Orbital pacing of Eocene climate during the Middle Eocene Climate Optimum and the chron C19r event: Missing link found in the tropical western Atlantic



[1] A high-resolution stratigraphy is essential toward deciphering climate variability in detail and understanding causality arguments of events in earth history. Because the middle to late Eocene provides a perfect testing ground for carbon cycle models to reconstruct the transition from a hothouse to an icehouse world, an accurate time scale is needed to decode climate-driving mechanisms. Here we present new results from ODP Site 1260 (Leg 207) which covers a unique expanded middle Eocene section (magnetochrons C18r to C20r, late Lutetian to early Bartonian) of the tropical western Atlantic including the chron C19r transient hyperthermal event and the Middle Eocene Climate Optimum (MECO). To establish a detailed cyclostratigraphy we acquired iron intensity records by XRF scanning Site 1260 cores. We revise the shipboard composite section, establish a cyclostratigraphy and use the exceptional eccentricity modulated precession cycles for orbital tuning. The new astrochronology revises the age of magnetic polarity chrons C19n to C20n, validates the position of very long eccentricity minima at 40.2 and 43.0 Ma in the orbital solutions and extends the Astronomically Tuned Geological Timescale back to 44 Ma. For the first time the new data provide clear evidence for an orbital pacing of the chron C19r event and a likely involvement of the very long (2.4 myr) eccentricity cycle contributing to the evolution of the MECO.

1. Introduction

[2] The Eocene period increasingly receives attention in paleoclimate research due to its high potential to provide deep insight to carbon cycle dynamics and climate evolution in Earth's history [Zachos et al., 2008]. One focus in this context is on the middle Eocene because it covers the commencement of the critical transition period in Earth history from the warm, high-diversity greenhouse of the early Eocene to the icehouse conditions of the early Oligocene. In particularly the late Lutetian to early Bartonian interval (44–38 Ma) is of interest due to the lack of knowledge for the presence and volume of polar ice sheets in this period [Prentice and Matthews, 1988; Zachos et al., 2001, 1994; Tripati et al., 2005; Edgar et al., 2007; Burgess et al., 2008; Koeberl and Montanari, 2009]. This interval also encompasses the Middle Eocene Climate Optimum (MECO), a ∼500 kyr long global warming event that challenges the understanding of the long-term (100 kyr to 1 myr) carbon cycle [Bohaty and Zachos, 2003; Bohaty et al., 2009; Edgar et al., 2010; Sluijs et al., 2013], and the chron C19r event, a transient hyperthermal event [Edgar et al., 2007]. To decipher the causal mechanisms of these climate events in the late Lutetian to early Bartonian a precise time scale is required.

[3] Currently, an accurate orbital time scale for the late Lutetian to early Bartonian is not available [Pälike and Hilgen, 2008; Vandenberghe et al., 2012]. Although a tuned record covering magnetochrons C18r to C21n from the Contessa Highway section (Italy) has been compiled [Jovane et al., 2010], the tuning to the 405 kyr eccentricity cycle has been seriously questioned [Vandenberghe et al., 2012]. Astronomical tuning to eccentricity clearly depends on the stability of the orbital solution [Laskar et al., 2004; Westerhold et al., 2007]. Recent improvements resulted in solutions considered to be now valid back to 50 Ma [Laskar et al., 2011a; Laskar et al., 2011b]. Comparison of orbital solutions with the eccentricity cycle pattern in geological data suggests a validity of the La2011 solution for eccentricity [Laskar et al., 2011b] back to ∼54 Ma [Westerhold et al., 2012]. However, so far no comparison has been done for the interval from 44 to 38 Ma where the three most recent orbital solutions (La2004 [Laskar et al., 2004], La2010 [Laskar et al., 2011a], La2011 [Laskar et al., 2011b]) start to diverge [Westerhold et al., 2012].

[4] Here we present new results from Ocean Drilling Program (ODP) Site 1260 [Shipboard Scientific Party, 2004] which was drilled during ODP Leg 207 [Erbacher et al., 2004] characterized by an expanded middle Eocene section of the tropical western Atlantic (Figure 1). The cyclic sedimentary succession, further distinguished by its superb bio and magnetostratigraphic age control, covers magnetochrons C18r to C20r in the late Lutetian to early Bartonian [Shipboard Scientific Party, 2004; Suganuma and Ogg, 2006; Edgar et al., 2007]. Our objectives are the construction of a cyclostratigraphic framework based on high-resolution XRF core scanning iron (Fe) intensity data, comparison of the geological data with orbital solutions, and subsequent astronomical tuning. The tuned record further allows us to assess the quality of the tuning of the Contessa section, establish astronomically tuned ages for magnetochron boundaries and climatic events, examine climate variability in detail, and evaluate the relation between climate variability and orbital forcing.

Figure 1.

Location map for ODP Site 1260 and the Contessa section on a 40 Ma paleogeographic reconstruction in Mollweide projection (http://www.odsn.de [Hay et al., 1999]).

