To fully understand the global climate dynamics of the warm early Eocene
with its reoccurring hyperthermal events, an accurate high-fidelity age
model is required. The Ypresian stage (56–47.8 Ma) covers a key interval
within the Eocene as it ranges from the warmest marine temperatures in the
early Eocene to the long-term cooling trends in the middle Eocene. Despite
the recent development of detailed marine isotope records spanning portions
of the Ypresian stage, key records to establish a complete astronomically
calibrated age model for the Ypresian are still missing. Here we present new
high-resolution X-ray fluorescence (XRF) core scanning iron intensity, bulk
stable isotope, calcareous nannofossil, and magnetostratigraphic data
generated on core material from ODP Sites 1258 (Leg 207, Demerara Rise),
1262, 1263, 1265, and 1267 (Leg 208, Walvis Ridge) recovered in the
equatorial and South Atlantic Ocean. By combining new data with published
records, a 405 kyr eccentricity cyclostratigraphic framework was established,
revealing a 300–400 kyr long condensed interval for magnetochron C22n in the
Leg 208 succession. Because the amplitudes are dominated by eccentricity,
the XRF data help to identify the most suitable orbital solution for
astronomical tuning of the Ypresian. Our new records fit best with the
La2010b numerical solution for eccentricity, which was used as a target
curve for compiling the Ypresian astronomical timescale (YATS). The
consistent positions of the very long eccentricity minima in the geological
data and the La2010b solution suggest that the macroscopic feature
displaying the chaotic diffusion of the planetary orbits, the transition
from libration to circulation in the combination of angles in the precession
motion of the orbits of Earth and Mars, occurred
The Ypresian stage, from 56.0 to 47.8 Ma, represents the first
Despite the recent progress in construction of astronomically calibrated age models for the Eocene using geochemical records (Cramer et al., 2003; Lourens et al., 2005; Westerhold et al., 2007; Galeotti et al., 2010; Jovane et al., 2010; Westerhold et al., 2012, 2014, 2015; Westerhold and Röhl, 2013; Littler et al., 2014; Lauretano et al., 2016), a fully consistent astrochronology for the Ypresian is not yet available. Two major issues have to be solved to achieve a complete Ypresian astronomical timescale (YATS): (1) the “50 Ma discrepancy” in magnetostratigraphy (Vandenberghe et al., 2012; Westerhold et al., 2015) and (2) the exact number of 405 kyr eccentricity cycles in magnetochron C23 (Lauretano et al., 2016).
The 50 Ma discrepancy arises from the short duration of magnetochron C23n.2n in Ocean Drilling Program (ODP) Site 1258 data (Suganuma and Ogg, 2006; Westerhold and Röhl, 2009) that results in a very unlikely abrupt global increase in spreading rate for this chron only at around 50 Ma (Vandenberghe et al., 2012). Assuming a 400 kyr longer duration for Chron C22r in the same record would result in lower spreading rates than calculated on the basis of the standard geomagnetic polarity timescale (GPTS) using the synthetic magnetic anomaly profile of the South Atlantic (Cande and Kent, 1992, 1995; CK95). However, the Site 1258 magnetostratigraphy is currently the only available record covering the entire Ypresian. Paleomagnetic interpretation for sediments from Site 1258 is difficult because the core was retrieved by rotary drilling. Minicores from Site 1258 gave a relatively good magnetostratigraphy but it is based on a rather subjective method of polarity interpretation (Suganuma and Ogg, 2006). After revision of the composite record of Site 1258, the interpretation was improved (Westerhold and Röhl, 2009), but the interpretation for Chrons C22r, C23n, and C23r remained ambiguous.
Eccentricity-modulated precession cycles in X-ray fluorescence (XRF) core
scanning iron (Fe) intensities from Site 1258 and Leg 208 sites helped to
set up the first cyclostratigraphic age models for the Ypresian (Westerhold
et al., 2007; Westerhold and Röhl 2009). High-resolution bulk and
benthic stable isotope data for the early (Cramer et al., 2003; Zachos et
al., 2010; Stap et al., 2010; Littler et al., 2014; Lauretano et al., 2015,
2016), middle (Kirtland Turner et al., 2014), and late Ypresian (Sexton et
al., 2011) showed exceptionally strong 100 and 405 kyr eccentricity cycle
variations that were partly utilized for age model construction. In order to
test the 50 Ma discrepancy and the astronomical age model at ODP Site 1258,
high-resolution benthic stable isotope records spanning 54 to 49 Ma were
compiled from ODP Site 1262 (Littler et al., 2014; Lauretano et al., 2015)
and Site 1263 (Stap et al., 2010; Lauretano et al., 2015,
2016) and were astronomically tuned to the La2010d (Laskar et al., 2011a)
orbital solution (Lauretano et al., 2016). In theory, the dominant
eccentricity-related cyclicity in this interval should enable development of
a robust astrochronology. However, a period of low-amplitude variability and
a major shift in benthic
To establish a complete YATS with consistent GPTS ages, a new deep-sea magneto-cyclostratigraphic record is needed to test the ODP Site 1258 magnetostratigraphy. In particular, the durations of Chrons C22 and C23 have to be evaluated. Here we present a new complete magnetostratigraphy spanning Chron C21n to C24n by integrating records from ODP Leg 208 Sites 1262, 1263, 1265, and 1267. New XRF core scanning data and core images are used for ultra-high-resolution correlation between the Leg 208 sites. New calcareous nannofossil data from Sites 1258, 1263, and 1265 are presented and combined with published datums for direct calibration to magnetostratigraphy and revision of age datums. After integration of Leg 208 and ODP Leg 207 Site 1258 isotope data with new XRF data and core images, a cyclostratigraphic framework was compiled that was subsequently astronomically tuned to the La2010 (Laskar et al., 2011a) orbital solution. This study provides the first complete and consistent bio-, chemo-, and magnetostratigraphy for the Ypresian stage. Our new data sets combining two regional records of ODP Legs 207 and 208 reveal unprecedented insight into possible challenges and uncertainties but also demonstrate the high potential of detailed chemostratigraphy, magnetostratigraphy, and biostratigraphy that will be the prerequisite for the major field of chronostratigraphy.
