Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
We reconstruct the latest Paleocene and early Eocene (∼57–50 Ma) environmental trends in the Arctic Ocean and focus on the Paleocene-Eocene thermal maximum (PETM) (∼55 Ma), using strata recovered from the Lomonosov Ridge by the Integrated Ocean Drilling Program Expedition 302. The Lomonosov Ridge was still partially subaerial during the latest Paleocene and earliest Eocene and gradually subsided during the early Eocene. Organic dinoflagellate cyst (dinocyst) assemblages point to brackish and productive surface waters throughout the latest Paleocene and early Eocene. Dinocyst assemblages are cosmopolitan during this time interval, suggesting warm conditions, which is corroborated by TEX86′-reconstructed temperatures of 15°–18°C. Inorganic geochemistry generally reflects reducing conditions within the sediment and euxinic conditions during the upper lower Eocene. Spectral analysis reveals that the cyclicity, recorded in X-ray fluorescence scanning Fe data from close to Eocene thermal maximum 2 (∼53 Ma, presence confirmed by dinocyst stratigraphy), is related to precession. Within the lower part of the PETM, proxy records indicate enhanced weathering, runoff, anoxia, and productivity along with sea level rise. On the basis of total organic carbon content and variations in sediment accumulation rates, excess organic carbon burial in the Arctic Ocean appears to have contributed significantly to the sequestration of injected carbon during the PETM.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
While many low-latitude and midlatitude and limited high southern latitude PETM climate proxy records exist, northern high-latitude records are rare because of limited access to suitable core drilling sites. Information from this region, however, is vital to assess the status of global climate under high CO2 concentrations because it defines end-members of important parameters, such as the meridional temperature gradient and atmospheric water transport. The few previous studies, from continental sections with generally poor dating, indicate that the Arctic Region was indeed warm during the latest Paleocene through early Eocene as indicated by floral [Greenwood and Wing, 1995], faunal [Markwick, 1998] and isotopic [Bice et al., 1996; Tripati et al., 2001] evidence. Recently, the first scientific drilling expedition was carried out in the Central Arctic Ocean (Integrated Ocean Drilling Program (IODP) Expedition 302, Arctic Coring Expedition (ACEX), Lomonosov Ridge). Initial ACEX results show that extremely high temperatures (between 17°C and 24°C) and low surface salinities prevailed in the Arctic around the PETM and during the early Eocene, while ventilation of the basin was sluggish [Backman et al., 2006; Brinkhuis et al., 2006; Moran et al., 2006; Pagani et al., 2006b; Sluijs et al., 2006; L. M. Waddell and T. C. Moore, Salinity of the Eocene Arctic Ocean from oxygen isotope analysis of fish bone carbonate, submitted to Paleoceanography, 2007].
We have expanded the initial palynological and organic geochemical data sets, and generated elemental abundances from high-resolution X-ray fluorescence (XRF) core scanning and from chemical analyses on discrete samples through upper Paleocene–lower Eocene sediments of the Expedition 302 section (cores 302-M0004A-11X through 302-M0004A-35X, from now on referred to as cores 11X through 35X). Hereby we augment previous studies [Backman et al., 2006; Brinkhuis et al., 2006; Pagani et al., 2006b; Sluijs et al., 2006] to better assess long-term late Paleocene–early Eocene environmental and climate trends, and to place the Arctic PETM record into a better constrained perspective.
Palynological processing was performed using standard methods [cf., e.g., Sluijs et al., 2003]. Briefly, samples of 1–2 cm stratigraphic thickness were freeze or oven dried and a known amount of Lycopodium spores was added to ∼2–4 g of material. Samples were then treated with 30% HCl and twice with 38% HF for carbonate and silicate removal, respectively. Residues were sieved using a 15-μm nylon mesh to remove small particles. To break up clumps of residue, the sample was placed in an ultrasonic bath for a maximum of 5 min, sieved again, and subsequently concentrated into 1 mL of glycerin water, of which 10 μL was mounted on microscope slides. Slides were counted for marine (e.g., dinocysts) and terrestrial palynomorphs (e.g., pollen and spores) to a minimum of 200 dinocysts. Average sampling resolution was ∼1 m but up to 20 cm through the PETM and the total sample number is 75. Marine and terrestrial palynomorph preservation was excellent for all samples. We follow dinocyst taxonomy of [Fensome and Williams, 2004]. Absolute quantitative numbers were counted using the relative abundance of Lycopodium spores [cf. Stockmarr, 1972].
2.2. Inorganic Geochemistry Including XRF Core Scanning
Standard analytical methods for discrete samples are discontinuous, time consuming, and expensive. Relatively fast nondestructive core logging methods yield continuous data at much finer scales (millimeter scale) than are practical for individual sampling methods [Röhl and Abrams, 2000]. We measured the elemental composition of sediments from Cores 19X through 35X at Bremen University, Germany, using the XRF core scanner II [Richter et al., 2006; Tjallingii et al., 2007]. This scanner allows high-resolution, nearly continuous, nondestructive analyses of major and minor elements at the surface of split cores. XRF data were collected every 1 cm down core over a 1 cm2 area using 30 s count time. For the mostly laminated sediments in Core 30X [Sluijs et al., 2006] we also applied a much higher resolution and collected data every 2 mm down core over a 2 mm × 1 cm area using 30 s count time. To investigate whether cyclicity in our data is related to orbital forcing we calculated spectra in the depth domain to identify the dominant cycle periods within Fe XRF scanner data of some selected cores. Blackman-Tukey power spectra have been calculated by the AnalySeries program [Paillard et al., 1996].