2. Materials and Methods

2.1. X-ray Fluorescence (XRF) Scanning

[5] We XRF scanned a 100 m long middle Eocene nannofossil radiolarian chalk record from ODP Site 1260 (2549 m present water depth, ∼2500–2600 m paleo water depth) (Figure 1) [Shipboard Scientific Party, 2004; Suganuma and Ogg, 2006; Edgar et al., 2007; Sexton et al., 2006]. The XRF Core Scanners acquire bulk-sediment chemical data from split core surfaces [Röhl and Abrams, 2000]. Acquired elemental intensities are proportional to the elemental concentrations in the sediment. Intensities are guided by the X-ray hardware, the energy level of the X-ray source, the count time, and the physical properties of the sediment, namely grain size and porosity [Röhl and Abrams, 2000; Röhl et al., 2007; Tjallingii et al., 2007; Westerhold et al., 2007]. In total, >117 m along the shipboard composite depth record from 32 to 135 m composite depth (mcd) [Shipboard Scientific Party, 2004] spanning the time interval from magnetochrons C18r to C20r (39.5–44 Ma) were scanned with extensively overlapping sections at the splice tie-points to test the shipboard splice. XRF core scanning records have been shown to be more consistent for hole-to-hole correlation and possess a significantly higher signal-to-noise ratio than standard shipboard data like physical properties [Röhl and Abrams, 2000; Westerhold et al., 2007]. In this study, we present the iron (Fe) intensity data for this interval. Elemental intensity data were originally obtained with two different generations of XRF Core Scanners (I and II). However, a reference section was scanned with both scanners to allow direct comparison of the data. This way all XRF I Fe intensity data were transformed into the corresponding XRF II Fe intensity data using the calculated regression (see supporting information Figure S1).1 The complete data set presented is available online in the PANGAEA database (http://doi.pangaea.de/10.1594/PANGAEA.817217).

[6] XRF Core Scanner data for ODP Site 1260 were collected directly at the split core surface of the archive half with XRF Core Scanner I and II at the MARUM-University of Bremen. The split core surface was covered with a 4 µm thin SPEXCerti Prep Ultralene foil to avoid contamination of the XRF measurement unit and desiccation of the sediment. XRF Core Scanner I data were collected every 2 cm down-core over a 1 cm2 area using a generator setting of 20 kV, 0.087 mA, and a sampling time of 30 s. XRF Core Scanner I was operated with a KEVEX Psi Peltier Cooled Silicon Detector and a KEVEX X-ray Tube 52500008-02 with molybdenum (Mo) target material. XRF Core Scanner II data were collected every 2 cm down-core over a 1.2 cm2 area using generator settings of 10 kV, a current of 0.15 mA, and a sampling time of 30 s. XRF Core Scanner II was operated with an AMPTEK XR-100CR Si-PIN Diode detector with 159 eV X-ray resolution, an AMPTEK PX2T/CR Power Supply/Shaper and Amplifier, and an Oxford Instruments 50W XTF5011 X-ray tube with rhodium (Rh) target material. Raw data spectra were processed by the analysis of X-ray spectra by Iterative Least square software (WIN AXIL) package from Canberra Eurisys.

3. Results

3.1. XRF Core Scanning Data

[7] The Site 1260 Fe intensity record (supporting information Tables S1–S3) reveals persistent and remarkably clear sedimentary cycles for the studied interval (Figure 2). The chron C19r event and the MECO show very prominent Fe peaks, which are related to dissolution of calcium carbonate in the deep ocean during the events [Edgar et al., 2007; Bohaty et al., 2009] and therefore relative enrichment of the clay fraction in the sediment [Röhl and Abrams, 2000; Röhl et al., 2007]. The aim of this study is to make use of the regular Fe data pattern to establish a cyclostratigraphic framework for the middle Eocene interval covering C18r to the top of C20r rather than interpreting the complete geochemical composition of the sediments using other elemental data derived from XRF scanning.

Figure 2.

The Fe intensity record from ODP Site 1260. The positions of the C19r event [Edgar et al., 2007] and the Middle Eocene Climate Optimum (MECO) [Bohaty et al., 2009] are indicated. The cores have been scanned in a combined effort at MARUM XRF Core Scanners I and II. XRF Core Scanner I data have been adjusted to XRF Core Scanner II data (see text). Also shown: core image of the spliced record, relative declination data (black crosses) with paleomagnetic interpretation [Suganuma and Ogg, 2006; Edgar et al., 2007], and the position for the top of plankic foraminfera O. beckmanni zone [Edgar et al., 2010]. The gray bar with question mark indicates the proposed position of C18n.2n at Site 1260.

