Except for a few discontinuous fragments of the Late Cretaceous/Early Cenozoic climate history and depositional environment, the paleoenvironmental evolution of the pre-Neogene central Arctic Ocean was virtually unknown prior to the IODP Expedition 302 (Arctic Ocean Coring Expedition–ACEX) drilling campaign on Lomonosov Ridge in 2004. Here we present detailed organic carbon (OC) records from the entire ca. 200 m thick Paleogene OC-rich section of the ACEX drill sites. These records indicate euxinic “Black Sea-type” conditions favorable for the preservation of labile aquatic (marine algae-type) OC occur throughout the upper part of the early Eocene and the middle Eocene, explained by salinity stratification due to freshwater discharge. The superimposed short-term (“Milankovitch-type”) variability in amount and composition of OC is related to changes in primary production and terrigenous input. Prominent early Eocene events of algae-type OC preservation coincide with global δ13C events such as the PETM and Elmo events. The Elmo δ13C Event has been identified in the Arctic Ocean for the first time.
 A major element in the global climate evolution during Cenozoic times was the transformation from warm Paleocene/Eocene oceans with low latitudinal thermal gradients into the more recent modes of circulation characterized by strong thermal gradients, oceanic fronts, cold deep oceans and cold high-latitude surface waters [Miller et al., 1987; Zachos et al., 2001]. From this early “Greenhouse” time interval continuous central Arctic Ocean sedimentary records, allowing a development of chronologic sequences of climate and environmental change through Cenozoic times and a comparison with global climate records, however, were missing prior to the Integrated Ocean Drilling Program (IODP) Expedition 302 (Arctic Ocean Coring Expedition–ACEX) [Backman et al., 2006; Moran et al., 2006]. For the central Arctic Ocean in only four short cores obtained by gravity coring from drifting ice flows on the Alpha Ridge (Figure 1), some older pre-Neogene organic-carbon- rich sediments representing isolated, discontinuous fragments of the Late Cretaceous/Early Cenozoic climate history and depositional environment, have been recovered [Jackson et al., 1985; Clark et al., 1986]. In general, all these data suggest a warmer (ice-free) Arctic Ocean with strong seasonality and high paleoproductivity at that time [Clark et al., 1986; Jenkyns et al., 2004]. This gap in knowledge could partly be filled by ACEX. For the first time in the history of scientific ocean drilling, ACEX visited Lomonosov Ridge in the central Arctic Ocean (Figure 1), a continental fragment originally split off from the Eurasian continental margin at ∼56 Ma [Wilson, 1963; Kristoffersen, 1990]. As the Lomonosov Ridge moved away from the Eurasian plate and subsided, sedimentation on top of this continental sliver began, providing a unique Cenozoic stratigraphic sequence.
 The almost 430 m thick sedimentary sequence recovered from Lomonosov Ridge during ACEX range in age from Quaternary to Late Cretaceous [Backman et al., 2006; Moran et al., 2006]. Whereas recent papers on ACEX material were restricted to short-term events such as the global Paleocene/Eocene Thermal Maximum (PETM) event [Pagani et al., 2006; Sluijs et al., 2006] and the middle Eocene “Azolla” freshwater event [Brinkhuis et al., 2006], we investigated the entire 200 m thick Paleogene sequence of the ACEX record (200 and 405 mcd (meters composite depth)) (Figure 2). This interval consists of organic-carbon-rich sediments, mainly representing the late Paleocene to middle Eocene, i.e., a time interval when the Arctic Ocean was still quite isolated from the world ocean (Figure 1). The sediments are dominated by very dark gray silty clay (Subunit 1/6; middle Eocene), very dark gray mud-bearing partly finely laminated biosiliceous ooze (Unit 2; middle Eocene), and very dark gray partly finely laminated clay (Unit 3; early Eocene to late Paleocene) (Figure 2). Based on the existing age model [Backman et al., 2006], the sequence between 200 and 390 mcd represents the time interval between 44.1 and 56 Ma, resulting in average sedimentation rates of about 1 to 3 cm/ky (Figure 2).
 Using high-resolution records (based on about 650 samples) of total organic carbon (TOC), stable carbon isotopes of the organic matter (δ13Corg), total organic carbon/total nitrogen (C/N) ratios, Rock-Eval parameters, and total organic carbon/total sulfur (C/S) ratios, as well as organic petrography data, we were able to present for the first time detailed information on the amount, composition, and variability of organic-carbon deposition in the central Arctic and its paleoenvironmental significance during early Cenozoic times. All data are available at http://doi.pangaea.de/10.1594/PANGAEA.476007.
