Apparent Diachroneity of Calcareous Nannofossil Datums During the Early Eocene in the High‐Latitude South Pacific Ocean

The late Paleocene to early Eocene interval is characterized by a series of carbon perturbations that caused transient warming (hyperthermal) events, of which the Paleocene‐Eocene Thermal Maximum (PETM) was the largest. These hyperthermals can be recognized in the pelagic sedimentary record as paired negative δ13C and δ18O excursions, in addition to decreased calcium carbonate and increased iron content caused by carbonate dissolution. However, current data are predominantly sourced from the equatorial‐to subequatorial regions. Here we present a new high‐latitude late Paleocene—early Eocene record, recovered during International Ocean Discovery Program (IODP) Expedition 378 on the Campbell Plateau off New Zealand, in the southwest Pacific Ocean. To construct an age model, we correlated our chemostratigraphic and biostratigraphic data to existing astronomically‐tuned age models from Walvis Ridge (South Atlantic Ocean) and Demerara Rise (equatorial Atlantic Ocean). Our results indicate that the Site U1553 composite section spans ∼7 million years of the latest Paleocene to early Eocene (50.5–57.5 Ma), and preserves many of the early Eocene hyperthermals; including a PETM interval that is more expanded than elsewhere in this region. However, construction of the age model also revealed discrepancies between the chemostratigraphic and biostratigraphic tie points used for correlation. This is likely due to latitudinal diachroneity in the calcareous nannofossil biostratigraphic datums, which are primarily based on low‐to mid‐latitude assemblages. Therefore, our study highlights the need to establish a revised calcareous nannofossil biozonation that is more appropriate for high‐latitude age models.

Since the discovery of the PETM, numerous studies have reported similar but less pronounced paired negative shifts in δ 13 C and δ 18 O during the late Paleocene and early Eocene in both outcrop sections (e.g., Agnini et al., 2009;Coccioni et al., 2012;Galeotti et al., 2010;Slotnick et al., 2015) and within deep sea sediment cores (e.g., Cramer et al., 2003;Kirtland Turner et al., 2014;Lauretano et al., 2015;Littler et al., 2014;Stap et al., 2010;Westerhold et al., 2007;Westerhold et al., 2017;Zachos et al., 2010).These hyperthermals have been labeled from A to L (Cramer et al., 2003) and M to W (Lauretano et al., 2016), and can also be recognized as peaks in X-Ray fluorescence (XRF)-derived iron intensity data caused by carbonate dissolution (e.g., Röhl et al., 2000Röhl et al., , 2007;;Westerhold, Röhl, Donner, & Zachos, 2018;Zachos et al., 2010).More recently, labels based on the position of these events relative to magnetostratigraphy have also been assigned (Kirtland Turner et al., 2014;Westerhold et al., 2017).Although these smaller hyperthermals are relatively understudied compared to the PETM, previous research indicates that significant paleoceanographic and paleoecological changes also occurred during these warming events (e.g., Agnini et al., 2016;D'Onofrio et al., 2016;Gibbs et al., 2012;Harper et al., 2020;Jennions et al., 2015;Luciani et al., 2016;Thomas et al., 2018), and therefore can provide valuable information as to how the earth-ocean system responds to different magnitudes of carbon perturbation.However, before such analyses can be conducted, it is essential to have a robust age model that can be reliably used to constrain the pacing and timing of such changes.
Currently, our best astronomically-calibrated Paleogene age models are restricted to low-and mid-latitude sites such as Demerara Rise (ODP Leg 207; equatorial Atlantic Ocean), Walvis Ridge (ODP Leg 208; mid-latitude South Atlantic Ocean), and Shatsky Rise (ODP Leg 198; subtropical Pacific Ocean) (e.g., Barnet et al., 2019;Littler et al., 2014;Röhl et al., 2007;Westerhold et al., 2017;Westerhold, Röhl, Donner, & Zachos, 2018;Westerhold et al., 2020;Zachos et al., 2010).In comparison, higher latitude early Paleogene age models are scarce, although do exist for Maud Rise in the Atlantic sector of the Southern Ocean (ODP Sites 689 and 690; Bains et al., 1999;Cramer et al., 2003;Kennett & Stott, 1991;Röhl et al., 2007;Thomas et al., 1990), for the Kerguelen Plateau (ODP Sites 738 and 1135; Jiang & Wise, 2009) and the Mantelle Basin (International Ocean Discovery Program (IODP) Site U1514; Vahlenkamp et al., 2020) in the southern Indian Ocean, and for the Campbell Plateau in the South Pacific Ocean (Deep Sea Drilling Program (DSDP) 277; Hollis, 1997;Hollis et al., 2015;Shepherd et al., 2021).However, almost all of these age models are based on δ 13 C data combined with low-resolution, shipboard biostratigraphic datums that may be unreliable.Furthermore, sediment cores from ODP Site 1135 and DSDP Site 277 were both spot-cored and do not provide a continuous record of the late Paleocene and early Eocene.IODP Site U1514 is the only high-latitude site with an astronomically tuned age model; however, this is based only on high-resolution XRF data as the carbon isotope data is of a very low-resolution (Vahlenkamp et al., 2020).In addition, some of the shipboard biostratigraphic datums from this site are inconsistent with the chemostratigraphically based age model interpretation.This highlights the need to use a combination of geochemical and biostratigraphic data to construct more reliable age models at the high-latitudes, so that the pacing and timing of paleoceanographic and paleoecological changes during the early Eocene hyperthermals can be better constrained.
For this reason, herein, we present a new chemostratigraphic and biostratigraphic record from the high-latitude southwest Pacific Ocean, spanning ∼7 million years of the late Paleocene to early .The resultant age model allows us to evaluate the reliability of calcareous nannofossil biohorizons on a global scale.Furthermore, the new stable isotope record combined with our age model have the potential to provide new insights into high-latitude climate responses during carbon cycle perturbations, which will further our knowledge of early Paleogene paleoceanographic conditions.et al., 2022a).This new site was drilled very close to legacy DSDP Site 277, which consists of only one spot-cored hole with relatively low core recovery (Kennett et al., 1975;Shackleton & Kennett, 1975).