Quantitative XRF measurements were performed on 154 selected discrete samples. Sediment samples were freeze-dried and homogenized in an agate mill. All samples were analyzed for major and minor elements using a Philips PW 2400 X-ray spectrometer. 600 mg of sample was mixed with 3600 mg of a mixture of dilithiumtetraborate/lithiummetaborate (50% Li2B4O7/50% LiBO2), preoxidized at 500°C with NH4NO3 (p.a.), and fused to glass beads. Analytical precision and accuracy were better than 5% for all elements, as checked by in-house and international reference rocks. Total sulfur (TS) was determined using an Eltra CS 500 IR analyzer. Precision and accuracy of TS analyses were better than 3%.
2.3. Organic Geochemistry
A total of 65 powdered and freeze-dried sediment samples (1–3 g dry mass) were extracted with dichloromethane (DCM)/methanol (2:1) by using accelerated solvent extraction (Dionex ACE). The extracts were separated by Al2O3 column chromatography using hexane/DCM (9:1) and DCM/methanol (1:1) to yield the apolar and polar fractions, respectively. The polar fractions were analyzed for glycerol dialkyl glycerol tetraether (GDGT) lipids by high-performance liquid chromatography/mass spectrometry [Hopmans et al., 2000; Schouten et al., 2007a]. The relative abundance of GDGTs were used to calculate TEX86 [Schouten et al., 2002] and TEX′86 [Sluijs et al., 2006]; reproducibility was within ∼1°C; and the Branched and Isoprenoid Tethraether (BIT) index. The BIT index describes a ratio between soil bacteria-derived and marine chrenarchaeota-derived GDGTs (see Hopmans et al.  for more detailed method description). TEX86 differs from TEX′86 in that the latter does not include the the GDGT lipid containing three cyclopentane rings, which was unusually abundant in strata close to the PETM because of a high terrestrial GDGT component [Sluijs et al., 2006]. TEX86 and TEX′86 are calibrated by core top analysis to mean annual mean SST. However, because the export of GDGT lipids to the seafloor was likely dominant during summer in the Arctic Ocean, the signal is likely skewed toward summer SSTs [Sluijs et al., 2006].
2.4. Age Model
Approximately 42% of the upper Paleocene and lower Eocene sediments were recovered during IODP Expedition 302 (ACEX; Cores 35X–11X) in Hole 4A on the Lomonosov Ridge (∼87° 52.00′ N; 136° 10.64′ E; 1288 m water depth; Figure 1) [Backman et al., 2006]. Sediments are devoid of calcareous microfossils and also barren of siliceous microfossils below ∼315 m composite depth (mcd). Organic dinoflagellate cysts (dinocysts) are the only age indicative microfossils present, except for a few agglutinated benthic foraminifera [Backman et al., 2006]. The lower Paleogene sediments overlie a basement of Campanian age; the unconformity occurs within Core 35X at ∼405 mcd [Backman et al., 2006]. The onset of the PETM is here based on identification of the CIE in both total organic carbon (TOC) and the C29n-alkane, derived from higher plant waxes, and the occurrence of the dinocyst species Apectodinium augustum, which is diagnostic of the PETM [Bujak and Brinkhuis, 1998; Steurbaut et al., 2003; Sluijs et al., 2007a]. It occurs between Core 32X and the poorly recovered core catcher 31X-CC, implying a possible depth between 385.10 and 388.61 mcd [Pagani et al., 2006b; Sluijs et al., 2006] (Figure 2). The termination of the PETM is in general somewhat subjective for pelagic sections because of the exponential recovery of the CIE [Röhl et al., 2007]. On the basis of the δ13C of TOC and of C29n-alkane, the termination of the CIE is estimated at ∼378.41 mcd [Pagani et al., 2006b; Sluijs et al., 2006]. The position of the CIE termination is also supported by the concomitant last occurrence of the dinocyst species Apectodinium augustum, which has only been recorded from the CIE [Bujak and Brinkhuis, 1998; Steurbaut et al., 2003; Sluijs et al., 2007a]. As the CIE reflects an isotopic anomaly within the global exogenic carbon pool, the termination of the CIE in the Arctic organic carbon records should be synchronous with the CIE termination of well known pelagic PETM sections from the Atlantic and Pacific Oceans (e.g., [Röhl et al., 2000; Zachos et al., 2003, 2005]), even considering local factors. While the paleomagnetic signal is weak throughout the recovered section [Backman et al., 2006], one clear polarity reversal was recognized at 399.63 mcd and identified as the chron C25n–C24r reversal based on the overlying PETM [Backman et al., 2006].