 Total organic carbon, total nitrogen, and total sulfur were determined by LECO elemental analyser technique. Stable carbon isotopes of the bulk organic matter (δ13Corg) has been determined by standard technique on ground carbonate-free sediments samples, using a Finnigan MAT Delta-S mass spectrometer equipped with a FLASH elemental analyser and a CONFLO III gas mixing system for the online determination of the carbon isotopic composition. The standard deviation (1σ) is generally better than δ13C = 0.15‰.
 As indicator for the composition of the organic matter in immature sediments, the hydrogen index (HI) obtained by Rock Eval pyrolysis, was used. The HI value corresponds to the quantity of hydrocarbons per gram TOC (mgHC/gC), generated by pyrolytic degradation of the kerogen during heating from 300°C up to 550°C. In immature sediments, HI values <100 mgHC/gC indicate a dominantly terrigenous (higher plant) source of the organic matter whereas HI values of ≫100 mgHC/gC indicate the presence of significant amount of aquatic algae (marine and/or freshwater) and/or microbial biomass. The immaturity of the organic matter, prerequisite for the use of the HI values as OC source indicator, was verified by the low Rock-Eval Tmax values of <435°C as well as vitrinite reflectance values <0.5% (see Auxiliary Material, Figure S1 and Table S1). More details about the analytical method and data interpretation are described by Tissot and Welte  and Meyers .
 To support the organic-geochemical bulk parameters, organic petrographical analysis has been carried out on a selected set of samples. Macerals, organic particles determined by microscopy under incident and fluorescent light, allow a precise and quantitative distinction of aquatic (marine or freshwater) and terrigenous organic matter (for details and classification see Taylor et al. ). Vitrinite/huminite, inertinite, detritus (vitrinite/huminite and inertinite with grains <10 μm) and liptinites (e.g., sporinite and cutinite) are counted as terrigenous particles, whereas aquatic particles are represented mainly by lamalginite, dinoflagellate cysts, and liptodetrinite and telalginite (e.g. Botryococcus or Pediastrum). Amorphous organic matter occurs as unstructured and weakly fluorescing bituminite partially including well preserved lamalginite.
3. Results and Discussion
3.1. Amount and Origin of Organic Carbon
 The upper Paleocene to middle Eocene ACEX sediments are characterized by high-amplitude variations in TOC content between 1 and 5%; only in the early Eocene between 370 and 375 mcd lower values of 0.5 to 1% occur (Figure 2). In general, TOC minima are higher in the biosiliceous oozes of Unit 2 than in the silty clays of Unit 3. In the overlying (middle Miocene?) sediments separated by a major hiatus from the Paleogene sequence and characterized by prominent gray and black color banding (“Zebra” Subunit 1/5; 195–200 mcd), very high TOC values of 5 to 14.5% were determined.
 In order to interpret the TOC record in terms of paleoenvironmental conditions, additional information about the composition of the organic-carbon fraction is needed. As one parameter for classification of the organic carbon (OC) source, i.e., to distinguish between terrigenous (higher plant) and aquatic (freshwater and/or marine) OC, the hydrogen index (HI) derived from Rock-Eval pyrolysis [Tissot and Welte, 1984], was determined (see methods). Elevated HI values of 100 to 350 mgHC/gC occur in the middle Eocene sediments (Unit 2) and the upper part of the early Eocene (220–345 mcd) (Figure 2). These values increase to 250 to 450 mgHC/gC when considering the dead OC present in the entire record with a background value of about 0.45% TOC (see Auxiliary Material, Figure S2), indicating that hydrogen-rich algae-type OC of aquatic (i.e., marine, brackish to freshwater) origin occurs in significant to major amounts. Based on the correlation between HI values and kerogen microscopy data determined in immature Mesozoic/Cenozoic marine sediments [Stein et al., 1986] and first maceral data obtained from ACEX sediment samples, about 40 to 80% of the OC seems to be of aquatic origin (see Auxiliary Material, Figure S1). In the underlying early Eocene and–especially–late Paleocene as well as the upper part of the overlying middle Eocene section (Unit 1/6), all characterized by low HI values of 50 to 150 mgHC/gC (Figure 2), terrigenous OC seems to be dominant (about 80% of the OC). The OC of the late Paleocene interval displays a characteristic similar to surface sediments on Lomonosov Ridge [Stein et al., 2004], as also supported by kerogen microscopy data showing a predominance of terrigenous macerals (see Auxiliary Material, Figure S1 and Table S1). Two exceptions from this general picture, however, are obvious. Peak values of HI reaching 250 to >300 mgHC/gC and indicating the presence of major amount of algae-type OC also occur around 385 mcd (Paleocene/Eocene boundary) and 368 mcd (early Eocene). Similar peak HI values were determined in Subunit 1/5 (195–200 mcd; Middle Miocene?). All three intervals with peak HI values coincide with prominent minima in the δ13Corg record (Figures 2 and 3) .