3.2. Revised Composite Depth Record

[8] The Fe data reveal a first discrepancy in the shipboard splice between the lower part of Core 1260B-1R and the upper part of Core 1260A-7R around 48 mcd (supporting information Figure S2a). Here the shipboard splice is lacking 40 cm of sediment equivalent to one sedimentary cycle. Around 77 mcd the base of Core 1260B-5R and the top of Core 1260A-10R are mismatched by 68 cm, an equivalent of almost two sedimentary cycles in the new composite record (supporting information Figure S2b). Another missing 77 cm long interval was identified between the base of Core 1260B-9R and the top of Core 1260A-14R at 115 mcd (supporting information Figure S2c). The last readjustment had to be made at 123.5 mcd between Core 1260A-14R and the top of Core 1260B-10R. The shipboard splice lacks 61 cm here. In total the revised composite record (rmcd) assembled down to 135.48 rmcd is now 2.46 m longer than the shipboard composite (supporting information Tables S4 and Table S5). Two coring gaps not covered by either of the two drill holes will be discussed in section 4.2. The change in composite depth with respect to the shipboard results is in the order of 2% showing the difficulty in correlation between holes and splice construction without fingerprint excursions (e.g., ash beds) or data with significantly higher signal-to-noise ratio than standard shipboard data.

3.3. Time Series Analysis

[9] Evolutionary Wavelet Power Spectra of the Site 1260 Fe intensity data record clearly show strong spectral power located at distinct periods (Figure 3) (Wavelet software was provided by C. Torrence and G. Compo http://paos.colorado.edu/research/wavelets). Multitaper Method (MTM) spectra were calculated by the kSpectra Toolkit software from SpectraWorks using 3 tapers and a resolution of 2 [Ghil et al., 2002]. Confidence levels are based on a robust red noise estimation [Mann and Lees, 1996].

Figure 3.

Spectral analysis of Site 1260 Fe data. (a) Wavelet power spectrum of Fe intensity data in the depth domain from 45 to 135 revised meters composite depth (rmcd). The contours are normalized linear variances with blue representing low spectral power, and red representing high spectral power. The black contour lines enclose regions with >95% confidence. Shaded regions on either end indicate the cone of influence where edge effects become important. Distinct bands that run across the spectra indicate the dominance of Milankovitch frequencies. (b–g) MTM power spectra of Fe intensity data from various intervals in the depth domain. The spectra have been calculated by the kSpectra Toolkit using 3 tapers and resolution of 2; bars indicate bandwidth. Background estimate and hence confidence levels (90%, 95%, and 99%) are based on a robust red noise estimation [Mann and Lees, 1996]. In preparation for analysis outliers and the long-term trend were removed and the time series linearly resampled at 2 kyr intervals. (h) Wavelet power spectrum of Fe intensity data using the magnetostratigraphy and the CK95 time scale exhibiting precession and eccentricity cycles.

[10] The most dominant cycles are 40, 55, and 66 cm long. Longer cycles of 2.1–3.1 and even 8–12 m length are present (Figure 3). The dominance of the 40, 55, and 66 cm cycles in the spectral analysis is related to the high amplitude of these cycles in the XRF Fe intensity data. The magnetostratigraphy as well as the diagnostic frequency ratios unambiguously relate these bands to Milankovitch cycles (Figure 3h). Applying either the Cande and Kent [1992, 1995] or the GPTS2004 [Ogg and Smith, 2004] ages for the Magnetic Polarity Times Scale (GPTS) reveals that the 40, 55, and 66 cm cycles are related to precession, and the 2.1–3.1 as well as 8–12 m cycles are related to the short (95 kyr) and long (405 kyr) eccentricity cycles, respectively. A dominant obliquity component is not obvious.

[11] The wavelets exhibit shifts of the dominant cyclicity through time, and their understanding in relation to either changes in sedimentation rates or the dominant orbital frequency is important for the cyclostratigraphic interpretation and subsequent orbital tuning. From the base of the record to 120 rmcd the 54 cm and the 2.3 m cycles are prevailing (Figure 3f). At 120 rmcd this pattern shifts to a 66 cm/3.1 m cycles persistent up to 90 rmcd (Figure 3e). Further up, 75–90 rmcd, cyclicity shifts back to the 55 cm length (Figure 3d), where after the dominance shifted again to 66 cm (Figure 3c). Finally, around 55 rmcd, the record is characterized by 40 cm cycles. Assuming the longer 2.1 and 3.1 m cycles represent the 95 kyr component of the short eccentricity then the dominant 55 cm and 66 cm cycles correspond to the 23 and 19 kyr components of precession.