Recently, Stein et al.  recognized a transient negative excursion in δ13C of TOC at ∼368 mcd, and assigned it to ETM2. This interval, however, lacked good biostratigraphic or magnetostratigraphic control and this δ13C excursion could also represent another event, such as the X event. Ongoing high-resolution multiproxy studies have recorded the dinocyst species Cerodinium wardenense from this interval (at ∼368.3 mcd) and a quasi-concomitant decrease in abundance of Hystrichosphaeridium tubiferum (A. Sluijs and H. Brinkhuis, unpublished results, 2007). C. wardenense has a reported last occurrence in the North Sea close to the nannofossil zone boundary NP10–NP11, which closely coincides with the top of the H. tubiferum zone of [Bujak and Mudge, 1994], and is just below magnetochron reversal C24r–C24n.3n [Luterbacher et al., 2004]. These analogous stratigraphic characteristics, and the general similarity of early Eocene dinocyst assemblages from the Arctic Ocean and the North Sea, suggest these events are synchronous between the two oceans. Considering that the X event is dated within Chron C24n.1n and NP12 [Röhl et al., 2005], and ETM2 within latest Chron C24r and earliest NP11 [Lourens et al., 2005], the presence of C. wardenense in this interval confirms the identification of ETM2 by Stein et al. . The onset of the CIE of ETM2 is now recognized at 368.9 mcd (Figure 2).
As yet, no other age indicative dinocysts have been recorded from the lower Eocene of the Expedition 302 cores. As indicated below, the dinoflagellate markers commonly used in the North Atlantic region were most likely not adapted to low surface water salinities that prevailed in the Arctic Ocean during this time interval. The lowermost middle Eocene is characterized by mass abundances of remains of the freshwater fern Azolla [Brinkhuis et al., 2006]. Age-equivalent sediments with abundant Azolla have also been recorded at North Atlantic Ocean Drilling Program (ODP) Site 913 [Eldrett et al., 2004]. The last abundant occurrence has there been calibrated to magnetochron C21r, which is also the assumed age for the top (298.9 mcd) of abundant Azolla fossils on Expedition 302 cores [Brinkhuis et al., 2006].
To ultimately estimate sediment accumulation rates, we applied the orbitally tuned, but still floating, age model defined on multiple-drilled and complete late Paleocene to early Eocene pelagic sediments from Walvis Ridge ODP Leg 208 [Westerhold et al., 2007]. Owing to uncertainties in the astronomical calculations far back in geological time [Laskar, 1999], absolute orbitally tuned ages cannot yet be provided. Nevertheless, Westerhold et al.  provide the most accurate durations available for intervals between the termination of Chron C25n and the onset of the CIE of the PETM (1.113 Ma), and between the onset of the PETM-CIE and the onset of the CIE of ETM2 (1.827 Ma). For the duration of the PETM-CIE, we use 170 ka, recently defined in marine pelagic sections [Röhl et al., 2007], which is consistent with a revised age model from the terrestrial Bighorn Basin deposits [Sluijs et al., 2007b]. For the interval between the termination of the PETM-CIE and the onset of the CIE of ETM2 we subtract 170 ka from 1.827 Ma [Westerhold et al., 2007], arriving at 1.657 Ma.
The age model of Westerhold et al.  stops at the termination of chron C24r. Hence, for estimating sedimentation rates between chron C24r and the termination of the Azolla phase we adapt the recent geomagnetic polarity timescale of Ogg and Smith . The termination of the Azolla phase within Chron C21r was dated at 48.6 Ma, relative to Cande and Kent  [Eldrett et al., 2004]. The timescale of Ogg and Smith , however, has a slightly different age and duration for Chron C21r, which indicates an age of 48.1 Ma for the termination of the Azolla phase. The onset of Chron C24n.3n is dated at 53.8 Ma [Ogg and Smith, 2004], giving a duration of 5.7 Ma between the onset of C24n.3n and the termination of the Azolla phase. The onset of the CIE of ETM2 is 0.17 Ma older than the base of Chron C24n.3n [Westerhold et al., 2007], which provides a duration of 5.90 Ma between the onset of ETM2 and the termination of the Azolla phase.
3.1. Long-Term Late Paleocene and Early Eocene Trends
Palynomorph assemblages from the upper Paleocene and lowermost Eocene (Cores 35X–27X) are characterized by abundant terrestrial palynomorphs (Figure 3a) and sediments that contain abundant amorphous organic matter, mostly of terrestrial origin. In this interval, as well as near the termination of the PETM, spores produced by the fern family Osmundaceae are consistently present and abundant (Figure 3a). Upper lower Eocene are characterized by abundant aquatic palynomorphs, notably dinocysts, and substantially lower BIT indices, while large spores derived from Osmundaceae are nearly absent (Figure 3a).
Throughout the upper Paleocene and lower Eocene, dinocyst assemblages are generally dominated by the peridinioid dinocyst genera Senegalinium spp. and Cerodinium spp. Other consistently present peridinioid taxa include Deflandrea and Phthanoperidinium. These cysts were likely produced by dinoflagellates that required nutrient-rich conditions [Dale, 1996; Röhl et al., 2004; Sluijs et al., 2005] and were tolerant to low surface water salinities [Pross and Brinkhuis, 2005; Brinkhuis et al., 2006]. In many sections, abundant occurrences of these dinocyst genera are accompanied by the records of abundant terrestrial spores and pollen, and freshwater algae [Sluijs et al., 2005] and are grouped in low-salinity-tolerant dinocysts (Figures 3a). Gonyaulacoid genera include Spiniferites, Areoligera and Membranospaera spp.