Site Location
The late Paleocene to early Eocene sediments at Site U1553 were recovered from rotary-drilled Holes C and D between 409.21 and 461.7 m below sea-floor (mbsf), consisting of the lithological Subunit IIIb and lithological Unit IV (Röhl et al., 2022b).Lithological Subunit IIIb represents the top ∼10 m of the study interval and consists of white, homogenous nannofossil chalk with foraminifera.The rest of the study interval is represented by lithological Unit IV, which is predominantly composed of limestones and is distinguished from Subunit IIIb by the increased lithification and variable degree of bioturbation (Röhl et al., 2022b).These sediments were deposited at high paleolatitudes (∼59-61°S; Figure 1) in lower to middle bathyal water depths based on the benthic foraminiferal assemblages (∼1,000 m; Hollis, 1997).The PETM interval at Site U1553 was first identified offshore within lithological Unit IV based on the nannoplankton assemblages (Röhl et al., 2022b).At the onset of the PETM, there is an abrupt switch from white, non-bioturbated limestones to darker green, bioturbated limestones (Röhl et al., 2022b).Unlike at many other PETM sites there is no obvious dissolution horizon (e.g., Bralower et al., 2014;Zachos et al., 2005) indicating that sedimentation occurred above the calcite compensation depth (CCD) during the entire study interval.
High-resolution shipboard XRF core scanning data generated at the Gulf Coast Core Repository (GCR) was previously used to make a composite section (splice) of Holes U1553C and -D (Drury et al., 2022;Wilkens et al., 2022).The splice shows that the record from 440 to 496 m core composite depth below seafloor (CCSF) is relatively continuous, with minimal core recovery gaps (Figure 2a).This is particularly true of the interval between 450 and 491 m CCSF, where core recovery is generally >83%: exceptional for a rotary-drilled hole.In comparison, the lowest and highest parts of the splice (below 491 m CCSF and above 450 m CCSF respectively) have poorer core recovery (35%-45%), due to the presence of more lithified limestones and/or cherts (Röhl et al., 2022b).Despite these challenges, our composite section still provides one of the most complete records of the early Paleogene from the high-latitude South Pacific to date.

Core-Images and XRF Data
Line-scanned core images and XRF shipboard data (Röhl et al., 2022b) can be used to identify hyperthermal events, which are reported to contain darker, carbonate-depleted sediment with higher XRF-Fe values (Röhl  (Müller et al., 2018) by using the integrated combined rotation file (Cao et al., 2020;Müller et al., 2019;Torsvik et al., 2019;Young et al., 2019Young et al., ). et al., 2000;;Westerhold et al., 2017;Zachos et al., 2005).To generate the core splice images of Holes C and D, we used the software tool Code for Ocean Drilling Data (Wilkens et al., 2017).XRF-Fe counts consist of 2025 data points, with an average spacing of 2.3 cm (Figure 2a).

Stable Isotope Analyses
Paired negative δ 13 C and δ 18 O excursions are arguably the best way to identify the early Eocene hyperthermal events.For this reason, we measured the bulk stable carbon and oxygen isotopes of 454 sediment samples from Holes U1553C and -D.Each sediment sample was dried, finely ground by hand with an agate mortar and pestle, and analyzed in the Stable Isotope Laboratory at MARUM-Center for Marine Environmental Sciences, University of Bremen on a Finnigan 253plus gas isotope ratio mass spectrometer with a Kiel IV automated carbonate preparation device.Data are reported in delta-notation versus Vienna Peedee Belemnite (V-PDB).The instrument was calibrated against the house standard (ground Solnhofen limestone), which in turn was calibrated against the NBS 19 calcite.Over the measurement period, the standard deviation of the house standard was 0.03‰ for δ 13 C and 0.06‰ for δ 18 O.
Bulk stable carbon and oxygen isotope ratios for the same interval were independently analyzed at the Marine Core Research Institute, Kochi University.We analyzed 354 bulk sediment samples from Holes U1553C and -D, which is a different sample set from that analyzed at MARUM.The samples were freeze-dried and finely ground by hand with an agate mortar and pestle.The bulk carbon and oxygen isotopes of the powdered samples were measured using a GV Instruments IsoPrime with a Multicarb preparation system (Wythenshave, Manchester, United Kingdom).The calibration standard material was IAEA-603 (calcite, a reference material certified by International Atomic Energy Agency).The measured isotopic ratios were converted to deltanotation versus V-PDB.Repeated measurements (n = 70) of the IAEA-603 showed that the standard deviations of carbon and oxygen isotope ratios were 0.03 and 0.10‰, respectively.The total of 808 samples across a ∼7 Ma long time interval represents an average sampling resolution of 15,000 to 20,000 years, respectively.

