstrained depth habitat of G. venezuelana it was not possible to calculate sea surface temperatures for the Oligocene tropical Pacific. Paleotemperatures were calculated from adjusted benthic foraminiferal oxygen isotope measurements assuming constant ice volume and salinity. While we recognize that uniform ice volume is not correct this equation provides a gauge of general variation in deep sea temperatures.

[50] Our adjusted benthic oxygen isotope values vary around a median value of ∼2.4‰. At this time Antarctica was partially glaciated. Additional measurements of Mg/Ca ratios from the same interval and the same site [Lear et al., 2004] indicate that the average bottom water temperature for the time interval studied here was 3.6 ± 1.0°C, assuming modern seawater Mg/Ca. Combining this estimate with the equation of Erez and Luz [1983] results in an average δ18Osw of −0.62‰ during this part of the Oligocene, and bottom water temperatures that fluctuated in the range of ∼1–6°C.

[51] The 0.19 to 0.23‰ change during the Oi events that are not related to ice volume (section 4.8) would then be attributed to deep sea temperature changes of 0.76 to 0.92°C (with ∼0.25‰°C [Epstein et al., 1953]). In the planktonic foraminiferal record changes in δ18O ascribed to temperature range between 0.20 and 1.72°C. The largest change occurs around Oi-2a (70Ol), where planktonic foraminiferal oxygen isotope values shift by 1.01‰ (Table 1).

4.11. Oligocene δ13C Variations and the Carbon Cycle

[52] Enriched δ18O values and eustatic fall during the Oi events are associated with positive δ13C excursions in benthic foraminifera (Figure 2). The magnitude and cyclicity of the carbon isotope shifts is similar to that recorded in carbon isotope records of the Oligocene and Miocene [Miller and Fairbanks, 1985; Pisias et al., 1985; Woodruff and Savin, 1991; Flower and Kennett, 1993; Zachos et al., 1996; Paul et al., 2000; Zachos et al., 2001b], suggesting that this climatic regime is consistent for at least 10 million years.

[53] The high-resolution data from Site 1218 indicate that substantial, pronounced (1‰) oscillations in δ13C are also reflected in the planktonic foraminifera and Δδ13C records. Heavy planktonic foraminifera δ13C values (>1.4‰) and enhanced carbon isotope gradients are recorded at 29.1, 27.9 and 26.8 Ma, associated with the Oi events (Figure 4). This indicates that the carbon isotope variations were a global signal and influenced the entire water column.

[54] These increases in planktonic foraminifera δ13C match cycles within the lithological record and are associated with increased percent carbonate (Figure 8). The significant cyclic variations in δ13C and CaCO3 content may reflect either changes in carbonate export productivity or alternatively a secondary effect of dissolution/diagenesis. While diagenesis has been shown to influence the δ18O of planktonic foraminifera [Pearson et al., 2001] and limited empirical evidence exists to explain high-amplitude changes in planktonic foraminifera δ13C through diagenesis and the carbonate ion effect [Spero et al., 1997; Zeebe et al., 1999; Russell and Spero, 2000], the influence of diagenesis on the δ13C signal of planktonic foraminifera appears to be negligible. Foraminifera and calcareous nannofossils are still abundant during lower (80%) carbonate intervals, and samples contain a full assemblage of Oligocene planktonic foraminiferal fauna.

Figure 8.

Planktonic foraminifera carbon isotopes and benthic-planktonic carbon isotope difference from Site 1218, against calculated percent CaCO3 and Ca, Fe XRF measurements. Note the very close correspondence of carbonate contents and Δδ13C.

[55] Significant and coherent variation at short and long-term eccentricity frequencies are prevalent in δ13C records (Figure 5) and %CaCO3 records. The cyclic δ13C gradient between planktonic and benthic foraminifera has considerable implications regarding the nature of carbon burial during the Oligocene. Orbital forcing and particularly modulation by the eccentricity cycles affected both export productivity and carbon burial.

[56] The association and cyclic variations of δ13C in planktonic and benthic foraminifera, enhanced Δδ13C and increased carbonate content during each of the glacial events is significant. Increases in the water column carbon isotope gradient are also documented during Pleistocene glaciations [Hodell et al., 2003]. This suggests a direct relationship between the carbon cycle and glaciations, and supports the oscillation of the carbon cycle and pCO2 levels in relation to ice sheet development and expansion. Enhanced productivity and an increased rate of burial of organic carbon during glacial events would account for the high-amplitude δ13C variations. Glacioeustatic sea level fall during Oi events may have increased continental weathering rates, and ocean eutrophication by revealing and eroding nutrient rich sediments on the continental shelf and transferring them to the ocean.