New data for this study were generated from sediment cores retrieved during ODP Leg 207 (Site 1258; Erbacher et al., 2004) and Leg 208 (Sites 1262, 1263, 1265, 1267; Zachos et al., 2004). The carbonate-rich sequences drilled at the equatorial Atlantic Demerara Rise (Leg 207) and the southeast Atlantic Walvis Ridge (Leg 208, Fig. 1) were recovered from multiple holes combined with composite records. For Sites 1262 and 1265 the shipboard composite depth was used (mcd – meters composite depth) for analysis (Zachos et al., 2004). For Sites 1258, 1263, and 1267 the revised composite record was applied (rmcd – revised composite depth; Westerhold et al., 2007, 2015; Westerhold and Röhl, 2009) for further analysis.
Location of ODP Site 1258 and Leg 208 sites on a 50 Ma
paleogeographic reconstruction in Mollweide projection (from
To obtain a complete record for the Ypresian, published data had to be combined with the new XRF Fe core scanning data. The newly acquired data presented here have been measured on the three XRF core scanners at MARUM – Center for Marine Environmental Sciences, University of Bremen, with various hardware generations and under different settings (for details see Supplement). Site 1262 XRF data were obtained from 92.74–104.57 and 111.10–112.40 mcd, and were combined with data from Westerhold et al. (2007, 2012). Site 1263 data reported here are from 168.08 to 289.00 rmcd and were combined with data from Westerhold et al. (2007). Site 1265 XRF data were measured from 228.10 to 275.00 mcd and merged with data from Westerhold et al. (2007). New Site 1267 XRF data were generated from 153.12 to 236.53 rmcd Settings for XRF core scanning of Site 1258 are given in Westerhold and Röhl (2009). Here we report the data from 134.61 to 212.49 rmcd for Site 1258. Combined with published data, 18 000 new XRF core scanning data points (see Supplement) result in a total of more than 30 000 data points covering the latest Paleocene and Ypresian. This enormous data set is the prerequisite for correlating the five drill sites in detail.
To close a stratigraphic gap in the stable isotope record for Site 1263, a
section (230.50 to 239.00 rmcd) of nearby Site 1265 was selected. A total
of 369 sediment samples were collected, freeze-dried, and powdered (Table S8
in the Supplement). The
Published stable isotope data also used in this study have been compiled from Lourens et al. (2005), Zachos et al. (2005, 2010), McCarren et al. (2008), Stap et al. (2009, 2010), Sexton et al. (2011), Westerhold et al. (2012, 2015), Kirtland Turner et al. (2014), Littler et al. (2014), and Lauretano et al. (2015, 2016). All data are provided in the Supplement (and PANGAEA) relative to the Site 1263 depth and with the tuned age added.
Natural remanent magnetization (NRM) was measured on 400 discrete cube
samples (dimensions 2 cm
Smear slides were processed following the standard procedures described in
Bown et al. (1998) in order to investigate calcareous nannofossil assemblages.
High-resolution semiquantitative counting methods, which consist of counting
the number of forms ascribed to the same taxon detected in or normalized to a
prefixed area (i.e., 1 mm
Calcareous nannofossil biostratigraphic analyses were newly performed or refined for the early Eocene of ODP Sites 1258, 1263, and 1265. The biohorizons used in different low-latitude to midlatitude zonations (Martini, 1971; Okada and Bukry, 1980; Agnini et al., 2014), as well as additional biohorizons, were recognized. A set of 168 samples were analyzed from the Demerara Rise (equatorial Atlantic) which allows for the identification of 40 biohorizons. At Walvis Ridge (SE Atlantic), a total of 181 samples were studied from ODP Sites 1263 (77) and 1265 (104), which permits the detection of 27 biohorizons. New data were integrated with ship data or other published results to obtain a more complete and reliable data set. Tables of the calcareous nannofossil biohorizons are given in Tables S31–S35.