Calcareous Nannofossil Biostratigraphy
For the nannofossil biostratigraphy, we used the biohorizons of Agnini et al. (2007Agnini et al. ( , 2014)), which are based on the abundance patterns of several biostratigraphically-useful taxa.The stratigraphic positions of biohorizons in Holes U1553C and -D were obtained through analyses of the nannofossil assemblages observed in smear slide samples, which were processed using the standard procedures outlined in Bown and Young (1998).80 Samples (52 in Hole U1553C, 28 in Hole U1553D) were analyzed at a ∼30 cm sampling resolution in the intervals straddling the biohorizons, and another 25 samples were checked in the stratigraphic intervals between the biohorizons.Analyses were carried out using a light microscope at 1200× magnification under cross-polarized and transmitted light.The biostratigraphic markers were identified using the taxonomic concepts compiled in the online database Nannotax3 (https://www.mikrotax.org/Nannotax3/),and their occurrences were determined by counting the number of the respective specimens in a prefixed area (N/mm 2 ) (Backman & Shackleton, 1983).We added the taxonomic note "cf." (="confer," in Latin) to Fasciculithus tympaniformis, as some specimens exhibit an anomalous morphology likely caused by overgrowth.Therefore, we are confident about their genus-level-but not their species-level-designations.This is unlikely to affect our interpretations, as the Fasciculithus involutus group (which includes F. tympaniformis) is the last Fasciculithus group to become extinct (e.g., Agnini et al., 2014).

Age Model
In order to construct a high-resolution age model using the full potential of the sediment records from both Holes C and D, we combined data from outside the splice with the in-splice data by applying the revised splice adjustment (Drury et al., 2022;Wilkens et al., 2022).To improve the reliability of the existing shipboard U1553 age model (Röhl et al., 2022b), we combined our new bulk sediment stable isotope data (δ 13 C) with the updated calcareous nannofossil datums.We then correlated our chemostratigraphic record to published astronomically tuned sections from ODP Site 1262 on the Walvis Ridge (Zachos et al., 2010)  (Kirtland Turner et al., 2014).The high-resolution records from both sites allow intuitive correlation with Site U1553 using the amplitude and shape of several carbon isotope excursions (CIEs).Calcareous nannofossil biostratigraphic events were also correlated to those at Site 1262 (Agnini et al., 2007) using the revised age model of Westerhold et al. (2017).

Core Images and Geochemical Results
Core images from Site U1553 (Röhl et al., 2022b) generally show alternations of darker and lighter sediment layers, with thicknesses ranging from centimeters to decimeters (Figure 2a).Darker-colored sediments are more abundant in the lower part of the U1553 composite (late Paleocene to early Eocene section; 462-496 m CCSF), whilst lighter-colored sediments dominate the upper part (440-462 m CCSF).Concurrent with the change in sediment color, XRF-derived Fe content shows increased values in darker sediment layers and lower values in lighter sediment layers.
Bulk sediment δ 13 C values during the late Paleocene to early Eocene interval range between 0.2 and 3.2‰ (Figure 2b).Overall, a long-term, gradual decrease in δ 13 C is punctuated by several negative CIEs.The lower part of the record between 482 and 496 m CCSF is characterized by relatively high δ 13 C values ranging from 2.2 to 3.2‰, followed by the most prominent decrease of ∼2.5‰ between 481 and 483 m CCSF (i.e., during the onset of the PETM).Between 450 and 481 m CCSF, average δ 13 C values are 1.13‰, but show several negative excursions, with a decrease of up to 0.76‰ at 451,454,461,464,468 and 472 m CCSF.The negative excursions at 468 and 472 m CCSF are both followed by a second smaller δ 13 C decrease of less than 0.4‰.The uppermost section (440-443.5m CCSF) of the record is characterized by an overall increase in δ 13 C.

