Our new high-resolution stable isotope, XRF, and %CF records for Site 1209 are the most detailed and complete data sets generated to date for the Paleocene. Combining the orbital chronology with the detailed stable isotope record affords a precise chronology of the major global/regional trends and events of the Paleocene at unprecedented resolution and allows us to evaluate the relation between climatic and astronomical cycles.
4.1. Key Trends in Paleocene Benthic Stable Isotope Records
 The Site 1209 benthic stable isotope records (Figure 3) show the well-known long-term patterns established from numerous lower-resolution records [e.g., Shackleton et al., 1985; Charisi and Schmitz, 1995; Zachos et al., 2001a; Hollis et al., 2005]. The 3°C warming and subsequent 1.5°C cooling during a 500 kyr interval in the late Maastricht just prior to the K/Pg boundary has been observed in numerous deep-sea records throughout the oceans including the North Atlantic, South Atlantic, Tropical Pacific and Indian Ocean [Stott and Kennett, 1990; Li and Keller, 1998; Barrera and Savin, 1999]. The warming is relatively fast and has been attributed to a rise in atmospheric pCO2 driven by a second phase of the Deccan Trap volcanism centered at the C30n/C29r magnetochron boundary [Caldeira and Rampino, 1990; Li and Keller, 1998; Nordt et al., 2002; Ravizza and Peucker-Ehrenbrink, 2003; Wilf et al., 2003; Chenet et al., 2007; Robinson et al., 2009; Thibault and Gardin, 2010]. Immediately after the K/Pg boundary, from 65.3 to 64.8 Ma, deep-sea temperature increased gradually by 2°C. Decreasing δ13C suggests that the warming could have been driven by the 3rd phase of the Deccan Trap volcanism [Knight et al., 2003; Baksi, 2005; Knight et al., 2005; Chenet et al., 2007; Chenet et al., 2008; Chenet et al., 2009] which occurred after the K/Pg impact till the mid of magnetochron C29n. In contrast, modeling results suggest that flood basalt volcanism might not in itself emit carbon at a sufficient rate to cause noticeable warming, particularly during the early Paleocene [Caldeira and Rampino, 1990; Self et al., 2006]. Alternatively, diminished CO2 drawn down by a reduction in oceanic export production for ∼0.5 kyr after the K/Pg impact might have contributed to warming [Hsü and McKenzie, 1985; Zachos et al., 1989; D'Hondt et al., 1998].
Figure 3. (top) Paleocene composite multisite deep-sea benthic foraminiferal δ18O and δ13C records against age (option 1 of Westerhold et al. ). Black line is the five-point moving average of Site 1209 benthic isotope data from this study. Note that for benthic isotopes all δ13C data are corrected to N. truempyi [Katz et al., 2003], all δ18O data are corrected to N. truempyi and adjusted +0.45‰ [Shackleton et al., 1984]. Paleotemperature after Erez and Luz ; −1.2‰ ice-free SMOW. Primary components of North Atlantic Igneous Province (NAIP) and Deccan Trap volcanism are adapted from Sinton and Duncan  and Chenet et al. , respectively. Data source are Sites 384, 527, 550, 577, 689, 690, 702, 738, 865, 1209 from Cramer et al.  compilation; Site 215 from Zachos et al. [2001a] compilation; Site 761 from Quillévéré et al. ; Site 1049 from Quillévéré et al. ; and Sites 1262 and 1263 from McCarren et al. . (bottom) The paleogeographic reconstruction for the middle Paleocene (60 Ma) shows the location of Shatsky Rise ODP Site 1209 and different DSDP and ODP sites that make up the Paleocene portion of the stacked composite record (see Figure 1).
Download figure to PowerPoint
 The Dan-C2 isotope excursion which has been reported to occur near the top of C29r in a benthic δ13C record from a single Atlantic Ocean hole (1049C) [Quillévéré et al., 2008] is not present in our high-resolution benthic δ13C record from Shatsky Rise. The characteristic double peak in δ13C of bulk sediment, however, has been found at sites from Walvis Ridge (DSDP 527 and 528). Bulk sediment carbon and oxygen isotope records from the western Tethyan Contessa Highway section (Gubbio, Italy) also show the Dan-C2 event [Coccioni et al., 2010]. Either this event was restricted to the Atlantic and is thus not manifested in the deep equatorial Pacific, or Site 1209 contains an undetected stratigraphic gap or condensed interval. There is, however, no obvious evidence of the latter.