Values of the BIT index, a biomarker-based measure for the amount of river-transported soil organic matter relative to marine organic matter [Hopmans et al., 2004], are relatively high at an average of ∼0.4 in the upper Paleocene and lowermost Eocene, while they are low (∼0.1) in the upper lower Eocene (Figure 3a). TEX′86 records sea surface temperatures of 17° or 18°C on average in the uppermost Paleocene and lowermost Eocene, while they are 12°C to 13°C in the uppermost early Eocene.
XRF core scanner data and geochemical analyses on discrete samples both clearly reflect general lithological trends. The correlation between XRF scans and single samples analyzed by conventional XRF on borate glass beads is generally good (Figure 3b). This allows geochemical element intensities (count rates) to be transferred into concentration ranges, implying a significant improvement of the resolution for the discrete sample analyses.
Sediments from the upper Paleocene and lower lower Eocene are clay-rich (9.6% Al) with TOC values around 2% (Figure 3b and fifth column of Table 1) [Stein et al., 2006; Stein, 2007] and total sulfur contents slightly exceeding 2%. Biogenic silica is not present in significant amounts, as seen in the mean Si/Al ratio of 2.71, which is lower than the average shale value of 3.1 [Wedepohl, 1971; Brumsack, 2006]. Sediments are also devoid of CaCO3, considering the very low Ca/Al ratios (fifth column of Table 1). The upper lower Eocene reveals a sediment type with a high amount of biogenic silica (Si/Al ratio averaging three times the shale level) and extremely high Fe and S contents (sixth column of Table 1 and Figure 3b). The sediments are fully pyritized (Fe/S < 1), have high trace element concentrations and are enriched in diagenetic minerals like apatite or rhodochrosite resp. ankerite (higher Ca/Al, P/Al, and Mn/Al ratios), owing to significantly more pore space [Backman et al., 2006].
Table 1. Elemental Mean Values of the Upper Paleocene and Lower Eocene Part of the Integrated Ocean Drilling Program Expedition 302 Section in Comparison With Average Shale Values
The number of total palynomorphs per gram of dry sediment is particularly high within the lower part of the PETM (Figure 4) and relatively low before and after the PETM. The same holds for the number of dinocysts per gram sediment. Dinocyst diversity across the PETM is low compared to other sites [e.g., Powell et al., 1996; Crouch et al., 2003; Crouch and Brinkhuis, 2005] and apart from Apectodinium [Sluijs et al., 2006], assemblages in the lower part of the PETM are dominated by Cerodinium (Figure 5). Although consistently present before and during the initial stages of the PETM, normal marine taxa belonging to the Membranosphaera complex (cpx; consisting of the genera Membranosphaera, Histiocysta and Microdinium) and Spiniferites cpx (consisting of the genera Spiniferites, Achomosphaera, Hafniasphaera and Rottnestia) become abundant toward the termination of the PETM (Figure 5). Above the PETM, typical open marine inner neritic [Brinkhuis, 1994; Pross and Brinkhuis, 2005; Sluijs et al., 2005] Glaphyrocysta cpx (consisting of the genera Glaphyrocysta and Areoligera) become more abundant, particularly Glaphyrocysta ordinata (Figure 5). Other more or less consistently present taxa from the PETM interval include Caligodinium aceras, Diphyes colligerum, Alisocysta sp.2 of Heilmann-Clausen , Hystrichosphaeridium tubiferum, Muratodinium fimbriatum, Operculodinium spp., Thalassiphora pelagica, T. delicata and Wilsonidiumtesselatum.
During the PETM an increase in S, Fe/S, Fe/Al, Mn/Al, and trace element/Al ratios is recorded (Table 1 and Figure 3b). Si/Al and Ti/Al ratios are low during the PETM, whereas the Fe/Al ratio is slightly elevated. A distinct Fe maximum was identified in both XRF data sets at ∼383 mcd (Figure 6).
4.1. Sediment Accumulation Rates
Sediment accumulation rates are estimated for the uppermost Paleocene and lower Eocene of the IODP Expedition 302 section using depths and durations as described above (Figure 2). Generally, sediment accumulation rates are close to 1 cm ka−1, but are significantly higher (5.0 ± 1.2 cm ka−1) during the PETM (Figure 2). While there is notable uncertainty in this estimate due to the imprecise position of the onset of the PETM-CIE, it, nevertheless, implies that sedimentation accumulation rates increased threefold to fivefold during the PETM. This was most likely a result of enhanced siliciclastic input by rivers in response to an intensification of the hydrological cycle during the PETM, which is also supported by changes in geochemical data, such as Fe/Al ratios (Table 1), and biotic changes (see below).
4.2. Changes in Relative Sea Level
On the basis of palynological and geochemical records, upper Paleocene to lower Eocene sediments from Expedition 302 Hole M0004A can generally be divided in two major intervals, that is, upper Paleocene through lower lower Eocene (Cores 35X–27X, at ∼400–365 mcd) and upper lower Eocene (Cores 24X–12X, at ∼365–310 mcd) (Figure 3).
In the upper Paleocene and lowermost Eocene, as well as near the termination of the PETM, spores produced by the fern family Osmundaceae are consistently present and abundant (Figure 3a). These spores are very large and thick-walled (100–300 μm) and cannot have been transported over long distances, implying the site was close to shore. As the spreading Gakkel Ridge had separated the Lomonosov Ridge from the Siberian Shelf by the latest Paleocene [Jokat et al., 1992, 1995], the consistent occurrence of Osmundaceae spores indicates that parts of the Lomonosov Ridge in the vicinity of the drill site were still above sea level at least until near the onset of ETM2 (Figure 3a). This interpretation is supported by the general abundance of terrestrial palynomorphs and palynodebris and high BIT index values (Figure 3a).