Biostratigraphy
The upper Paleocene to lower Eocene nannofossil biohorizons delineated at Site U1553 are summarized in Table 1, and comprise most of those considered by Agnini et al. (2007Agnini et al. ( , 2014)).However, it was not possible to confidently assign some of the biohorizons due to the poor preservation of the nannofossil assemblages, this being inconsistent between different depth intervals and showing variability from diffuse overgrowth to etching.For example, dissolution-resistant taxa (e.g., Discoaster and Tribrachiatus) are often heavily overgrown, and interspersed among the well-preserved coccoliths of dominant taxa (e.g., Toweius and Chiasmolithus).For this reason, preservation problems hampered the recognition of the lowest occurrence biohorizons related to Tribrachiatus, that is, base (B) Tribrachiatus bramlettei and B Tribrachiatus contortus.
Additional challenges in the designation of biohorizons stemmed from the rarity or inconsistent occurrence of many of the biostratigraphic marker species at the start and/or end of their stratigraphic ranges, as well as the occurrence of some taxa above their previously published last occurrences.For example, the stratigraphic range of Fasciculithus is well documented as becoming extinct shortly after the PETM at many early Eocene sites (e.g., Bralower, 2002;Raffi et al., 2005;Self-Trail et al., 2012;Westerhold et al., 2017).However, it extends to at least ∼20 m above the PETM at Site U1553 (Figure 2c), causing the top (T) of the Fasciculithus tympaniformis biohorizon to appear significantly delayed with respect to its observed position at mid-and low-latitude sites (Agnini et al., 2014).This is largely due to the presence of Fasciculithus specimens with a non-typical morphology (i.e., F. cf.tympaniformis; Figure 3: 13-15), especially in the upper part of its extended range.These anomalous forms are unlikely to be a product of reworking, as reworked specimens of other late Paleocene taxa were not observed.Instead, the prolonged occurrence of Fasciculithus appears to be a primary paleobiological signal, which decreases the reliability of the top Fasciculithus biohorizon at Site U1553.
By comparing the biohorizons between Holes U1553C and -D, we observe that only a few of the events (the base of Discoaster multiradiatus, the top of Tribrachiatus contortus, the tentative top of Fasciculithus cf.tympaniformis and the top of Tribrachiatus orthostylus) are found at the same depth within uncertainty (Figure 2c).However, it is important to note that the oldest (B D. multiradiatus) and youngest (T T. orthostylus) of these events have a higher range of uncertainty than the other biostratigraphic datums.The <2 m inter-hole depth discrepancies in the base of Tribrachiatus orthostylus, top common Discoaster multiradiatus, the base of Sphenolithus radians and the base of Discoaster lodoensis, may simply be a result of variable nannofossil preservation, the rarity of marker species at the beginning of their stratigraphic ranges and/or differences in core recovery and sampling intensity between holes during certain depth intervals.Sphenolithus.The highest observed occurrence (HO) of F. cf.tympaniformis is also shown (the true top of F. cf.tympaniformis expected to be even higher up but is not well constrained herein).Note.The biochronology from several studies is summarized (Agnini et al., 2014;Gradstein et al., 2012;Westerhold et al., 2017).*Shipboard data (Röhl et al., 2022b).