 Several records indicate a global change in climate and paleoceanographic conditions at 63.0 Ma. This includes the onset of a long-term global cooling trend, a systematic increase in sedimentation rates and decreasing Fe intensity or increasing carbonate content. The increase in both Atlantic and Pacific carbonate sedimentation rates [Westerhold et al., 2008] suggests higher open ocean carbonate production and/or preservation, roughly coinciding with the second stage of the post-K/Pg recovery of the open ocean carbon cycle [D'Hondt et al., 1998; Adams et al., 2004]. Carbon isotopes, on the other hand, show only a minor response at this time consistent with relatively minor changes in overall carbon fluxes.
 The Site 1209 record indicates that the cooling trend which initiated at 63.0 Ma and continued to the base of the ELPE event at 58.5 Ma, was temporarily interrupted at 61.75 Ma by a transient warming of more than 2°C near the termination of Chron C27n, which we refer to as the Top Chron 27n event [Westerhold et al., 2008]. This warming also marks the onset of the Paleocene carbon isotope maximum (PCIM) [e.g., Shackleton et al., 1984] which is thought to represent massive burial of organic carbon caused by enhanced marine productivity or increased burial of terrestrial biomass [Shackleton, 1986; Corfield and Norris, 1996; Thompson and Schmitz, 1997; Kurtz et al., 2003; Hilting et al., 2008]. The isotope record across the Early Late Paleocene Event (ELPE) [Röhl et al., 2004; Petrizzo, 2005; Bralower et al., 2006] or Mid Paleocene Biotic Event (MPBE) [Bernaola et al., 2007] is sparse due to strong dissolution. The few data points are insufficient to determine whether or not the ELPE was a global hyperthermal-like event [see Bernaola et al., 2007].
 The PCIM is punctuated by several rapid, but small shifts in the isotope record. At 58.2 Ma, δ13C and δ18O decrease in 200 kyr consistent with a gradual input of light carbon accompanied by a small rise in temperature. At 57.7 Ma a rapid positive shift of 0.5‰ in δ13C data is accompanied by a decrease of 0.4‰ in δ18O, roughly coincident with a prominent rise in carbonate sedimentation rates in both the South Atlantic and in the Pacific [Westerhold et al., 2008]. The positive shift in δ13C is fast and might be more consistent with a ventilation change that results in enhanced carbonate preservation at Site 1209. Other records are required to determine if the signal is global, in which case it could reflect a relatively rapid increase in burial rates of organic carbon countered by an increase in carbonate preservation and burial [Kump and Arthur, 1999].
 The long-term 3°C warming and decline in δ13C (∼1‰) that precedes the PETM essentially initiates at 57.3 Ma with a rapid warming of 1.5°C. The decrease might have resulted from a long-term input of 12C-enriched carbon into the atmosphere and ocean. The second phase of the North Atlantic Igneous Province (NAIP) initiated around 57.5 Ma [Sinton and Duncan, 1998]. Presumably, the increased crustal production was accompanied by increased mantle outgassing of CO2 and thus likely sustained the long-term warming trend that began at ∼57.5 Ma and lasted into the early Eocene [Vogt, 1979; Arthur, 1980; Corfield and Cartlidge, 1992; Zachos et al., 2001a; Demicco, 2004]. This gradual trend was briefly interrupted at the end of the Paleocene by the strong carbon isotope anomaly and warming of the PETM [e.g., Zachos et al., 2008].
4.2. Orbital Cyclicity in the Paleocene Benthic Isotope Record
 Our new data show that the deep-sea benthic isotope record is characterized by periodic cycles or excursions that are superimposed on the long-term trend. In general, carbon isotope minima are associated with oxygen isotope minima and increased dissolution, indicated by peaks in Fe intensity as well as minima in the coarse fraction (Figure 2). Carbon isotope excursions are related to the short (100 kyr) and long (405 kyr) eccentricity cycle with maxima in eccentricity being associated with carbon isotope minima (Figure 4). Wavelet analysis of benthic δ13C, benthic δ18O and coarse fraction supports the dominance of the 405 kyr cycle in the benthic Paleocene record. (Figures S5, S6, and S7). Precession and obliquity related variations in the data cannot be detected with confidence, because low sedimentation rate and bioturbation smooth signals over 10 to 25 kyr. Cross-spectral analysis (Figure 4c) between eccentricity and isotope data reveals a 40–60 kyr phase lag of δ13C and a 0–50 kyr phase lag of δ18O for long eccentricity cycles. For short eccentricity cycles we found a roughly 10 kyr lag of δ13C and almost no lag of δ18O. This frequency-dependent phase lag of δ13C has been detected previously and is attributed to a nonlinear energy transfer from precession to eccentricity bands due to the long residence time of carbon in the oceans [see Pälike et al., 2006]. A similar interaction of the carbon cycle, solar forcing, and oxygen isotope variations has been documented for the Oligocene and Miocene [Zachos et al., 2001b; Holbourn et al., 2007] as well as the Paleocene and late Cretaceous [Herbert, 1997; Zachos et al., 2010].