A marine transgression in the lower part of the PETM was initially suggested by Sluijs et al. , based on a relative decrease in terrestrially derived organic matter recorded in palynomorph assemblages, BIT index and Rock Eval Hydrogen Index. Corroborating this interpretation is the near absence of Osmundaceae spores during most of the PETM (∼387–382 mcd; Figure 4). As discussed above, these large spores are not likely to be transported over long distances and we surmise that during the initial stages of the PETM the site was located too far from shore for the spores to have been deposited. Dinocyst assemblages were barely influenced by changes in sea level but largely reflect changes in salinity and runoff (see below).
Upper lower Eocene strata are characterized by low abundances of terrestrially derived organic matter (few terrestrial palynomorphs low BIT values, and the absence of Osmundaceae spores; Figure 3a). Hence it appears that during the late early Eocene the drill site was located further from shore than during the late Paleocene–earliest Eocene, which most likely reflects the gradual subsidence of the Lomonosov Ridge during the early Eocene (M. O'Regan et al., Mid-Cenozoic tectonic and paleoenvironmental setting of the central Arctic Ocean, submitted to Paleoceanography, 2007).
4.3. Salinity and Runoff
In the upper lower Eocene, very high percentages of low-salinity-tolerant dinocysts are recorded (Figure 3a), despite the more distal setting relative to underlying Upper Paleocene to lowermost Eocene. In addition, dinocyst diversity is low and very few typically open marine dinocysts (e.g., Spiniferites, Areoligera and Membranospaera spp.) are recorded. Dinocyst assemblages in this interval are very similar to those recorded in the interval characterized by abundant remains of the freshwater fern Azolla, near the top of the lower Eocene [Brinkhuis et al., 2006]. This indicates that brackish or fresh surface waters already persisted during the upper lower Eocene, suggesting that only a small decrease in salinity or a change in another critical environmental parameter, perhaps phosphorus concentrations, caused the inception of conditions favorable for massive Azolla growth.
The trend from nearly 100% low-salinity-tolerant dinocyst taxa during the lower part of the PETM to assemblages dominated by typical marine taxa, such as Membranosphaera, Spiniferites and Glaphyrocysta, toward the termination of the PETM, and after, likely reflects an increase in surface salinity. This interpretation is consistent with the concomitant termination of photic zone euxinia, indicated by the decrease and subsequent absence of isorenieratane [Koopmans et al., 1996] (Figure 4), suggesting a less stratified water column, and δD-reconstructed surface water salinities [Pagani et al., 2006b] during the recovery phase of the PETM.
The number of palynomorphs and dinocysts per gram of dry sediment is particularly high within the lower part of the PETM (Figure 4). This is likely, at least in part, associated with enhanced runoff as a climate response during the PETM [Pagani et al., 2006b; Sluijs et al., 2006]. This increased nutrient supply to surface waters, resulting in more dinocyst production, and transported more terrestrial palynomorphs into the basin. Alternatively, absolute quantitative numbers, particularly dinocysts, are highest within the laminated interval (∼387–382.5 mcd), which also contains isorenieratene derivatives indicating photic zone euxinia (Figure 5). This suggests that despite the good palynomorphs preservation throughout the whole section, preservation may have affected absolute palynomorph concentrations. However, the influence of preservation may have been minimal given that terrestrial palynomorph concentrations remain unaffected at the termination of water column anoxia at ∼382.5 mcd (Figure 4).
Arctic Ocean surface temperatures during the late early Eocene appear to have been ∼4°C lower than during the late Paleocene and earliest Eocene (Figure 3a). This result differs the general assumption that climate during the late early Eocene was warmer than during late Paleocene and earliest Eocene [e.g., Zachos et al., 2001]. A possible scenario is that the Arctic region did not follow this global trend. Quantitatively, such a cooling of the Arctic Ocean is consistent with the stagnation of ocean heat transport from lower latitudes [Huber and Sloan, 2001; Huber et al., 2003; Huber and Nof, 2006], which could have been established through a decrease of oceanic exchange between the Arctic and adjacent oceans [cf. Brinkhuis et al., 2006]. This scenario would be consistent with the recorded 100% low-salinity-tolerant dinocyst assemblages, as a decrease in ocean exchange with adjacent oceans would also limit salt transport into the Arctic Ocean. Alternatively, the recorded early Eocene sea surface cooling trend could be attributed to local effects, such as the more distal position relative to the coast and a slightly deeper setting.