All data generated within and available data compiled for this study are
combined in the data set file and available open access online at
Fe intensity data reveal the cyclic pattern commonly observed for the interval from PETM to ETM-2 (Westerhold et al., 2007; Littler et al., 2014), with higher values in darker, more clay-rich layers. A decrease in carbonate content around 51 Ma for all sites (Zachos et al., 2004) is reflected by overall higher Fe intensities. Multiple distinct peaks in Fe data for Chrons C22 and C23n correspond to strikingly bundled sets of clay-rich dark intervals for all sites. At Site 1262 the XRF Fe record ends with the shoaling of the carbonate compensation depth (CCD) above the site in Chron C21r around 93 mcd (Zachos et al., 2004). Generally, the records from Sites 1262 and 1267 are of lower resolution than those from Sites 1263 and 1265 due to the regional decline in carbonate accumulation rates with increasing water depth. Site 1263 shows the most persistent high-resolution XRF Fe intensity signal. A gap in the 1263 record from 229.15 to 233.68 rmcd caused by a drilling disturbance (Zachos et al., 2004) can successfully be bridged by the records from Sites 1265 and 1267. The high-resolution XRF Fe data show consistent patterns that are required to do a detailed site-to-site correlation and integration of Leg 208 Sites 1262, 1263, 1265, and 1267. New XRF Fe intensity data for Site 1258 between PETM and ETM-2 reveal the same eccentricity-modulated precession cycles as observed from Leg 208 sites.
XRF core scanning Fe intensity data from four Leg 208 sites are shown in Fig. S1 in the Supplement and given in Tables S1 to S7. All data plotted versus Site 1263 depth are given in Fig. 2 from ETM-2 to Chron C20r. Data for the interval from the PETM to ETM-2 are plotted in Fig. S7 in the Supplement. Due to the large and very detailed data set and the fact that most of the data from PETM to ETM-2 have been published previously, priority for figures in the main paper is on the interval from ETM-2 to Chron C20r.
To obtain a complete bulk stable isotope record for Leg 208 sites, the gap in
the Site 1263 record was bridged by incorporating bulk data from Site 1265
(Fig. S1a). Bulk stable isotope data from Site 1265 show cyclic variations
between 1.6 and 2.2 ‰ and match with overlapping data from Site 1263
(Fig. 2; Westerhold et al., 2015, 217 to 227 rmcd of 1263). As observed
previously (Zachos et al., 2010; Littler et al., 2014), lighter bulk
To correlate and integrate Leg 208 and Site 1258 records, the new software
tool CODD (Code for Ocean Drilling Data; Wilkens et al., 2017) was utilized.
This tool greatly facilitates handling of large and complex data sets and
allows the use of core images for scientific analysis. For all sites in the
study, core images and all available data were assembled by holes. Then the
composite records were cross-checked and assembled. In order to be able to
use data generated outside of the splice, all cores were mapped onto the
splice using differential stretching and squeezing (Tables S36–S40). Site 1263
was chosen as the reference site to correlate all other sites to because it
has the highest sedimentation rates and the most detailed stable isotope data
(Lauretano et al., 2015, 2016; McCarren et al., 2008; Stap et al., 2010;
Westerhold et al., 2015). XRF Fe and core images primarily guided the
correlation between sites (Fig. 2). Stable isotope data are used to assess
the correlation between Sites 1258 and 1263 because the XRF Fe data of Site 1258 are
dominated by precession cycles and thus difficult to directly correlate to
the eccentricity-dominated cycles at Site 1263 (Westerhold and Röhl, 2009).
Figure 2 shows the outstanding match between the sites just by visual
comparison of the core images. Existing correlations between Leg 208 sites
(Röhl et al., 2007; Westerhold et al., 2007; Lauretano et al., 2015) were
updated and adjusted if needed. Correlation tie points are provided in
Tables S41–S44. The recent correlation between Sites 1262 and 1263 as well as Sites 1263 and
1258 by Lauretano et al. (2015, 2016) based on stable
isotope data were further refined as well. The primary modifications between
Sites 1262 and 1263 were made in a short interval (265 to 267 rmcd) of Site 1263. The
detailed comparison with Site 1265 reveals a gap of about three precession cycles
in XRF Fe at 284.40 rmcd at Site 1263 due to a core break (Fig. S6). No
additional mismatches were recognized, suggesting that the combined Leg 208
records represent the complete stratigraphic sequence for Walvis Ridge.
However, correlation of 1258 to 1263 shows overall good agreement except for
the interval from 230 to 235 rmcd of 1263. Fine-scale comparison of the benthic
Site 1258 and bulk Site 1265 stable
Vector analysis according to the method by Kirschvink (1980) without
anchoring to the origin of the orthogonal projections was applied to the
results of the alternating field demagnetization of NRM to determine the characteristic
remanent magnetization (ChRM). The maximum angular deviation (MAD) values
were computed, reflecting the quality of individual magnetic component
directions. Most of the MAD values are below 10 (Fig. S4, Tables S17–S30).
Figure S5 displays the demagnetization characteristics of a sample with
reversed polarity from C22r and a sample with normal polarity from C22n. As an example of samples with demagnetization behavior with
larger scatter (larger MAD), data from a sample at the C22n–C22r reversal are
plotted in Fig. S5. The larger MADs that can be identified at Leg 208 sites
in a few samples are not simply related to the intensity of their remanent
magnetization. The median destructive field (MDF) of the NRM demagnetization
is comparably low for most of the samples. It ranges from 2.6 to 24 mT (mean
6.1
Recognition of calcareous nannofossil events allows for the magnetic chrons to be clearly identified from C20r to C24r (Figs. 2 and S4, Tables S26–S30). Raw inclination, declination, and intensity data for each measurement step for Leg 208 sites are given in Tables S17–S20. Magnetostratigraphic interpretation is given in Tables S21–S25. Processed paleomagnetic data from Leg 208 sites, the basis for the magnetostratigraphic interpretation, are provided in Tables S26–S30 for each site. The assignment of error bars, as with all magnetostratigraphic data, is a subjective endeavor. The error bar for Leg 208 data marks the interval in which the inclination shifts from clearly reversed to clearly normal polarity or vice versa. Poorer sample resolution and/or ambiguous or transitional inclination values across a reversal will thus increase the error bar. We did not apply an inclination threshold value to mark a shift in polarity because the reversals occur at different seafloor depths at all sites. Drilling depth and compaction difference between sites might have affected the inclination at each site differently. A much more sharply defined length of error bars could be derived from higher-resolution data (e.g., by analyzing u channels), which is beyond the scope of this study.