3.4. Cyclostratigraphy

[12] To construct a cyclostratigraphy we follow the metronome approach of Herbert et al. [1995] assuming that each cycle represents a duration equal to the mean period of the orbital parameter that forced its formation. We used direct cycle counting and Gaussian band-pass filtering of the Fe data (Figure 4). Before band-pass filtering, we removed the C19r event Fe peak from the data to avoid filter disturbance. We started counting precession cycles at the first cycle below the very high Fe peak at 44 rmcd, where we propose here the location of the onset of the peak-MECO interval at 1260 [Bohaty et al., 2009] (Figure 4). Available biostratigraphic data point to a hiatus at the top of the study interval between core sections 1260A5R-CC (36.10 rmcd) and 1260A6R-1 (37.40 rmcd ±1.3m) [Edgar et al., 2007; Shipboard Scientific Party, 2004; Edgar et al., 2010]. Because the top of the planktic foraminifera O. beckmanni zone at 37.40 rmcd (Figure 2) has been reported to occur in the normal chron C18n.2n [Wade et al., 2011] we suggest that the C18r/C18n.2n reversal should be located in the dark layer possibly representing the peak-MECO around 44 rmcd. The polarity boundary therefore might just be located above the highest paleomagnetic data point at 43.86 rmcd [Suganuma and Ogg, 2006] (Figure 2). In any case, this uncertainty does not affect the cyclostratigraphy and orbital tuning of the sediment below 44 rmcd.

Figure 4.

Cycle counting of Site 1260 Fe data (red line). Fe data, paleomagnetic interpretation, and core images are plotted on the new revised composite depth (rmcd, this study). The small green numbers indicate the number of precession cycles counted from the base of the peak-MECO [Bohaty et al., 2009] event. Precession cycles (green) and short eccentricity cycles (black) have been extracted by Gaussian filtering. The stable long eccentricity cycle (405 kyr) is indicated in blue. Counting all three orbital frequencies results in a consistent duration of 3.848–3.895 myr for the studied interval.

[13] We counted 185 precession cycles as well as 40.5 short and 9.75 long eccentricity cycles for the 44.25–134.90 rmcd interval (Figure 4). Assuming an average duration of 21 kyr for a precession cycle, 95 kyr for a short eccentricity cycle and 405 kyr for a long eccentricity cycle [Herbert et al., 1995; Röhl et al., 2001, 2003; Laskar et al., 2004; Röhl et al., 2007; Westerhold et al., 2007; Westerhold and Röhl, 2009] we obtain a duration of 3.848–3.948 myr with the precession cycle counts resulting into an estimate of 3.885 myr. The precession cycle counts include three potential cycles in the core gap at 57 rmcd, the assumption that the C19r event represents one precession cycle, and the core gap at 95.73 rmcd spans two precession cycles. These notions are validated by the short eccentricity cycle pattern and the correlation to the La2004 orbital solution for precession (see section 4.2). The cycle counting also provides new estimates for the duration of magnetochrons C20n to C19n. We observe that 9 ± 1 precession cycles (189 ± 21 kyr) encompass chron C19n, 44 ± 1 precession cycles (924 ± 21 kyr) span chron C19r, and 61 ± 1.5 precession cycles (1281 ± 32 kyr) cover chron C20n (Table 1).

Table 1. Duration of Magnetochrons
ChronCK95 (myr)GPTS2012 (myr)Contessa Highway (myr)ODP Site 1260
Cycle Count (myr)Tuned (myr)
C19n0.2640.2360.260189 ± 21200
C19r1.0150.9111.030924 ± 21890
C20n1.2531.1311.2501281 ± 321298

4. Discussion

4.1. Age Calibration Based on the Stable 405 kyr Eccentricity Cycle

[14] The eccentricity modulated precession cycles present in the Fe data allow us to develop a high-resolution cyclostratigraphy and compare the eccentricity modulation in the data with the orbital solutions for eccentricity. The extraction of the modulation by eccentricity is the prerequisite to establish an absolute time scale based on orbital tuning [Shackleton and Crowhurst, 1997; Westerhold et al., 2007; Zeeden et al., 2013]. The first-order calibration is the identification of the very long eccentricity cycle minima in the geological data. Using the age scale derived from relative cycle counting (Figure 5a), the amplitude modulation of the Fe intensity data show a clear minimum around 5.2 myr (45–50 rmcd). A second although less pronounced minimum is observed around 8.0 myr relative time scale (∼115 rmcd). The La2004, La2010, and the La2011 solutions all show a prominent very long eccentricity minimum at 40.2 Ma [see Westerhold et al., 2012, Figure 2 therein]. We use this as an initial tie point and correlate the minimum at 5.2 myr relative age scale to very long eccentricity cycle 100 (Figure 5a). The second minimum will be in very long eccentricity cycle 107, which is consistent with the minimum observed in La2010a, La2010d, and La2011 [see Westerhold et al., 2012, Figure 2d therein]. The correlation suggests that the La2010a solution is closest to the geological data around 43 Ma but due to the less clear expression of the minimum in the Fe data this awaits further confirmation. Thus, the cyclostratigraphy based on the Fe record from Site 1260 starts at 40.1 Ma with a 405 kyr cycle identification numbering system (Figures 4 and 5b) reaching back to 44.0 Ma spanning 405 kyr cycle 100–109.

Figure 5.