4.5. Environmental Implications of the Apectodinium Acme
Prior to the PETM, the genus Apectodinium was only locally and periodically abundant at low latitudes, while during the PETM it became dominant and made up >40% of the dinocyst assemblage on a global scale [Heilmann-Clausen, 1985; Powell et al., 1996; Bujak and Brinkhuis, 1998; Heilmann-Clausen and Egger, 2000; Crouch et al., 2001, 2003; Egger et al., 2003; Sluijs et al., 2006, 2007a, 2007b]. In the Arctic Ocean during the PETM, Apectodinium is slightly less abundant, making up ∼20% of the assemblage. Typical low-salinity-tolerant dinocysts are dominant. A similar trend is recorded on the New Jersey Shelf, where Apectodinium became outnumbered by typical low-salinity-tolerant dinocysts during the PETM [Sluijs et al., 2007a]. This indicates that, while Apectodinium was successful at low salinities other taxa were able to dominate dinocyst assemblages under such ecological conditions. Given the low-latitude preference of Apectodinium, it must have required high temperatures [Bujak and Brinkhuis, 1998; Crouch et al., 2001], perhaps a minimum of 20°C [Sluijs et al., 2006, 2007a]. In addition to temperature, Apectodinium perhaps required stratified surface waters [Crouch et al., 2003]. Moreover Apectodinium has morphological characteristics identical to cysts of modern heterotrophic dinoflagellates, which has fueled the hypothesis that Apectodinium was a heterotrophic dinoflagellate [Bujak and Brinkhuis, 1998]. Basic predator-prey abundance models indicate that with higher nutrient supplies, ecosystems should become relatively enriched in organisms that are higher up in the food chain, i.e., heterotrophic. Increasing nutrient levels may therefore have contributed to the Apectodinium acme. If so, the global character of the acme implies that at least neritic and coastal settings underwent significant eutrophication on a global scale [Bujak and Brinkhuis, 1998; Crouch et al., 2001], a hypothesis corroborated by many proxy data (see overview of Sluijs et al. [2007b]). For the Arctic region, this interpretation is consistent with enhanced nutrient input by rivers, higher TOC concentrations, and productivity-related trends in the algal-derived C17n-alkane [Pagani et al., 2006b]. Modern dinoflagellate blooms usually last for several days to weeks [e.g., Dale, 1996]. Conceivably, Apectodinium blooms during the PETM had similar dynamics, in which case the Apectodinium acme would imply a change in specific seasonal conditions of surface waters. This may include any of the above environmental factors. However, even a combination of these factors was likely not truly unique in the early Paleogene, suggesting that some critical environmental factor that stimulated massive Apectodinium production has not yet been identified.
4.6. Water Column and Seafloor Oxygenation
Previous studies in the Arctic region have concluded that bottom waters were at least temporally devoid of oxygen, based on laminated sediments, the absence of fossils of benthic organisms and organic geochemical evidence such as biomarkers [Backman et al., 2006; Brinkhuis et al., 2006; Sluijs et al., 2006; Stein et al., 2006]. Moreover, photic zone euxinia developed during the PETM [Sluijs et al., 2006]. To further assess the evolution of bottom water oxygenation we apply XRF analyses.
In sediments from the upper Paleocene and lower lower Eocene, sulfur is very abundant and the redox sensitive major elements Fe and Mn are present in concentrations slightly (Fe) or significantly (Mn) below the average shale level, in accordance with a generally reducing suboxic paleoenvironmental setting. This is also reflected in elevated As/Al, Mo/Al, U/Al, and V/Al ratios (Table 1). The Fe/S ratio is, however, significantly higher than 1, which points toward incomplete pyritization of this sequence, either because of a lack of seawater sulfate or, less likely, enhanced sedimentation rates [e.g., Rachold and Brumsack, 2001] (Figure 2).
During the mid early Eocene the clay-rich mudstones are replaced by biogenic silica-rich sediments (Si/Al ratio averaging three times the shale level). High trace element/Al ratios also point toward euxinic conditions, possibly comparable to Modern deep water conditions in the Baltic Sea [Huckriede and Meischner, 1996].
An additional advantage of combining XRF scanning and discrete sample analyses is that it may be possible to distinguish real changes in composition from the effects of end-member mixing. For example, the Ti or TiO2 spikes recorded in Core 28X (∼375 mcd) are possibly associated with ash layers. While turbidites could be an alternative explanation, these are normally characterized by a coarse-grained base, and sudden increases in elemental concentrations resulting into asymmetric Ti distributions. The high-resolution XRF scans do not exhibit sharp increase of Ti counts, but rather symmetrical peaks, so turbidite layers can most likely be excluded.
The occurrence of laminated sediments, from Core 31X at the onset of the PETM to 382.5 mcd, implies the absence of bioturbating organisms. This is corroborated by the absence of benthic foraminiferal linings, which are present below and above the laminated interval (Figure 4). Together, this suggests bottom waters became depleted in oxygen during the PETM. Within the laminated interval, derivatives of the characteristic pigment isorenieratene, derived from the brown strain of photosynthetic green sulfur bacteria, are recorded (Figure 4). These bacteria require euxinic (anoxic and sulfidic) conditions to thrive [Sinninghe Damsté et al., 1993], which implies that during the lower part of the PETM photic zone euxinia developed at the drill site coincident with bottom water anoxia [Sluijs et al., 2006]. This can be explained by increased surface ocean stratification due to enhanced Arctic precipitation and runoff; a hypothesis corroborated by hydrogen isotope data [Pagani et al., 2006b; Sluijs et al., 2006], a threefold to fivefold increase in sediment accumulation rates (Figure 2) and increased concentrations of terrestrial and marine palynomorphs (Figure 4).