Having multiple magnetostratigraphic records from the same region combined
with the established high-resolution correlation allows the quality of the
paleomagnetic data to be evaluated and inconsistencies to be identified. This,
again, is crucial for resolving the 50 Ma discrepancy because a single
magnetostratigraphic record from one succession could contain significant
unresolved errors. Plotting all Leg 208 ChRM data and the published Site 1258
magnetostratigraphic interpretation against Site 1263 depth immediately shows
how consistent but also dynamic the magnetostratigraphy of each site can
be (Fig. 2). For example, Chron C22n is clearly too short at Site 1265, which
could be related to the condensed interval in this part. Sites 1262 and 1267,
however, are consistent with the Site 1258 Chron C22n thickness. At Site 1263
the top of Chron C22n is compromised by drilling disturbance, and the base of
Chron C22n is spread over a larger interval than at the other Leg 208 sites.
Chron C21n is well captured at Sites 1258, 1263, and 1267, and the base
at Site 1265 is also captured well. The top of Chron C23n is consistent between Sites 1258, 1262,
1263,
and 1267. The signal from Site 1265 is a bit noisy and a clear identification
for the top of Chron C23n is difficult. The normal interval labeled
as C23n could probably only be C23n.2n. The bottom of Chron C23n is consistent within
error at Sites 1262, 1265, and 1267, with Site 1262 giving the best signal. The
ChRM of Site 1263 does not provide an interpretable signal below 260
rmcd,
preventing the identification of the base of C23n and the entire Chron C24n.
Clearly, comparison to Site 1258 reveals that Chron C23n is too short at Site
1258, probably due to the position of the base Chron C23n. Chron C24n can be
identified in Leg 208 but has relatively larger error bars than the other
chrons. The top of Chron C24n spreads out from
High-resolution correlation between the sites allows us to investigate how
reliably calcareous nannofossil datums can be determined, especially over the
depth transect of Leg 208. Therefore, key taxa were targeted to be identified
at Sites 1263 and 1265 to be compared to the high-resolution work at Site 1262
(Agnini et al., 2007) and low-resolution shipboard data at Site 1267.
Biostratigraphic datums are transformed in biochronological data using the
integrated bio-, magneto-, and astrocyclostratigraphic age model developed in this
study (Tables S31–S35). Age estimations of calcareous nannofossil
biohorizons are generally consistent through the Walvis Ridge sites and in
agreement with recently published biochronological data (see Agnini et al.,
2014, for review). Most of these biohorizons, in particular almost all the
bioevents used in previous and more recent calcareous nannofossil
biozonations, were proven to represent reliable data and powerful tools for
highly resolved correlations. A total of 18 biohorizons across
the study interval (i.e., decrease in diversity of
Data from ODP Site 1258 were produced to biostratigraphically frame the study
succession but these data are also used to investigate the degree of
reliability of calcareous nannofossil data over wide areas. In general, the
stratigraphic positions as well as the ranking and spacing of the biohorizons
detected at this site are in fair agreement with data from Walvis Ridge. The
two biochronological data sets presented for Walvis Ridge and Demerara Rise
sites showed that the ages calculated for some biohorizons, the B of
Time series analysis of early Eocene records that are used here already
showed that the dominant cyclicity in multiple proxy records is related to
eccentricity. The interval from PETM to ETM-2 is dominated by eccentricity-modulated precession cycles that are an impressive recorder of Earth orbital
variations through time and the climatic response to them (Lourens et al.,
2005; Zachos et al., 2010; Littler et al., 2014). Data from this interval not
only allowed the construction of high-precision cyclostratigraphies (Lourens et al.,
2005; Westerhold et al., 2007) but also the test of these astrochronologies as
well as the theoretical astronomical solutions (Westerhold et al., 2007, 2012; Meyers, 2015). As observed in a compilation of late
Paleocene to early Eocene stable isotope data, the
Based on the previous effort we construct a new consistent age model now spanning the entire Ypresian for sites from Leg 208 and Site 1258 by combining a wealth of available information with new high-resolution data, also making extensive use of the previously untouched spliced core images. Published benthic and bulk stable isotope data were combined for Leg 208 and Site 1258 (Figs. 2 and S7), plotted at the Site 1263 revised meters composite depth (rmcd) and detrended for long-term trends (Fig. S9). Data have been linearly interpolated at 2 cm spacing and were then smoothed applying the Igor Pro smooth operation using binomial (Gaussian) smoothing and 30 001 points in the smoothing window. Benthic data from Site 1263 located in the disturbed drilling interval were removed from the combined record. The methods for time series analysis are those of Westerhold et al. (2015). Presence of the short and long eccentricity cycle in the isotope data is well documented (Zachos et al., 2010; Littler et al., 2014; Lauretano et al., 2015, 2016). Strong eccentricity-related cyclicity is clearly present in the evolutive wavelet power spectra of both isotope and XRF Fe data (Fig. S10) for the entire Ypresian, applying the magnetostratigraphic interpretation and using either the CK95 (Cande and Kent, 1995) or the GPTS2012 (Vandenberghe et al., 2012) ages for reversals.