Comparison of relative cycle counting timescale with present astronomical solutions for orbital eccentricity. (a) Results of relative timescale versus tuned time scale. From top to bottom: Site 1260 Fe intensity data (red line) versus relative cycle counting age in Ma. We assume that the average duration of one precession cycle is 21 kyr and set the peak-MECO onset arbitrarily to 5.0 Ma. Amplitude demodulation of the precession cycle for the cycle counting (light green) and tuned (dark green) age model are plotted with the proposed position of the very long eccentricity minima (gray bars). The amplitude modulations have been extracted with the program ENVELOPE [Schulz et al., 1999]. Orbital solution La2010d [Laskar et al., 2011a] and La2011 [Laskar et al., 2011b] are plotted with the stable long eccentricity cycle number as counted backward in time from the present [cf. Wade and Pälike, 2004]. Ages for magnetochron boundaries and the C19r event are on the tuned time scale. (b) Fe intensity data (red), benthic stable δ13C data (black line with crosses) [Edgar et al., 2007], core images, and magnetochrons plotted on rmcd depth. The Gaussian band-pass filter (0.10 ± 0.03) for the proposed 405 kyr cycle as depicted from Fe (red) and benthic δ13C (black) is plotted above. Note the difference in filters around 65 rmcd which could lead to an overestimation of cycles only using the Fe data for this interval.

4.2. Orbital Tuning

[15] Based on the comparison with the benthic stable isotopes record [Edgar et al., 2007] Fe intensity peaks in the darker clay-rich layers correspond to lighter values in δ13C (supporting information Figure S3). As shown in previous Paleogene successions [Lourens et al., 2005; Pälike et al., 2006; Zachos et al., 2010; Westerhold et al., 2011] we assume that lighter carbon isotope values and therefore Fe peaks (representing more clay-rich layers) correlate with both eccentricity and precession maxima. It has to be noted here that it is still not known whether the lighter carbon isotope values and Fe peaks in the Eocene correspond to precession minima or maxima. Despite this uncertainty we chose to tune the Fe peaks to precession maxima of La2004 (supporting information Figure S4). Precession solutions are not available for La2010 and La2011. The accuracy of the tuned age for the base of C20n is limited as La2004 differs from La2010 and La2011 in time intervals older than 42.7 Ma (supporting information Figure S4). To account for the phase relation uncertainty and the uncertainty in the La2004 precession solution we estimated and report absolute ages with an error of ±20 kyr. Sedimentation rates based on the cycle counting and the tuned age model are consistent (supporting information Figure S5).

[16] The extracted amplitude modulation (AM) of the tuned Fe intensity record and the AM based on the cycle counting (Figure 5a) are quite similar. The tuned AM shows some more distinct patterns which are most likely introduced by the tuning process itself [Huybers and Aharonson, 2010]. The excellent correlation between these two records suggests that the amplitude modulation of the precession cycle in this particular record can be used to anchor the record to eccentricity. The modulation of the precession cycle by eccentricity in the Site 1260 record is not as well expressed as for example in Eocene sediments from the Walvis Ridge [Lourens et al., 2005; Röhl et al., 2007; Westerhold et al., 2007]. As revealed by the MTM spectra (Figure 3) the precession related cycles have a higher amplitude than the eccentricity component in the Fe data. Site 1260 is located in low latitudes and thus could be prone to record the dominant high precession and eccentricity forcing [Short et al., 1991; Laepple and Lohmann, 2009]. The relatively expanded Site 1260 record is characterized by well pronounced precession cycles. In contrast, lower sedimentation rates smooth precession and attenuate eccentricity as observed in Blake Nose [Röhl et al., 2001, 2003] and Walvis Ridge records [Lourens et al., 2005; Westerhold et al., 2007]. Therefore, we are confident that the orbitally tuned age model is very robust and provide new durations and ages for magnetochrons and their boundaries (Figure 5a; Tables 1 and 2). The orbitally tuned age for the C19r event [Edgar et al., 2007] is 41.508 ± 0.02 Ma, for the long-term warming of the MECO [Bohaty and Zachos, 2003; Bohaty et al., 2009] is 40.45–40.05 Ma, and the peak-MECO if located at 44 rmcd [Bohaty et al., 2009] 40.05 ± 0.02 Ma. The Lutetian/Bartonian boundary recently was defined to coincide with the top of chron C19n [Vandenberghe et al., 2012] and therefore its astronomically tuned age is 41.061 ± 0.02 Ma.