[57] Previous studies have suggested that the increased benthic foraminifera δ13C values during the Oi events are related to changes in the deep sea carbon reservoir and enhanced organic carbon burial [Miller and Fairbanks, 1985; Woodruff and Savin, 1991; Paul et al., 2000]. While previous studies demonstrated a very strong response of benthic δ13C values on eccentricity timescales during the late Oligocene [Zachos et al., 2001b; Billups et al., 2002], additional information about the link between the global carbon budget has been provided by studies from earlier times during in the Eocene and Paleocene. In particular, Cramer et al. [2003] described a sequence of re-occurring light δ13C events during eccentricity (and thus climatic precession amplitude) maxima. They proposed a model that simulates the shape of carbon isotope curves very well as a nonlinear response of the carbon budget to astronomical forcing, and which explains the strong response at timescales >0.1 Myr by the residence time of carbon in the oceans [Cramer, 2003].

[58] While the substantial shifts and maxima in the planktonic foraminifer δ13C record and increased δ13C gradient between the surface and the deep ocean are to a large extent attributed to the global carbon reservoir signal, the planktonic carbon record reveal much more pronounced short-term eccentricity variations (110 kyr). The corresponding changes in percent carbonate and increases in planktonic foraminiferal δ13C suggest an enhanced production of biogenic material and mass accumulation rates related to changes in surface water productivity. Indeed, strongly changing productivity patterns and advective flow of eutrophic waters in the area have been suggested by sediment mass accumulation rate studies for the Pacific equatorial region, with a pattern of productivity that was quite different to the single upwelling driven equatorial belt that we see at present [Moore et al., 2004]. Cyclic shifts in the boundary between the North Equatorial Current and the Equatorial Counter Current, related to orbital modulation of wind patterns may account for changes in surface water productivity.

5. Summary and Conclusion

[59] We presented stable isotope data from both planktonic and benthic foraminifera from ODP Site 1218 at a ∼6 kyr resolution that provide a detailed, continuous, 3.6 Myr record of climate and paleoceanographic change in the equatorial Pacific during the Oligocene. Together with additional collaborative studies from the earlier and later Oligocene, they will form part of a Pacific stable isotope reference section. Increases in both planktonic and benthic foraminifera δ18O are used to constrain the magnitude and timing of major fluctuations in ice volume and global sea level change. Glacial episodes, modulated by obliquity and eccentricity frequencies, occurred at 29.16, 27.91 and 26.76 Ma. These correspond to glacioeustatic sea level fluctuations of 50 to 65 m. High amplitude cyclic variations are recorded in the carbon isotope signal of planktonic and benthic foraminifera, the water column carbon isotope gradient and estimated percent carbonate. Maxima in δ13C and the increased Δδ13C values are associated with each of the glacial events. Alteration of high-latitude temperatures and Antarctic ice volume thus had a significant impact on the global carbon burial and equatorial productivity. During the Oligocene ice volume and temperature variations were driven by the 405 kyr and 1.2 Myr orbital cycles, which exerted a prominent influence on climate, equatorial productivity and global sea level changes, and allows the development of a new astronomical cycle based naming scheme.

Acknowledgments

[60] We are grateful for the reviews of J. Zachos and M. Huber. We thank H. Brinkhuis, J. Zachos, J. Backman, P. Pearson and J. Herrle for discussions, C. Chilcott for mass spectrometer assistance at the University of Edinburgh and R. Hesketh and K. Marsden for help with sample preparation, and K. Hajnal, P. Torssander and Å. Wallin for help with benthic isotope measurements and sample preparation in Stockholm. This research used samples provided by the Ocean Drilling Program (ODP). ODP is sponsored by the U.S. National Science Foundation and participating countries under the management of Joint Oceanographic Institutions, Inc. Financial support was provided by the Swedish Research Council and the UK Natural Environment Research Council (NERC) to H.P., and by NERC (reference NER/I/S/2000/00954) and a Rapid Response Award from UK ODP to B.S.W. XRF measurements were obtained at Bremen University with funding through the European Paleostudies programme.