We extracted the 405 and 100 kyr component of the data as detected in the evolutive spectra and plotted the filter over the data to investigate where the signal originates. The first-order tuning was done identifying the 405 kyr cycle in all data consistently. The advantage of having the high-resolution XRF Fe data is the ability to detect distinct modulation of the amplitude in the 100 kyr period related to the 405 and 2.4 myr eccentricity cycle modulations. In Fig. S11 in the Supplement, for example, around 242 and 265 rmcd of Site 1263 the XRF Fe data from Site 1263 and 1267 show very low amplitude variations separated by four 405 kyr cycles. If the amplitude modulation (AM) in the data is mainly driven by eccentricity (Zachos et al., 2010; Littler et al., 2014), then these intervals represent the 2.4 myr eccentricity nodes with minor amplitude variations on the 100 kyr level. Identification of the 2.4 myr minima is very important because they function as major tie points for orbital tuning and test for consistency with astronomical solutions (Westerhold et al., 2012, 2015; Zeeden et al., 2013). The starting point for the stable 405 kyr cyclostratigraphy is eccentricity cycle 119 at 48.0 Ma, which also represents a 2.4 myr eccentricity minimum (Fig. S11). Records presented here reconnect to astrochronologies that cover the Eocene cyclostratigraphic gap (Westerhold et al., 2015) from Site 1263.
Two enigmas had to be solved before a final stable 405 kyr cyclostratigraphy
was set up. First, the question of whether two or three 405 kyr cycles are
present at Site 1263 in the critical interval from 254 to 265 rmcd of Site 1263 needed to be answered.
Second, the question of which orbital solution is appropriate for more detailed orbital
tuning on the short eccentricity level needed to be answered. The first issue is complicated by two
The eccentricity-modulated precession cycles at ODP Site 1258 can help to
test the effects of different numbers of 405 kyr cycles in the interval from
68 to 95 rmcd spanning Chron C23 (Fig. S14). The thicknesses of high-frequency cycles at Site 1258 change from
The principal terms for the precession of the Earth are given by the
combination of the fundamental secular frequencies (
Correlation ties for astronomical calibration from 54.5 to 46 Ma. The top
panel shows the numerical solutions La2011 (Laskar et al., 2011b) and
La2010a-d (Laskar et al., 2011a), including the 405 kyr cycle number counted
backwards from today (Wade and Pälike, 2004). Below the Site 1267 core
image, which shows the best expressed dark layer pattern of all sites, we
plotted the detrended benthic (black) and bulk (grey) combined
After a stable 405 kyr cyclostratigraphic framework is established, the orbital age model can be refined by tuning the carbon isotope data to an orbital solution on the short eccentricity level. Uncertainties in the ephemeris used to construct the orbital solutions currently limits their accuracy to roughly 48–50 Ma (Laskar et al., 2011a, b; Westerhold et al., 2015). Going beyond 50 Ma, the modulation pattern of short eccentricity recorded in the geological data can help to find the correct orbital solution (Laskar et al., 2004, 2011a, b). In particular, knowledge of the exact positions of the very long eccentricity minima, primary anchor for accurate orbital tuning (Shackleton and Crowhurst, 1997; Westerhold et al., 2007; Zeeden et al., 2014), can help to constrain astronomical solutions (Laskar et al., 2004; Westerhold et al., 2012). Beyond an age of 50 Ma, the position of the very long eccentricity nodes in available orbital solutions (La2004 – Laskar et al., 2004; La2010 – Laskar et al., 2011a; La2011 – Laskar et al., 2011b) are much more uncertain. Tuning to the La2010 or La2011 solution on the 100 kyr level is possible on the basis of a 405 kyr cyclostratigraphy but should be evaluated with great care. Before a bulk stable isotope record for Site 1258 and a benthic stable isotope record for Site 1263 were generated, only Site 1258 XRF Fe data provided a record sufficient for attempting astronomical tuning of the early Eocene (Westerhold and Röhl, 2009). Here we use a larger and more diverse data set from multiple sites on a stable 405 kyr cyclostratigraphy to test which orbital solution is the most appropriate one for detailed tuning.