Table 2. Absolute Ages of Magnetochrons
ChronCK95 Age (Ma)GPTS2012 Age (Ma)Contessa Highway Tuned Age (Ma)ODP Site 1260 Tuned Age (Ma)
C18n (o)40.13040.14540.120 
C19n (y)41.25741.15441.25041.061
C19n (o)41.52141.39041.51041.261
C20n (y)42.53642.30142.54042.151
C20n (o)43.78943.43243.79043.449
C21n (y)46.26445.72446.310 

4.3. Duration of Chron C19 and C20n

[17] The astronomical calibration of the Site 1260 Fe record provides new absolute age estimates for magnetochron boundaries C19n and C20n (Figure 6 and Table 2). Our absolute ages for chron boundaries are slightly younger than in GPTS2012 [Ogg, 2012; Vandenberghe et al., 2012]. The ages in GPTS2004 [Ogg and Smith, 2004; Luterbacher et al., 2004] are based on considering an ash layer in C21n.33 of 45.6 Ma (FC 28.02 Ma) and are most likely too young (see discussion in Westerhold et al. [2012]). For GPTS2012 this C21n ash-based tie point was replaced by the 40Ar/39Ar age of 43.35 ± 0.5 Ma (FC 28.201 Ma) for the Mission Valley Ash (California, USA) near the base of C20n [Prothero and Emry, 1996; Walsh et al., 1996; Smith et al., 2010]. Our absolute ages are ∼400 kyr (C20n) and ∼200 kyr (C19n) younger than in CK95 [Cande and Kent, 1995]. CK95 used the relatively old age for the ash layer in C21n.33 of 46.8 Ma (FC 28.02 Ma). The durations of C19n and C19r (Table 1) as derived from the new Site 1260 record are very similar to GPTS2012. In contrast, due to the revised age for the top of chron C20n its duration is consistent with CK95 but >150 kyr longer than in GPTS2012. Our results support the perception that the marine-magnetic anomaly width assignments as done by Cande and Kent [1992] need minor revision and thus remove systematic differences between Normal/Reversed marine magnetic anomaly widths and Normal/Reversed polarity chron durations [Ogg, 2012].

Figure 6.

Comparison of magnetochron boundary ages and durations for C18 to C20 based on estimates from Cande and Kent [1995] (CK95), Ogg and Smith [2004] (GPTS2004), Ogg [2012] (GPTS2012), Contessa [Jovane et al., 2010], and ODP Site 1260 (this study). Black lines track changes between different estimates. Radioisotopic age (FC at 28.201 Ma) of Mission Valley ash as used in GPTS2012 [Vandenberghe et al., 2012; Ogg, 2012] is plotted (pink dot). Magnetochron C18n at Site 1260 (gray shade) indicates that this chron has not been at this site but is inferred here based on correlation from other sites.

[18] Comparing our Site 1260 record with the tuned Contessa Highway section [Jovane et al., 2010] shows that the Contessa record seems to be offset by a full 405 kyr cycle at C20n and a half 405 kyr cycle at C19n. This offset is identical to the offset between Site 1260 and CK95 suggesting a problem in the tuning as already noted in the GPTS2012 [Vandenberghe et al., 2012]. It is not possible to decipher the long eccentricity cycle in both magnetic susceptibility and carbonate data of the Contessa section [Jovane et al., 2007], which might be the reason for the offset of a 405 kyr cycle at C20n. The difference for chron C19n could be explained by a discrepancy of the tuning of the Contessa section for C19r. Jovane et al. [2010] related thicker carbonate beds to long eccentricity minima. If this relation continued throughout the record than the carbonate beds at the C19n/C19r boundary in Contessa should correlate to a 405 kyr minimum rather than a maximum as described in Jovane et al. [2010, Figure 5 therein]. Based on our tuning of the 1260 record the C19n/C19r and the C19r/C20n boundaries are located in a 405 kyr minimum consistent with simultaneous thicker carbonate beds in the Contessa Highway section. Between these magnetochron boundaries one more 405 kyr minimum is present about 200 kyr prior to the C19r event. We propose that the C19r event should be located in the thicker marly bed at 124.8 m in Contessa and that the thicker carbonate beds around 122.8 m in Contessa correlate to a 405 kyr minimum. The carbonate beds at Contessa increase in the upper part of C19r indicating higher sedimentation rates. Disregarding this increase most likely resulted into an underestimate of the eccentricity cycle thickness and also leads to the counting of too many cycles in the upper part of C19r as the case in Jovane et al. [2010]. For the same reason the duration of C19n is longer than at Site 1260.

[19] Based on late to late middle Eocene cyclostratigraphies [Pälike et al., 2001; Röhl et al., 2004; Pälike et al., 2006] our orbital calibration of the Site 1260 Fe intensity records successfully extends the astronomical timescale into the middle Eocene (back to 44.0 Ma). We provide a robust link to the stable part of the astronomical solutions (La2004, La2010, and La2011) and demonstrate that there is the need for future orbital solutions to validate very long eccentricity minima at 40.10–40.45 as well as 42.90–43.30 Ma.