Correlation of Chemostratigraphic Data
The late Paleocene to early Eocene hyperthermal events have been identified worldwide as negative CIEs, which are often coincident with darker sediment layers containing increased XRF-derived Fe intensities due to the dissolution of carbonate (e.g., Lourens et al., 2005;Röhl et Zachos et al., 2010).Although this relationship is observed at Walvis Ridge (Röhl et al., 2007;Westerhold et al., 2007;Zachos et al., 2005), Shatsky Rise and Demerara Rise (Westerhold et al., 2017), the correlation of higher Fe intensities with lower δ 13 C values is not observed for the hyperthermal events at Site U1553.This decoupling between Fe and δ 13 C is due to the shallower depositional water depth of Site U1553 (∼1,221 m; Röhl et al., 2022a) compared to other pelagic sites of the early Eocene (2,387-4,755 m).For this reason, it is very likely that sediment deposition occurred above the CCD at Site U1553 and was not affected by the dissolution of calcium carbonate, even during the hyperthermals.Therefore, unlike in other studies (e.g., Röhl et al., 2007;Westerhold et al., 2007;Westerhold, Röhl, Donner, & Zachos, 2018;Zachos et al., 2005;Zachos et al., 2010), our XRF-derived Fe record cannot be reliably used to construct an age model.
Stable carbon isotope records have also previously been utilized to establish reliably calibrated orbital chronologies during the late Paleocene to early Eocene (Kirtland Turner et al., 2014;Zachos et al., 2010).For this reason, we correlated our bulk sediment stable carbon isotope data with those from two astronomically tuned sites in the Atlantic Ocean: ODP Site 1262 on the Walvis Ridge (Zachos et al., 2010) and ODP Site 1258 on the Demerara Rise (Kirtland Turner et al., 2014).In addition to their orbitally constrained age models (Westerhold et al., 2017), these sites have high-resolution bulk sediment δ 13 C records that allow for the unique recognition and correlation of hyperthermal events.
For the detailed correlation, we compared the amplitude and shape of the different CIEs as well as the overall pattern of the stable isotope record (Figure 4).The PETM is widely known as the largest CIE in the late Paleocene to early Eocene interval.At Site 1262, it is characterized by an abrupt 2.25‰ decrease in δ 13 C (Zachos et al., 2010).In the Site U1553 record, a comparable abrupt δ 13 C decrease occurs at 482 m CCSF (Figure 4), marking our first tie point.The paired hyperthermal events H1, H2 and I1, I2 show the typical double minima (Lauretano et al., 2015) with the first CIE (i.e., H1 and I1) being slightly larger in both cases.We identified these paired excursions at 470-472 m CCSF and 466-468 m CCSF.The J event marks the onset of the Early Eocene Climate Optimum (Westerhold, Röhl, Donner, & Zachos, 2018).In the bulk sediment record of Site 1262, the J event has a similar amplitude to I2, but has a more gradual recovery.We correlate this to the decrease in δ 13 C at 464 m CCSF at Site U1553.The K-event (Thomas et al., 2018) has an amplitude similar to H1, and was found in our record at 461 m CCSF (Figure 4).The L event has a lower amplitude signal compared to the K event.However, the onset of this event appears to be missing in the U1553 record due to a coring gap at around 459 m CCSF.The hyperthermals M and N are characterized by an asymmetric shape at Site 1258 with sharp onsets and gradual recovery, making them recognizable.We correlate these events to the negative carbon excursion at 454 m CCSF (M) and 451 m CCSF (N), respectively.By comparing the amplitudes, the alternation of more positive and negative δ 13 C values and the relative position of the hyperthermal events O, P, Q, R and S in relation to the N event at Site 1258 (Figure 4), we correlate the CIE above the coring gap at 441 m CCSF to the S event.Besides the individual hyperthermal events, there are two prominent shifts in our δ 13 C record: (a) a negative shift before H1 at ∼476 m CCSF and (b) a positive shift of ∼1‰ after the N-event at ∼450 m CCSF.Both of these shifts are observed in the benthic foraminiferal δ 13 C record of Site 1209 (Westerhold, Röhl, Donner, & Zachos, 2018), with the latter also being reported in the benthic Nuttallides truempyi δ 13 C record from Site 1263 (Lauretano et al., 2016).This further strengthens the correlation of the hyperthermal events between sites.The less pronounced CIEs immediately before and after the PETM (D1, D2, E1, E2 and F) could not be identified, most likely because of their low amplitudes (Figure 4).

Age Model Construction
We compared our δ 13 C record to the calcareous nannofossil datums in Table 2.However, doing so revealed multiple discrepancies between the chemo-and biostratigraphic tie points, especially above the PETM recovery (ca.55.5 Ma).The most striking discrepancy is the top occurrence of Fasciculithus cf.tympaniformis, the highest observed occurrence of which is recorded shortly below the K event (∼53 Ma; Figure 3; Figure S1 in Supporting Information S1) at Site U1553 and not during the recovery of the PETM as expected (55.531Ma; Westerhold et al., 2017).Another major discrepancy is observed in the first occurrence of Discoaster lodoensis (53.68 Ma), which should occur prior to the K event (52.85 Ma), but occurs 10 m above it according to our chemostratigraphic correlation (Figure 4).We also observe a delay in the base of Tribrachiatus orthostylus (54.321Ma), Sphenolithus radians (54.169Ma) and the top of Tribrachiatus contortus (54.117Ma).These three events are expected to be found prior to H1/ETM-2 (Agnini et al., 2007) but appear 9 m above H1/ETM-2 in the sediment records of U1553.The base of Discoaster multiradiatus prior to the PETM (57.32 Ma) agrees with the chemostratigraphic age model.Due to these discrepancies, we evaluated the validity of two alternative age models (Figure 5a): one based on chemostratigraphic tie points (option 1; Table 2) and the other based only on the biostratigraphic tie points (option 2; Table 2).To represent the biostratigraphic age model for option 2, we applied the biohorizons from Hole C, as these datums generally have smaller uncertainties and are in slightly closer agreement to the δ 13 C tie points than the datums from Hole D (Figure 5a).We also removed the T F. tympaniformis datum as a tie point, as the presence of non-typical morphotypes (i.e., F. cf.tympaniformis) for tens of meters above the PETM means that precise delineation of this biostratigraphic event at Site U1553 is highly unreliable.