The XRF data support a scenario of reduced oxygenation during the PETM (second, third, and fourth columns of Table 1 and Figure 6). The section below the PETM (405 to 396 mcd; fourth column of Table 1) is relatively low in S, Fe/Al, Mn/Al, and trace element/Al ratios. This suggests less intense reducing conditions and possibly low pyritization. During the PETM, conditions were much more reducing, as reflected in higher pyritization (higher Fe/S ratio, higher S contents) and elevated trace element/Al ratios. Interestingly, Si/Al and Ti/Al ratios are low during the PETM, whereas the Fe/Al ratio is slightly elevated. This may indicate more intense weathering, which also correlates with the high percentages of low-salinity-tolerant dinocysts (Figure 5), and a shift toward Al-rich (possibly kaolinite) and Fe-rich terrigenous detrital input.
A distinct Fe maximum was identified in both XRF data sets at ∼383 mcd (Figure 6). This is a critical horizon, as it marks the onset of the recovery of the CIE in both δ13CTOC and δ13CC29-n and the termination of the strictly laminated interval (Figure 4). Fe/S ratios are above unity in many samples from Core 30X (Table 1), pointing toward incomplete pyritization. This could be due to various reasons, such as a lack of sulfate (i.e., insignificant seawater influence), a lack of organic matter to metabolize (which is rather unlikely considering the high TOC content), or very high sedimentation rates. On the basis of high-resolution scans we see a reflection of this phenomenon from 383.0–383.2 mcd (Figure 6), where a triple peak in Fe counts is not accompanied by peaks in S counts. This may indicate the presence of a significant amount of freshwater, which would provide the required Fe, but is low in sulfate. This interval may represent an almost freshwater environment, which also agrees with the palynomorph assemblages, BIT index and Rock Eval Hydrogen Index data [Sluijs et al., 2006].
4.7. Orbital Forcing of Sedimentary Cycles
It has been widely accepted that XRF measurements usually show a significantly higher signal-to-noise ratio and a more consistent hole-to-hole agreement than any of the shipboard physical property measurements (e.g., density, color reflectance, magnetic susceptibility), and are therefore a great archive for time series analysis of relatively complete sections [Pälike et al., 2001; Röhl et al., 2003, 2004; Jaccard et al., 2005; Tjallingii et al., 2007; Westerhold et al., 2007]. While the uppermost Paleocene and lower Eocene just represents an incompletely recovered section (Figure 2) [Backman et al., 2006], the XRF scans for the recovered cores provide a suitable data set to further investigate the details of climatic signals in the record [Pälike et al., 2008; D. Spofforth et al., Palaeogene record of elemental concentrations in sediments from the Arctic Ocean obtained by XRF analyses, submitted to Paleoceanography, 2007]. Owing to the limited core recovery we restricted the time series analysis to the adjacent cores 27X and 28X (Figure 7) and removed the ETM2 horizon from the data in order to avoid spectral analysis disturbance. Early Eocene sediments between the PETM and ETM2 are meanwhile well known from other sections, such as the ODP Leg 208 sites [Lourens et al., 2005; Westerhold et al., 2007] and ODP Site 1051 [Cramer et al., 2003; Röhl et al., 2003], and therefore good control exists on the total duration of the interval. Spectral analyses of the interval prior to ETM2 (Figure 7) reveal related frequencies of the dominant cycles at 1.45, 4.5, and 7.1–7.6 cycles per meter which equal 0.7, 0.26, and 0.13–0.14 m cycle−1. The periodicities have a nearly perfect 1:2.5:5 frequency ratio and are diagnostic for a combination of the short (100 ka) eccentricity cycle, weak obliquity cycle (41 ka), but an especially pronounced and split precession (23.7 ka, 22.4 ka) cycle. This provides robust evidence for a match of the sedimentary rhythm to orbital forcing and a dominance of precession cycles modulated by eccentricity.
4.8. Organic Carbon Sequestration in the Arctic During the PETM?
As discussed above, sediment accumulation rates during the PETM increased threefold to fivefold, while on average wt % TOC are approximately 1% higher during the PETM (2.8%) than just prior to the PETM (∼2.0%; Figure 4). If we assume that anoxic conditions, along with high TOC concentrations, and elevated sedimentation rates were a basin-wide phenomenon, we can roughly estimate the scale of excess organic carbon burial. For this purpose we calculate the amount of organic carbon buried per 1000 years on the Lomonosov Ridge prior to and during the PETM, using % TOC, bulk density and above sedimentation rates. Given a bulk sediment density of 1.3 g cm−3 [Backman et al., 2006], sediment accumulation rates of 1.2 cm ka−1 and an average TOC content of 2.0 wt %, late Paleocene organic carbon burial rates at the drill site approximated 0.30 g C m−2 a−1. During the PETM, with a similar bulk sediment density [Backman et al., 2006], sediment accumulation rates of 5.0 cm ka−1 and an average TOC content of 2.8 wt %, carbon burial rates at the drill site approximated 1.81 g C m−2 a−1. This implies that organic carbon accumulation rates increased sixfold during the PETM. This increase was probably partly caused by improved preservation of organic matter because of anoxic conditions in the lower photic zone and on the seafloor. Moreover, marine productivity likely increased, as suggested by elevated numbers of dinocysts per gram of sediment (Figure 4) despite increased sedimentation rates (Figure 2), and trends in the δ13C of algal-derived C17 n-alkane [Pagani et al., 2006b].