4.4. Orbital Forcing of the Chron C19r Event and the MECO

[20] The new high-resolution records in combination with the astronomically calibrated time scale provide unprecedented insight into middle to late Eocene, namely the duration and onset of the chron C19r event [Edgar et al., 2007] and the MECO [Bohaty and Zachos, 2003; Bohaty et al., 2009]. The chron C19r event is characterized by a dark clay rich layer as also expressed by a distinct peak in Site 1260 Fe intensity, a negative carbon isotope excursion (CIE), and strong dissolution due to the shoaling of the Carbonate Compensation Depth (CCD; Figure 7) [Edgar et al., 2007]. All these characteristics of the C19r event are indicative of early Paleogene hyperthermal events like the PETM [Zachos et al., 2005] although with a smaller magnitude similar to the Eocene Thermal Maximum 2 at ∼53 Ma [Lourens et al., 2005; Edgar et al., 2007; Stap et al., 2010]. However, the C19r event occurred in slightly cooler climate several million years after the Early Eocene Climate Optimum (EECO, ∼51 Ma) and ∼1.0 myr prior to the onset of the MECO. There are no significant CCD changes one million years before and after the event as expressed in regular Fe intensity variations. At 40.5 Ma the average Fe values rise toward the peak-MECO at 40.05 ± 0.02 Ma indicating a shallowing of the CCD [Edgar et al., 2007] as observed in other ocean basins [Bohaty et al., 2009]. Although the duration of the C19r event can be estimated by orbitally tuning to be ∼20 kyr, carbonate dissolution could have condensed the record at Site 1260. With respect to the cyclostratigraphic framework we estimate the maximum duration in the order of 40–50 kyr, otherwise the phase relation between eccentricity related cycles and orbital forcing will move out of phase. This estimate is similar to durations observed at other transient hyperthermals in the early to middle Eocene [Stab et al., 2009; Sexton et al., 2011].

Figure 7.

Orbital pacing of the C19r event and the MECO. The upper plot shows Fe intensity (red) (this study), benthic δ18O (blue), and δ13C (green) [Edgar et al., 2007] of Site 1260 on tuned age scale plotted against the combined eccentricity, obliquity and precession (E + T − P, black) of La2004 [Laskar et al., 2004] from 39.5 to 43 Ma. Core images also on tuned time scale. The lower plot shows ETP from 30 to 45 Ma. According to the tuning of Site 1260 the C19r event corresponds to an exceptionally strong peak in insolation (green arrow). The 400 kyr MECO warming trend (light shaded yellow) prior to the peak-MECO CIE (darker shaded yellow) is associated with a very long eccentricity (2.4 myr) minimum.

[21] Through correlating the Site 1260 Fe intensity record to orbital insolation calculations it becomes quite clear that the C19r event occurred at an unusually high value of the ETP curve (green arrow in Figure 7). In fact, the peak at 41.520 Ma is the highest insolation value (∼573 W/m2) of the last 45 Ma using the La2004 solution at 65°N at 21 June. Only at 22.501 Ma an insolation peaks of similar magnitude exceeding 570 W/m2 is present. This exceptional strong insolation could have crossed a threshold leading to the rapid release of carbon into the ocean–atmosphere system causing the C19r event in a similar way as proposed for early Eocene hyperthermals [Lourens et al., 2005; Galeotti et al., 2010; Zachos et al., 2010; Lunt et al., 2011]. At 41.913 Ma the insolation reaches higher values (Figure 7), however, no carbon isotope excursion or peculiar Fe intensity peak at this time points to an event resembling the C19r event. Here we speculate that insolation was not strong enough to trigger an C19r-like event at 41.913 Ma. The short duration of 20–50 kyr, the rapid recovery and a potential orbital trigger of the C19r event are very similar characteristics also observed for other hyperthermals found in the early middle Eocene (48–50 Ma) that have been linked to release of dissolved organic carbon due to orbitally driven changes in ocean ventilation [Sexton et al., 2011]. However, to decipher the causal mechanism involved by investigating which phases of orbital changes caused which type of climate change and when requires to know the exact phase relationship between precession and geochemical data. An integrated approach of modeling scenarios, similar to what has been done by Sloan and Huber [2001], and higher resolution geochemical data are urgently needed to tackle this problem. Other problems that need to be overcome are the current uncertainties in orbital models for obliquity and precession at 41.5 Ma and the uncertain exact duration of the C19r event, which cannot be determined precisely at Site 1260 alone due to carbonate dissolution at the site.