Table 2
List of Utilized Tie Points for the Early Eocene Site U1553 Age Model: Hyperthermal Events (Hyp; Cramer et al., 2003) In order to match our age models with those of ODP Sites 1258 and 1262, we offset all three vertical axes of δ 13 C with a range of 3.5‰, but changed the absolute values in order to provide an overlap between all records (Figures 5b and 5c).Comparison of both options indicates that the CIE coinciding with option 1's K event, is shifted back by 1.2 Ma to the onset of H1 in age model option 2 (Figures 5b and 5c-dashed lines).Similarly, the negative excursion that coincides with H1 in option one is shifted back by 0.6 Ma in option 2. The large offsets for the K and the H1 event are a result of the delayed biostratigraphic events at Site U1553 (Figure 4) compared to ODP Sites 1258 and 1262 (Westerhold et al., 2017).
Overall, option 1 shows a very good correlation in the amplitude and shape of hyperthermal events before and after the PETM.Furthermore, the relatively stable δ 13 C values after the PETM and the negative shift around Previous biostratigraphic work conducted on sediments from nearby DSDP Site 277, suggests the presence of a >1 Myr hiatus after the PETM.This is based on the short depth interval (approx.40 cm) between the top of Fasciculithus spp.(55.64 Ma) and the base of Tribrachiatus orthostylus (54.37 Ma).In addition, the presence of a shorter (∼200 kyr) hiatus was proposed, based on the joint first occurrences of Tribrachiatus orthostylus (54.37 Ma) and Sphenolithus radians (53.17 Ma) within the same sample (Hollis et al., 2015;Shepherd et al., 2021).However, our biostratigraphic work at Site U1553 does not support the presence of either one of these hiatuses.Therefore, the hypothesized hiatuses at DSDP Site 277, are likely a result of single-hole, spot coring and are not a true stratigraphic signal.
As described for many other late Paleocene to early Eocene sections worldwide, the U1553 record exhibits a pronounced cyclicity (e.g., Kirtland Turner et al., 2014;Littler et al., 2014;Vahlenkamp et al., 2020;Westerhold et al., 2017;Zachos et al., 2010).Therefore, we plotted the eccentricity according to Laskar et al. (2011) and added a bandpass filter to highlight the long eccentricity cycle of 405 kyr (Figure 6a).We also plotted bulk sediment δ 18 O (as a rudimentary temperature proxy) and in-splice XRF-derived Ca/Fe data on our new chemostratigraphic age model to determine if paleoenvironmental trends were astronomically paced.As our δ 18 O record has a low signal-to-noise ratio, likely due to diagenetic overprinting (Hollis et al., 2015;Schrag, 1999;Schrag et al., 1995;Sexton et al., 2006), we applied a five-point running average (Figure 6c).Overall, the δ 13 C and δ 18 O isotopic signatures of the hyperthermal events closely follow the short and long eccentricity cycle with the exception of the PETM (Figures 6a-6c), which is consistent with previous studies.The δ 18 O signal reveals negative values during all of the hyperthermals and therefore demonstrates the expected warming during these events.In addition, the smaller hyperthermal events E1, E2 and F, which cannot be identified within our δ 13 C record, show clear negative excursions in the δ 18 O record, which coincides with the long eccentricity cycles 137 and 138 (Figure 6c).XRF-derived Fe on its own exhibits the dilution of carbonate (represented by Ca).To minimize the dilution effect, we applied the Ca/Fe ratio as shown in Figure 6d.The high positive Ca/Fe values during all hyperthermal events suggests either higher carbonate productivity and/or confirms that the sediments at Site U1553 were deposited above the CCD during the late Paleocene and early Eocene.Overall, our new age model and its comparability to stable isotope records of ODP Sites 1258 and 1262, highlights its applicability, and demonstrates the value of Site U1553 for future paleoceanographic research.