The excess carbon burial during the PETM at the drill site approximated 1.51 g C m−2 a−1. Assuming a 170-ka duration of the CIE of the PETM [Röhl et al., 2007; Sluijs et al., 2007b], and that the estimated rates are valid for the entire 3,000,000 km2 Arctic Ocean, the excess burial of organic carbon was on the order of 770 Gt for the PETM. For comparison, a similar amount of carbon derived from methane hydrates is estimated to be required to generate the entire CIE in the global exogenic carbon pool [Dickens et al., 1995; Dickens, 2001].
We realize many uncertainties could compromise this calculation, such as the missing (exact) detailed correlation of the termination of the CIE to the deep marine record on which the 170 ka duration is based [Röhl et al., 2007], the exact size of the Arctic sedimentary basin, or the hypothesis that organic carbon burial changed similarly throughout the entire Arctic basin. However, this estimate for excess organic carbon burial during the PETM should be within an order of magnitude of the real burial. Additionally, organic-rich PETM successions have been recorded from the Tethyan realm [Bolle et al., 2000; Crouch et al., 2003; Gavrilov et al., 2003]. This indicates that not only carbonate burial [Zachos et al., 2005; C. M. John et al., Continental margin records of the Paleocene-Eocene thermal maximum: Implications for global carbon cycle and hydrological cycling, submitted to Paleoceanography, 2007], but also organic carbon burial in the Arctic and other restricted marine areas may have played a quantitatively important role in the sequestration of injected carbon during the PETM. Interestingly, this should have dampened the magnitude and perhaps the duration of the CIE in sedimentary carbon by increasing the organic versus inorganic carbon burial ratio. This would suggest that injected 13C-depleted carbon was even more depleted than assumed or that the volume of injected carbon was larger than has been estimated using δ13C analyses only [Dickens et al., 1995; Dickens, 2000, 2001; Cramer and Kent, 2005; Higgins and Schrag, 2006; Pagani et al., 2006a; Schouten et al., 2007b].
We reconstruct long-term latest Paleocene and early Eocene (∼57–50 Ma) environmental and climate trends in the Arctic Ocean, and focus on the Paleocene-Eocene thermal maximum (PETM) (∼55 Ma), using strata recovered from the Lomonosov Ridge by Integrated Ocean Drilling Program Expedition 302. Palynological and organic geochemical data suggest the drill site was close to shore during the latest Paleocene and early early Eocene and that parts of the Lomonosov Ridge were still subaerial. The site became located farther from shore during the late early Eocene, likely related to gradual subsidence of the Lomonosov Ridge. Organic dinoflagellate cyst (dinocyst) assemblages point to brackish and productive surface waters throughout the latest Paleocene and early Eocene. This suggests that low salinities in the Arctic surface waters were already prevalent prior to freshwater conditions that prevailed during the mass growth of the hydropterid fern Azolla during the earliest middle Eocene. Dinocyst assemblages are cosmopolitan throughout the latest Paleocene and early Eocene, suggesting warm sea surface temperatures, which is corroborated by TEX′86-reconstructed values of 14°–18°C.
Sediments from the upper Paleocene and lower lower Eocene are clay- and organic-rich without biogenic silica and XRF analyses show that total sulfur contents slightly exceed 2%, reflecting reducing conditions within the sediment. For the upper lower Eocene high trace element/Al ratios furthermore point toward euxinic conditions, possibly comparable to the Modern deep waters in the Baltic Sea.
The presence of ETM2 in the IODP Expedition 302 cores is confirmed by dinocyst stratigraphy. Spectral analysis reveals that the cyclicity, recorded in XRF scanning Fe data from early Eocene strata close to ETM2 is related to precession.
During the PETM bottom waters become more reducing, based on elevated sulfur and Fe/Al, Mn/Al, and trace element/Al ratios. Element ratios further evidence more intense weathering and a shift toward Al-rich (possibly kaolinite) and Fe-rich terrigenous detrital input. This is corroborated by a strong increase in sediment accumulation rates during the PETM, as well as higher absolute abundances of dinocysts (low-salinity-tolerant taxa) and terrestrial palynomorphs per gram of sediment. Our evidence supports a sea level rise across the onset of the PETM at this site. Finally, on the basis of total organic carbon content and variations in sediment accumulation rates, excess burial of organic carbon in the Arctic Ocean appears to have been quantitatively significant during the PETM and potentially contributed to the sequestration of the injected carbon.
This research used samples and data provided by the Integrated Ocean Drilling Program (IODP). Funding for this research was provided by Utrecht Biogeology Centre and the LPP Foundation to A. Sluijs and H. Brinkhuis, by the Netherlands Organisation for Scientific Research to A. Sluijs (VENI grant 863.07.001), S. Schouten, F. Sangiorgi, J. Sinninghe Damsté, and H. Brinkhuis, by the Deutsche Forschungsgemeinschaft (DFG) to U. Röhl and H.-J. Brumsack. We thank Ruediger Stein for discussions on organic carbon burial and Erica Crouch and Deborah Thomas for their insightful reviews. We thank the IODP Expedition 302 Scientific Party for discussions, N. Welters, J. van Tongeren, E. Hopmans, and V. Lukies for technical support, and W. Hale and A. Wülbers (IODP Bremen Core Repository) for core handling.