[22] The MECO is an episode of gradual warming over 500 kyr accompanied by an increase in CO2 and a shoaling of the CCD but does not exhibit a pronounced shift in δ13C as typical for early Eocene hyperthermal events [Bohaty and Zachos, 2003; Bohaty et al., 2009; Bijl et al., 2010]. A transient (∼50 kyr) carbon isotope excursion (CIE) that coincides with a 1.5°C warming is only reported from the peak-MECO [Bohaty et al., 2009]. The actual cause for the increase in atmospheric CO2 during the MECO is unknown. Recent carbon cycle modeling proposes an imbalance in long-term carbon fluxes increasing the atmospheric pCO2 leading to global warming [Sluijs et al., 2013]. Although this model is not tied to a specific mechanism an increase in volcanic outgassing is suggested as one potential mechanism. As a response to the warming, thermal expansion of the ocean is hypothesized to lead to a sea-level rise that causes a redistribution of carbonate burial from the deep-ocean to the continental shelves and a stabilization of shelf sediment weathering. The net effect then leads to higher pCO2 and a shallowing of the CCD [Sluijs et al., 2013]. Data to test this scenario and to explain why the MECO abruptly ended with a transient CIE are still lacking. Although the Contessa Highway section reveals thicker carbonate beds in the warming phase of the MECO in the upper chron C19r it is not clear if this is related to long eccentricity minima or redistribution of carbonate sedimentation to the shelves. The increase in sedimentation rates in the Contessa succession at least suggests increased burial in this interval which would balance observed deep-sea decrease in carbonate accumulation rates [Bohaty et al., 2009]. Nevertheless, our new record provides a couple of more details for future model exercises. First, the onset of the MECO is now dated at 40.50 ± 0.02 Ma. Second, the warming trend leading to the CIE at the peak-MECO covered 415 kyr which verifies previous estimates [Bohaty et al., 2009]. Third, the commencement of the CIE, at 44 rmcd at Site 1260, was at 40.085 ± 0.02 Ma. And fourth and even more importantly, the MECO warming trend coincides with a very long eccentricity cycle minimum (pink bar in Figure 7). A recent investigation comparing ODP Site 1051 and 1260 suggested that the top of the MECO interval could be truncated at Site 1260 [Edgar et al., 2010]. Until additional data are available to test this we propose to use the current estimate for the duration of the MECO and regard this as a minimum estimate. It needs to be noted here that previous estimates for the duration of the MECO have been based solely on linear interpolation between magnetostratigraphic age tie points [Bohaty et al., 2009]. Thus, our orbitally tuned age model is an enormous progress over previously developed age models for late middle Eocene cores where the MECO has been identified. The placement of the MECO in a very long eccentricity cycle minimum is of high interest because when eccentricity is low, the amplification of precessional insolation changes weakens, and seasonal variability declines [Horton et al., 2012]. Less seasonal contrast leads to less hot-summer/cold-winter contrasts and hence might result in less hot-summer-driven monsoonal rainfall. If so, the precipitation for a 400 kyr period of relatively low eccentricity might have been decreased causing less weathering in the tropics and less input of solutes to the ocean resulting into a rise of the CCD [Pälike et al., 2012]. In the following, the reduced input of nutrients most likely lowered productivity resulting in a decrease in mass accumulation rates (supporting information Figure S5) and in a less effective biological pump sequestering atmospheric CO2. The delay in sequestration could have led to slowly rising pCO2 and therefore global warming over 500 kyr if volcanic and metamorphic CO2 outgassing remained constant or was even elevated. Unfortunately, a decrease in weathering is thought to be inconsistent with warming and rising pCO2 [Berner et al., 1983; Sluijs et al., 2013]. Further climate modeling should further test to what degree orbital configuration affects precipitation distribution and intensity strongly influencing runoff and weathering fluxes.

5. Conclusion

[23] High-resolution XRF core scanning Fe intensity data obtained from ODP Site 1260 (Demerara Rise, tropical western Atlantic) have been used to compile a revised composite section that provides new insights in the timing and duration of paleoceanographic events and their relation to orbital influences. Dominant eccentricity modulated precession cycles in the Fe record allow a detailed cyclostratigraphy for a ∼4 myr interval in the late Lutetian to early Bartonian and support the influence of orbital pacing. The extracted eccentricity modulation pattern from the Fe record has been compared to latest orbital solutions showing a consistent very long eccentricity minimum in La2010 and La2011 at 40.2 Ma. A second very long eccentricity minimum was identified at 43.0 Ma, although not well represented in the orbital solutions. Additional high-resolution cyclostratigraphic records are required to further compare to orbital solutions for this interval of the time scale. Orbital tuning of our record provides ages for magnetochron boundaries that are consistent with GPTS2012 and suggests an offset in tuning of the Contessa Highway section in the upper chron C19r. Based on the new astronomical age of the C19r event (41.508 ± 0.02 Ma), we demonstrate how this hyperthermal event was triggered by exceptionally high insolation forcing. In addition, our data point to orbital pacing playing a role in MECO warming as it is located in a very long eccentricity minimum at 40.2 Ma. The time scale presented extends the astronomically tuned timescale into the middle Eocene and provides a sound basis to further test carbon cycle models.


[24] This research used data acquired at the XRF Core Scanner Lab at the MARUM-Center for Marine Environmental Sciences, University of Bremen, Germany, and samples and/or data provided by the Ocean Drilling Program (ODP). We are indebted to Roy Wilkens (Hawaii) for core images analysis work. We thank V. Lukies for assisting in XRF core scanning as well as W. Hale and A. Wülbers for core handling. Funding for this research was provided by the Deutsche Forschungsgemeinschaft (DFG). We thank Jim Ogg, two anonymous reviewers, Editor Janne Blichert-Toft and Associate Editor Matthew Huber for thorough reviews and comments improving the paper. The data reported in this paper are tabulated in the supporting information and archived at the Pangaea database (doi.pangaea.de/10.1594/PANGAEA.817217).