The compiled records from the Ypresian, in particular the XRF core scanning Fe intensity data, show prominent minima in eccentricity-related modulation of the data in the intervals 212–220, 240–245, 260–267, 277–285, and 297–307 rmcd of Site 1263 (Figs. 3, S7). Starting with the first common node in the 2.4 myr cycle of La2010 and La2011 at 405 kyr cycles 118 to 119 (47.5–48.0 Ma), we go back in time and compare the position of the very long eccentricity cycle minima to the data amplitude minima of the Ypresian records. Correlating the modulation minima at 212–220 rmcd of Site 1263 to the node at 405 kyr cycles 118 to 119 anchors the records to the astronomically tuned middle to late Eocene timescale (Westerhold et al., 2015). With the application of the stable 405 kyr framework introduced above, the preceding data modulation minima at 240–245, 260–267, 277–285, and 297–307 rmcd of Site 1263 require very long eccentricity minima at 405 kyr cycles 123–124, 128–129, 132, and 135–136 (Fig. 3). The first three minima can be observed in the different orbital solutions, suggesting that basically all the solutions back to 52 Ma could be used as target curves for tuning. Beyond 52 Ma, only the La2010b and La2010c solutions show a minimum at 405 kyr cycle 132. Going further back in time to 56–57 Ma, the minimum before ETM-2 (Lourens et al., 2005; Westerhold et al., 2007; Meyers, 2015) and the minimum before the PETM (Westerhold et al., 2007; Zachos et al., 2010; Littler et al., 2014) in the data even match very long eccentricity minima in La2010b and La2010c in 405 kyr cycles 135–136 (54.3–54.9 Ma) and 140 (56.4 Ma).
Extraction of the AM using statistical methods like
those implemented in the astrochron package (Meyers, 2014) or the ENVELOPE
(Schulz et al., 1999) routine are important for independently testing the
visual recognition of cycle patterns (Hinnov, 2013; Hilgen et al., 2014). AM
analysis on XRF core data using the ENVELOPE routine was applied at ODP Sites
1262 (52–60 Ma) and 1258 (47–54 Ma) records in Westerhold et
al. (2012) in order to search for the very long eccentricity minima.
Meyers (2015) used
Following the approach of Zeeden et al. (2015), we extracted the short
eccentricity cycle (100 kyr) and applied a broad bandpass filter (0.004 to
0.016 cycles kyr
Comparison of the amplitude modulation (AM) of the short eccentricity cycle between the La2004, La2010, and La2011 orbital solutions and Fe intensity data from ODP Sites 1258 (red), 1262 (orange), and 1263 (blue). For the orbital solutions we also plotted the 405 kyr AM. The short eccentricity AM of Sites 1258, 1262, and 1263 Fe intensity data are plotted on the 405 kyr scale model. The very long eccentricity minima are highlighted by light blue bars in the orbital solutions and the Fe intensity data. Statistical and visual recognition of the cycle pattern suggests that the La2010b and La2010c solutions are most consistent with the geological data.
Bio- and magnetostratigraphic data, benthic and bulk stable
isotopes, XRF core scanning Fe intensities, and core images for the new
astrochronology for the Ypresian stage. Inclination data from ChRM analysis
of each site from Leg 208 are shown. For Site 1258 the results of Suganuma
and Ogg (2006) are given as a code between
The position of the very long eccentricity minima in the AM of XRF Fe
intensity data in the interval from 46 to 59 Ma (blue bars in Fig. 4) best
fits with minima in the La2010b and La2010c orbital solutions. In contrast,
the minima do not match minima in the La2004 solution, suggesting that this
solution is not appropriate for testing geological data in this period of
time. The La2010a–d and La2011 solutions fit geological data back to
50 Ma. Beyond 50 Ma these solutions diverge (as discussed in Westerhold et
al., 2012). Only the La2010b and La2010c solutions exhibit very long
eccentricity minima at
Quantitative evidence supporting the correct eccentricity node identification
can also be derived from the emergence of obliquity cycles in the data at the
nodes. Obliquity is not present in the Paleocene and early Eocene (Littler et
al., 2014; Zeebe et al., 2017) of the investigated records. However, Site 1258 Fe
intensity data show some obliquity-related cycles at around 80 rmcd (also
see Fig. S14) and from 55 to 60 rmcd corresponding to the end of the very
long eccentricity node at 52 Ma and the beginning of the very long
eccentricity node at 50 Ma. At another potential node (48 Ma), Site 1258 Fe
data do not clearly exhibit obliquity cycles but rather low-amplitude modulations
of precession-related cyclicity (Westerhold and Röhl, 2009, see Fig. 9
therein). Considering all these observations, they provide some independent
evidence for the existence of eccentricity nodes at 50 and 52 Ma. The nodes
at
Based on our observations and the statistical analysis (Hinnov, 2013; Hilgen et al., 2014), we decided to fine-tune the records to the La2010b orbital solution (Fig. 5). It has to be noted here that there is hardly any difference between the La2010b and La2010c solution in the Ypresian from 46–56 Ma. Therefore, it does not matter if La2010b or La2010c is chosen as a target curve. Consequences of the match between orbital solutions and the geological data as well as the implications of the new age model for magnetostratigraphy need to be discussed. The tie points for the tuned short eccentricity age model are given in Table S47.
Nonlinear response of climate is critical in the Ypresian. Multiple carbon
cycle perturbations are documented as negative carbon isotope excursions
(CIEs) and the dissolution of carbonates at the seafloor, both pointing to
massive releases of
We have established the first complete astrochronology for the entire Ypresian stage (YATS), compiling, integrating, and synthesizing geochemical and bio- and magnetostratigraphic records at unprecedented precision. The result is a complex stratigraphy that can function as a reference, allowing synchronization of paleoclimate records essential to understanding cause and consequences of events in the early Eocene.