Diachroneity of Early Eocene Calcareous Nannofossil Events in the High Latitude Southern Hemisphere
Our Site U1553 age model provides strong evidence for the diachroneity of calcareous nannofossil bioevents between Site 1262 (Agnini et al., 2007;Westerhold et al., 2017) and Site U1553.A possible explanation for this diachroneity is the latitudinal difference between sites with well-defined biostratigraphic tie points (Agnini et al., 2007(Agnini et al., , 2014) ) and Site U1553.Indeed, the traditionally used Paleogene calcareous nannofossil biozones are predominantly based on low-to mid-latitude assemblages (Agnini et al., 2014;Martini, 1971;Okada & Bukry, 1980) and therefore include many warm-water taxa (e.g., Discoaster spp.) that only have rare or spotty occurrences at high latitude sites such as U1553.Despite the apparent latitudinal diachroneity of nannofossil datums, it is striking that the sequence of biostratigraphic events at Site U1553 is identical to those at low-and mid-latitude sites.This suggests that the first and last occurrences of biostratigraphically important nannofossil taxa relative to one another were the same on a global scale, but that this evolutionary sequence occurred geologically later at Site U1553 compared to lower latitude sites.One of the most surprising results of our study is the >2.5 Myr delay in the highest observed occurrence of Fasciculithus at Site U1553 compared to ODP Site 1262 (Table 3).Although not explicitly stated in the literature, data from Maud Rise (ODP 690) in the high-latitude South Atlantic Ocean, also show that Fasciculithus is present until ∼20 m above the PETM interval (within nannofossil biozone NP10), and for this reason, this marker taxon could not be used to delineate the Paleocene/ Eocene boundary (Cramer et al., 2003;Pospichal & Wise, 1990).Rare Fasciculithus specimens were also consistently observed above this reported last occurrence, although these specimens were described as reworked.
The reasons why these Fasciculithus specimens were considered reworked are unclear, and may simply be a result of the authors deciding that this taxon had already extended too far above its previously reported stratigraphic range.For this reason, it is possible that the delayed extinction of Fasciculithus at Site U1553 is a primary ecological signal, which might also be a feature of other high-latitude sites.
As Fasciculithus is commonly interpreted to have preferred warmer surface waters (e.g., Bralower, 2002;Gibbs et al., 2006;Mutterlose et al., 2007), it is particularly unusual that it persists for longer than its previously documented stratigraphic range at a high-latitude site such as Site U1553.Although our study is the first to observe possible diachroneity in the top occurrence of Fasciculithus, its bottom occurrence during the Paleocene was shown to have a <1 Myr discrepancy between four different basins sampling a range of paleolatitudes (Fuqua et al., 2008).This previous study found that temperature only had a secondary control on the origination of this genus, with its first occurrence observed earlier at a higher-latitude site on the Exmouth Plateau, Indian Ocean (ODP Site 761, paleolatitude of ∼40°S) compared to a low-latitude site on the Shatsky Rise, Pacific Ocean (ODP Site 1209; paleolatitude of ∼15°N).It was therefore hypothesized that the first occurrence of Fasciculithus was primarily controlled by a decrease in surface ocean nutrient levels, which in turn was driven by increased biological pump efficiency.The complex interplay of processes that govern the efficiency of the biological pump are known to vary greatly over various spatial and temporal scales (see review in Honjo et al., 2014), therefore, it is reasonable to expect this mechanism to drive diachroneity in the occurrences of nannofossil taxa (in this case Fasciculithus) that had an ecological preference for specific surface ocean nutrient conditions.As it is likely that surface water nutrient availability was the primary control on the first occurrence of Fasciculithus, it is plausible that it also had a greater influence than temperature on the last occurrence of this taxon during the early Eocene.This is supported by data indicating that the latitudinal temperature gradient was almost non-existent during the early Eocene (Bijl et al., 2009), suggesting that alternative environmental variables (other than temperature) had a larger control on nannofossil distribution patterns at this time.Unfortunately, independent temperature and/or nutrient availability proxy data do not currently exist for the early Eocene interval at Site U1553, so we are unable to properly test this hypothesis.However, it is an interesting possibility that should be explored more fully in the future, both at Site U1553 and at other early Eocene sites.
Other nannofossil biozones were also shown to be diachronous at Site U1553, with the most extreme example being the base of Discoaster lodoensis.At ODP Sites 1258 and 1262-low to mid-latitude sites with orbital age control-this datum is placed before the I2 event (Westerhold et al., 2017), but occurs >2 Myr later at Site U1553 (Table 3).Some of this apparent diachroneity could be explained by the relatively low core recovery of U1553C-27R (about 30%, Röhl et al., 2022b), but this alone cannot explain the entire discrepancy.Like Fasciculithus, the genus Discoaster is interpreted as a warm-water taxon (e.g., Aubry, 1998;Bralower, 2002;Tremolada & Bralower, 2004); therefore, the delay in the first occurrence of D. lodoensis at high versus low-to mid-latitude sites is not unexpected.However, detailed scrutiny of the literature revealed that the first occurrence of Discoaster lodoensis is globally hetereogenous, even for sites that sample similarly high paleolatitudes.For example, the FO of Discoaster lodoensis at IODP Site U1514 and ODP Site 752 (Huber et al., 2019;Millen, 2012) is also delayed when compared to ODP Sites 1258 and 1262, but is observed much earlier (below ETM-2; ∼54.05 Ma) at ODP Site 690 (Pospichal & Wise, 1990).This heterogeneity could partially be explained by: (a) the low sampling resolution of the biostratigraphic analyses that have been conducted at Sites U1513, 752 and 690, (b) the rare and "spotty" occurrence of Discoaster at high-latitude sites, and/or (c) the common poor-preservation of Discoaster specimens (generally due to overgrowth) that makes taxonomic identification to species-level challenging.Future research should therefore establish the extent to which the diachroneity of the Discoaster lodoensis datum is a primary signal.Gradstein et al. (2012).*Shipboard data (Röhl et al., 2022b).
In addition to these more extreme examples, the base of Tribrachiatus orthostylus, the top common of Discoaster multiradiatus, the base of Sphenolithus radians and the top of Tribrachiatus contortus, all occur shortly before the onset of the H1 event (54.05) at ODP Site 1262 (Westerhold et al., 2017), but occur 0.5 to 1 Ma later at Site U1553 (Table 3).As reported in the IODP Expedition 378 Proceedings volume (Röhl et al., 2022b), the preservation of calcareous nannofossils is generally moderate to good.However, during a few restricted (cm-scale) intervals, the preservation of nannofossils decreased, perhaps due to the presence of thin layers that are enriched in silica (e.g., cherts) or were more affected by post-depositional processes such as cementation and recrystallization.Although this made it challenging to confidently identify taxa that are more susceptible to overgrowth (e.g., Tribrachiatus contortus and Discoaster lodoensis), these poorly-preserved intervals are too short and sporadic to explain the multiple >0.5 Myr-long diachroneities that we observe in the nannofossil datums.Therefore, despite ruling out various potential causes of the apparent diachroneity in calcareous nannofossil biostratigraphic events at Site U1553, the lack of comparable data from nearby high-latitude sites means that it is not currently possible to determine the exact drivers of this phenomenon.For this reason, future work should focus on developing further high-latitude age models to determine whether the apparent latitudinal diachroneity in nannofossil bioevents is characteristic of all early Paleogene high-latitude sites, or whether it is a specific feature of Site U1553.