Figure 4. Spectral analysis of Site 1209 benthic δ13C and δ18O data. (a) Benthic δ13C and δ18O data from Site 1209 (55–66 Ma, black line) and bulk δ13C data from Site 1262 (Walvis Ridge, South Atlantic, 55–57.5 Ma, gray line; data from Zachos et al. ). The 405 kyr filter output for Site 1209 data is dashed in intervals with sparse data coverage. The software package AnalySeries 2.0 [Paillard et al., 1996] was used to filter the time series for variability at a period of 405 kyr (0.024 ± 0.0007). For comparison, we plotted the Laskar et al.  solution for eccentricity and the stable 405 kyr cycle [Laskar et al., 2004, equation 47]. Note that the dashed gray grid lines are spaced by 405 kyr and that the original bulk δ13C data from Site 1262 have been offset by 1‰ to be plotted on the benthic δ13C y axis. (b) MTM power spectra for Site 1209 benthic δ13C and δ18O data. The spectra have been calculated by the kSpectra Toolkit using three tapers and a 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]. Spectral peaks related to the short (100 kyr) and long (405 kyr) eccentricity cycles are marked by an arrow. In preparation for analysis outliers were removed and the time series linearly resampled at 2 kyr intervals. The long-term trend was removed by a 2.5 Myr (0.0004 Hz) FFT high-pass filter. (c) Blackman-Tukey cross-spectral phase estimates between stable isotopes and the La2004 eccentricity curve, converted to lag times in kyr (with 99% confidence intervals).
Download figure to PowerPoint
 For a short interval at Shatsky Rise (230–234 rmcd; about 9 eccentricity cycles) helium isotope data [Marcantonio et al., 2009] and dust flux estimates [Woodard et al., 2011] show that changes in eolian dust flux are synchronous with Fe intensity data and thus likely eccentricity driven. In addition, the extraterrestrial 3He data from Site 1209 point to climate-forced changes in carbonate preservation anti-phased to the dust flux [Marcantonio et al., 2009]. Coarse fraction data presented here document enhanced preservation of calcium carbonate in eccentricity minima, supporting the helium isotope results. Our data from the central Pacific, along with the new bulk isotope data from the South Atlantic [Zachos et al., 2010], confirm the importance of eccentricity oscillations in driving Paleocene climate variations. Simple box modeling using parameterized forcing suggests that the climate system responses to complex orbital variations via expansion and contraction of biosphere productivity [Pälike et al., 2006]. The pronounced 100- and 405-kyr periodicity observed in the data might be a result of the long residence time of carbon in the ocean (∼100 kyr) which acts like a low-pass filter that dampens the signal, particularly at high frequencies [Cramer et al., 2003]. It is not known if the climatic response in the Paleocene and Eocene is achieved through ocean export production or carbon accumulation in the ocean or on land [Zachos et al., 2010]. The subdued amplitude of the 405-kyr cycles in benthic δ13C in the interval 2–3 Myr after the K/Pg might be consistent with observation that export of carbon to the deep sea had not fully recovered to preextinction levels [Coxall et al., 2006]. This dampening, the condensed intervals and the limited resolution of stable isotope data from 55.5 to 58 Ma in Site 1209 also hamper the clear recognition of the very long 2.4 Myr eccentricity cycle.