Just recently the first geologic evidence confirming the chaotic behavior of
the solar system through the identification of a chaotic resonance transition
during the Coniacian (
With the new cyclostratigraphy based on the stable 405 kyr eccentricity
cycle for the Ypresian, we can test if this switch is present in the
observations. Previously, the identification of the very long eccentricity
cycle in geological data for the Ypresian used the XRF Fe intensities from
Sites 1258 and 1262 only (Westerhold et al., 2007; Westerhold and Röhl,
2009). The multiple proxy data now provide a much clearer picture, as
described in Sect. 4.2, and show a very good match between geological data
and the La2010b and La2010c numerical orbital solutions (Laskar et al.,
2011a) for the Ypresian. The important feature shared between data and models
is the position of the very long eccentricity minima expressed as areas
of low-amplitude modulation in the data itself (Figs. 3, 4, 5). In the
La2010b and La2010c solutions, the transition from libration to circulation
occurs at
Comparison of magnetochron boundary ages in millions of years.
None of the available orbital solutions perfectly fit the geological data. However, it is important that we isolated the transition in the data, which is also present in the La2010b and La2010c solutions. The short eccentricity cycle pattern both in the solutions and the geological data will not match perfectly beyond 50 Ma when the uncertainty in the solutions increase (as discussed in Westerhold et al., 2012). Still, the geological data and La2010b/c solutions are very similar from 53.5 Ma to the PETM. In the interval from 51 to 52 Ma, the most difficult part to tune in the Ypresian, multiple hyperthermal events and the shift in carbon isotope data make a direct comparison much more difficult. It has to be noted here that the eccentricity solutions from La2010b/c might not be completely reliable in this interval. Despite the uncertainties, we provide a tuned age model to La2010b/c because the match in the 53.5 Ma to the PETM interval is good enough to do so. If in doubt, the provided 405 kyr age model can still be used.
The point in time when the transition occurs in the numerical solutions is
sensitive to the initial conditions of the planetary ephemeris used for
back calculation of the planetary motions. The initial conditions depend on
the accuracy of the observational data used to make a least squares fit of the
model to the data. The La2010b/c solutions used the INPOP08 Ephemeris (Fienga
et al., 2009; Laskar et al., 2011a). In contrast, the La2010d solution used
the INPOP06 Ephemeris (Fienga et al., 2008) and the La2011 solution the INPOP10a
ephemeris (Fienga et al., 2011). The very similar long-term behavior of
INPOP06 and INPOP10a lead to the conclusion that these ephemerides are more
stable than INPOP08 (Laskar et al., 2011b; Westerhold et al., 2012). Although
the INPOP10a ephemeris is considered to be more accurate than INPOP08 (Fienga
et al., 2011), the geological data provide evidence that the latter is closer
to reality. Identifying the transition from libration to circulation at
Our results support the application of La2010b or La2010c solutions for eccentricity to construct astronomical age models back to 60 Ma. Beyond 60 Ma an accurate solution for eccentricity is not possible at the moment (Laskar et al., 2011b), but the stable 405 kyr cycles will still provide a good target to establish astrochronologies (Laskar et al., 2004, 2011a, b). Here again, the geological data should be examined to find the very long eccentricity minima in very early Cenozoic and Mesozoic strata (Meyers, 2015) and provide a landmark for developing more precise orbital solutions.
Combining the new astrochronology with the revised magnetostratigraphy for
the Ypresian allows us to consider the significance of the abrupt global
increase in spreading rates in Chron C23n.2n, which is also known as the
50 Ma discrepancy on the Paleogene timescale (Vandenberghe et al.,
2012). The unusual peak in spreading rates in the South Atlantic (Fig. 6a) is
independent of the age model used for the magnetostratigraphic interpretation
of Site 1258 (Westerhold and Röhl, 2009; Westerhold et al., 2012;
Lauretano et al., 2016). The new multi-site magnetostratigraphic data from
Leg 208 sites reveal that Chron C23n is too short in the Site 1258
magnetostratigraphic interpretation (Suganuma and Ogg, 2006) and the likely
reason for the computed peak in spreading rates. Application of the tuned age
model to the integrated magnetostratigraphy results in a moderate but
distinct jump in spreading rates at the C23n.2n–C23r reversal from 12 to
19 km myr
Comparison of magnetochron boundary durations in millions of years.
Astronomical calibration of the refined magnetostratigraphy in the marine records resulted in an improved GPTS for the Ypresian (Fig. 6b, Table 1). Duration of polarity zones are now consistent with the reversal thickness relationships observed in the South Atlantic and within error for the mean width of magnetic anomalies as published in Table 4 of Cande and Kent (1992) (Tables 2 and S46). The improved magnetostratigraphy shows a 376 kyr shorter duration for C22r, a 335 kyr longer duration for C23n.2n, and a 283 kyr shorter duration for C24n compared to previous marine records (Westerhold et al., 2015). The 50 Ma discrepancy in seafloor spreading rates is now eliminated. It was clearly the effect of the difficult and incomplete identification of Chron C23n at Site 1258. Moreover, the durations are consistent within error to the GPTS2012 (Vandenberghe et al., 2012) except for Chron C20r, which is difficult to access due to the relatively large error for the radioisotopic ages of the Mission Valley and the Montanari ash (Fig. 6) (for discussion see Westerhold et al., 2015). At this point more precise estimates of the mean width of magnetic anomalies and their error are required to be able to evaluate and improve the GPTS in the late Eocene. However, comparison to the GPTS models from terrestrial successions corroborates the finding (Westerhold et al., 2015) that the model of Tsukui and Clyde (2012) more closely resembles the marine GPTS than the model of Smith et al. (2010, 2014) (Fig. 6). Issues in the correlation of the Layered tuff, Sixth tuff, and Main tuff to local magnetostratigraphic records in the terrestrial records from the Green River Formation need to be resolved (see Tsukui and Clyde, 2012) to understand the current discrepancies in various terrestrial and marine GPTS models.