Conclusions
Here, we present a new late Paleocene to early Eocene age model (50.5-57.5 Ma), reconstructed using sediments recovered from two holes at IODP Site U1553 in the high-latitude southwest Pacific Ocean.Our chemostratigraphic correlation reveals several characteristic paired negative carbon and oxygen isotope excursions, indicative of the early Eocene hyperthermals, underlining the global impact of carbon perturbations.In addition, the Site U1553 age model represents one of the most stratigraphically complete late Paleocene-early Eocene sections from this region to date.Therefore, our new record will be critical for future research on paleoenvironmental and paleoecological changes during warmer worlds at understudied high-latitude sites.Our study highlights that the traditionally used calcareous nannofossil biostratigraphic datums-calibrated to low-and midlatitude assemblages-cannot be reliably utilized at the high southern latitudes.Future biostratigraphic work should focus on devising a high-latitude biozonation scheme that can be calibrated to astronomically tuned records from lower latitude sites.Critically, the results of our study underline the necessity to check other highlatitude records, to determine whether the latitudinal diachroneity of calcareous nannofossils is the rule rather than the exception.
During IODP Expedition 378, Pleistocene to early Paleocene sediments were recovered from five holes at Site U1553, located on the Campbell Plateau off the south coast of New Zealand at 52°13.4′S, 166°11.5′E(Röhl

Figure 1 .
Figure 1.56 Ma paleogeographic reconstruction in Robinson projection showing the position of IODP Site U1553 (black star-this study) and the reference low-to mid-latitude Ocean Drilling Program Sites 1258 (blue star) and 1262 (orange star) used for chemostratigraphic correlation.The reconstruction highlights the high-latitude position of Site U1553 at ∼60°S during the early Eocene.Paleomap was generated with free software tool GPlates(Müller et al., 2018) by using the integrated combined rotation file(Cao et al., 2020;Müller et al., 2019;Torsvik et al., 2019;Young et al., 2019).
and Site 1258 on the Demerara Rise

Figure 2 .
Figure 2. The late Paleocene-early Eocene record from IODP Site U1553: (a) Core images and XRF-Fe (Röhl et al., 2022b) of the splice from Holes U1553 Holes C and -D (Drury et al., 2022; Wilkens et al., 2022) (b) Bulk sediment δ 13 C (c) Position of calcareous nannofossil biostratigraphic events.(T = Top occurrence, Tc = Top common occurrence, B= Bottom occurrence, cf.= confer).Genus names abbreviations are D.: Discoaster, F.: Fasciculithus, T.: Tribrachiatus, and S.:Sphenolithus.The highest observed occurrence (HO) of F. cf.tympaniformis is also shown (the true top of F. cf.tympaniformis expected to be even higher up but is not well constrained herein).

Figure 5 .
Figure 5. (a) Age-depth plot with tie points for the age model.Green: correlation of bulk sediment δ 13 C-hyperthermal events, black: biostratigraphy of calcareous nannofossil events in U1553C and gray: U1553D.The circle shows the highest observed occurrence of F. cf.tympaniformis (b) Age model option 1: correlation of the bulk sediment δ 13 C record from IODP Site U1553 (black line-this study), with Ocean Drilling Program (ODP) Site 1258 (blue line) and ODP Site 1262 (orange line).(c) Age model option 2: constructed using the calcareous nannofossil biozone datums from Site U1553 Hole C compared to the same datums from ODP Sites 1258 and 1262 against the bulk sediment δ 13 C records for each site.The dashed lines indicate the substantial time offset between the two alternative age models.

Figure 6 .
Figure 6.Earth eccentricity variability in comparison with Site U1553 data.(a) Earth eccentricity (La2010b) and its 405 kyr bandpass filter generated with Acycle (Li et al., 2019).The numbers reflect the long eccentricity cycles counted backwards from today.(b) Combined bulk sediment δ 13 C. (c) Combined bulk sediment δ 18 O, on which a five-point running average was applied to provide better visibility.(d) XRF Ca/Fe ratio.

Table 1
Sample Intervals of Calcareous Nannofossil Biostratigraphic Events at Site U1553 Holes C and -D, With T =

Table 3
Westerhold et al. (2017)e Chemostratigraphic Age Model (Option 1) for Biostratigraphic Events and Their Time Difference Compared to Ages ofWesterhold et al. (2017) Note.T, Top occurrence; Tc, Top common occurrence; cf, Confer; B, Bottom occurrence; HO, Highest observed occurrence."Age for B Discoaster multiradiatus from