4.3. Top Chron C27n Event
 The Site 1209 benthic foraminiferal stable isotope record also provides unprecedented insight into the climatic processes around the Danian/Seladian boundary, in particular the Top Chron 27n event at 61.75 Ma [Westerhold et al., 2008]. The Top Chron C27n Event has also been identified in detailed cyclostratigraphic records from ODP Leg 208 on Walvis Ridge (South Atlantic) and ODP Leg 199 on Shatsky Rise (Western Pacific) and is characterized by exceptionally high peaks in iron concentration close to the top of magnetochron C27n. The Site 1209 benthic record (Figure 5 and Figure S4) shows that the event is characterized by a double peak in carbon isotopes with a total duration of 200 kyr, a feature common to subsequent early Eocene hyperthermals [Zachos et al., 2010]. The onset is transient with a 0.6‰ decrease in δ13C and a 0.5‰ decrease in δ18O (2°C increase in deep-sea temperature) right at the Chron C27n/C26r boundary and planktic foraminiferal zone P3a/P3b boundary. The sudden decrease in carbonate in combination with a clear negative excursion in δ13C and δ18O indicates that this event may represent an early Paleocene global warming episode. This event clearly corresponds to the Latest Danian Event (LDE) of Bornemann et al. , the Danian/Selandian transition event of Speijer  (Neo-Duwi event), and the carbon isotope excursions CIE-DS1 of Arenillas et al.  at Zumaia as well as ODP 761B off NW Australia [Quillévéré et al., 2002] (Figure 5). New paleomagnetic data from Zumaia show that the onset of the transient δ13C excursion CIE-DS1 occurred exactly at the C27n/C26r magnetochron boundary [Dinarès-Turell et al., 2010], identical to the data from Shatsky Rise. The characteristic double peak in carbon isotopes is also present in bulk δ13C data from Walvis Ridge Site 1262 (at 195 mcd in the work by Kroon et al. [2007, Figure F11]). Our isotope data as well as micropaleontological work at other sections [Speijer, 2003; Youssef Ali, 2009] suggests that this event represents an exceptional perturbation of the carbon cycle associated with a 2°C warming of the deep ocean.
Figure 5. Benthic foraminiferal δ13C and δ18O and stratigraphy from ODP Site 1209, Gebel Aweina in the Nile Basin (Egypt) [Bornemann et al., 2009], and ODP Site 761B (offshore NW Australia) [Quillévéré et al., 2002] and bulk rock δ13C from Zumaia (Spain) [Arenillas et al., 2008] across the Top Chron 27n event or Latest Danian Event (LDE). Please note that the Danian/Selandian boundary definition at Zumaia is ∼870 kyr above the C27n/C26r magnetochron boundary according to Bernaola et al. ; the Danian/Selandian boundary at Site 1209 is after GTS2004 [Luterbacher et al., 2004], close to the C27n/C26r magnetochron boundary. The peak LDE δ13C and δ18O values are coincident with the base of the planktonic foraminifera zone P3b (Igorina albeari subzone). Blue shaded area represents the correlation between sections for the LDE with duration of 200 kyr. Red lines mark correlation ties between Zumaia and Site 1209.
Download figure to PowerPoint
 The Top Chron C27n event marks the beginning of the long-term increase in δ13C and coincides with the major deceleration of South Atlantic spreading rates indicating a possible causal relationship. Interestingly this event is synchronous with two major land mammal turnovers (Shanghuan-Nongshanian Asian Land Mammal boundary and the Torrejonian-Tiffanian North American Land Mammal boundary) and thus lends support to the idea that the turnover was driven by environmental changes [Clyde et al., 2008]. Although the causal mechanism for the Top Chron 27n event is uncertain, the proximity to the first phase of the NAIP (Figure 3) [Saunders et al., 1996; Larsen and Saunders, 1998; Jolley and Bell, 2002; Pedersen et al., 2002] points to a similar triggering mechanism as proposed for the PETM with initial warming caused by a pulse in volcanic CO2 input and subsequent warming due to positive feedback loops in the climate system [Svensen et al., 2004; Zachos et al., 2005; Sluijs et al., 2007; Storey et al., 2007]. Increased productivity and deep-sea cooling starting more than 1 million years earlier, at around 63.0 Ma, should have favored the expansion of reduced carbon reservoirs (e.g., gas hydrates), one or more of which then rapidly released excess carbon in response to initial warming.[Dickens et al., 1995; Dickens, 2003]. While there are other subsequent thermal excursions, for example at 61.35 Ma, none are associated with a prominent carbon isotope excursion and dissolution horizon features that imply tight coupling between climate and the carbon cycle.