Overview of naming and age of hyperthermal events.
The GSSP of the Ypresian, which also marks the Paleocene–Eocene boundary, is
defined at the basal inflection of the CIE of the
PETM (Aubry et al., 2007), about two-thirds of the way down in magnetochron C24r (Westerhold et al., 2007) at the base
of Zone CNE1 where the top of the calcareous nannofossil
The top of the Ypresian stage (or base of the Lutetian) is defined at the
lowest occurrence of the calcareous nannofossil
The Ypresian stage is of special interest to recent scientific work because it allows the study of climate dynamics and feedbacks in a warm world (Zachos et al., 2008). In particular, the occurrence of multiple transient global warming events (hyperthermals) could help tremendously to understand the response of the climate system to a massive release of carbon to the ocean–atmosphere system (Dickens, 2003; Zachos et al., 2008; Lunt et al., 2011; Kirtland Turner et al., 2014). A paired negative excursion in the carbon and oxygen isotope composition of bulk sediment and benthic foraminifera associated with a more clay-rich layer, indicating dissolution of carbonate, are the characteristics of the deep-marine hyperthermal events (Zachos et al., 2005; Lourens et al., 2005; Leon-Rodriguez and Dickens, 2010). The early Eocene hyperthermals are paced by Earth's orbital eccentricity, except for the PETM (Zachos et al., 2010; Sexton et al., 2011; Littler et al., 2014; Lauretano et al., 2015, 2016; Laurin et al., 2016). After the discovery of a large number of early Eocene hyperthermal events, a magnetochron-based naming scheme was introduced (Sexton et al., 2011; Kirtland Turner et al., 2014). Because the scheme used the inaccurate magnetostratigraphy of Site 1258 for Chrons C21 to C24, we updated the naming scheme of Kirtland Turner et al. (2014). We also maintain the labeling system of Cramer et al. (2003), which was extended by Lauretano et al. (2016) for consistency. In Table 3 we provide an overview of the naming schemes, astronomical age, and position with respect to magnetostratigraphy.
The most important characteristic used to identify hyperthermal events is a
paired excursion in
A new complex cyclostratigraphy and refined bio-, chemo-, and magnetostratigraphy
have
been developed for key ODP records spanning the entire Ypresian stage from 56
to 47 Ma. Detailed correlation of ODP Sites 1258, 1262, 1263, 1265, and 1267
using the new CODD macros software tool revealed a 3–400 kyr condensed
interval at the Leg 208 sites during Chron C22n. New characteristic remanent
magnetization data from four Leg 208 sites show an overall consistent
magnetostratigraphy refining the Ypresian geomagnetic polarity timescale.
Multi-site ChRM data correlated on a centimeter to decimeter scale suggest that a
magnetostratigraphic record from a single site might contain significant
errors due to coring disturbance. Cyclic variations in synthesized XRF core
scanning and stable isotope data as well as lithological changes apparent in
core images have been successfully used to refine previous astrochronologies
and construct the first complete Ypresian astronomical timescale. In
the absence of independent, high-precision time-control-like radioisotopic dates
in Leg 207 and 208 sediments, our study clearly validates that it is crucial
to combine multiple records from multiple regions to help safeguard against
incompleteness that is otherwise difficult to assess both qualitatively and
quantitatively. The YATS not only provides updated absolute ages for bio- and
magnetostratigraphy but also a comprehensive list of the early Eocene
hyperthermal events. The new astronomically calibrated Ypresian GPTS resolves
the 50 Ma discrepancy, which was primarily caused by the imprecise
magnetostratigraphy of Site 1258. Comparing the eccentricity-related cyclic
pattern in XRF core scanning and stable carbon isotope data to numerical
orbital solutions suggests that the transition from libration to circulation
as predicted by the La2010b solution occurred
The data reported in this paper are tabulated in the
Supplement and archived in the PANGAEA
(
The authors declare that they have no conflict of interest.
We thank Henning Kuhnert and his team for stable isotope analyses at MARUM, Alex Wülbers and Walter Hale at the IODP Bremen Core Repository (BCR) for core handling, and Vera Lukies (MARUM) for assistance with XRF core scanning. We thank Frits Hilgen, Stephen R. Meyers, and the anonymous referee for their constructive critical reviews. This research used samples and data provided by the International Ocean Discovery Program (IODP). IODP is sponsored by the US National Science Foundation (NSF) and participating countries. Financial support for this research was provided by the Deutsche Forschungsgemeinschaft (DFG) and the National Science Foundation (NSF). The article processing charges for this open-access publication were covered by the University of Bremen.Edited by: Appy Sluijs Reviewed by: Frederik Hilgen, Stepen Meyers, and one anonymous referee