 As additional proxies for the OC composition, δ13Corg values and TOC/total nitrogen (C/N) ratios are often used [e.g., Meyers, 1997, and references therein]. C/N ratios of marine organic matter (mainly phytoplankton and zooplankton) are around 6, whereas terrigenous organic matter (mainly from higher plants) has C/N ratios of >15. Light isotopic δ13Corg values of −26 to −28‰ are typical of terrigenous organic matter (i.e., land plants using the C3 pathway of photosynthesis), and heavy δ13Corg values of −20 to −22‰ are given as characteristic for marine organic matter (marine algae) (Figure 3). It has to be mentioned, however, that δ13Corg values of marine phytoplankton may be influenced by several factors such as higher concentration of dissolved CO2, cell growth rate, cell size, and cell membrane CO2 permeability, resulting in much lighter δ13Corg values more similar to the δ13Corg values of terrigenous OC [Rau et al., 1989; Goericke and Fry, 1994].
 The δ13Corg and C/N records of the ACEX section do not follow the same behavior as described above. Heaviest δ13Corg values were determined in the late Paleocene (−24 to −22.5‰) and lower early Eocene (−26 to −24‰), i.e., in intervals with a clear dominance of terrigenous OC (Figure 3). In Subunit 1/6 (middle Eocene) showing variations between dominantly terrigenous OC and intervals with more aquatic OC, the δ13Corg values of the samples with a dominance of terrigenous OC are between −27 and −29‰, i.e., they are more similar to the carbon isotope ratio of modern (C3) plants. That means, there seems to be a trend from more heavy to more light δ13Corg values through time (Figure 3), possibly related to changes in plant communities or physiology [Lou et al., 1996; Dean and Arthur, 1999].
 The middle Eocene (Unit 2) and upper part of the early Eocene (upper Unit 3) characterized by the dominance of algae-type OC, display significantly lower δ13Corg values ranging between −26 and −29‰ (Figures 2 and 3). A very similar inverted relationship, i.e., heavier δ13Corg values of −24.5 to −26‰ coinciding with dominantly terrigenous OC and lighter δ13Corg values of −27 to −28.5‰ coinciding with dominantly algae-type (marine) OC, is also reported from various sites in the early Cretaceous North Atlantic (Figure 3) [Dean et al., 1986; Dean and Arthur, 1999; Langrock et al., 2003]. Our light middle to upper early Eocene values agree very well with the global record of average δ13Corg values of marine OC, giving −27 to −29‰ for the time interval between 45 and 55 Ma [Hayes et al., 1999]. This may suggest that major proportions of the algae-type OC in the middle Eocene interval of the ACEX record is of marine origin. A dominantly marine source of the algae-type OC is also supported by the predominance of marine diatoms [Backman et al., 2006]. As explanation for the light δ13Corg values of marine OC it has been postulated that Cretaceous and early Cenozoic marine photosynthesis resulted in greater isotope fractionation, possible because of higher CO2 concentrations in the atmosphere and oceans [Dean et al., 1986; Hayes et al., 1999; Zachos et al., 2001]. When interpreting the ACEX records, however, it has to be considered that isotopically light organic carbon may also be added to the pool of OC by sulfide-oxidizing bacteria living in euxinic environments, as described for the modern Black Sea [Arthur et al., 1994] and postulated for the middle to upper early Eocene Arctic Ocean from our data (see below).
 Furthermore, the ACEX δ13Corg record displays three prominent minima with values of −30 to −32‰, characterized by increased abundance of aquatic OC (Figures 2 and 3). The oldest one represents the PETM event already described in detail elsewhere [Pagani et al., 2006; Sluijs et al., 2006] and probably associated with a massive greenhouse gas input [Dickens et al., 1995]. A second similar event, although of smaller magnitude, occurs at about 368 mcd (early Eocene) and may correlate with the “Elmo Event” probably representing another global thermal maximum [Lourens et al., 2005]. The third interval with very light δ13Corg values is from the middle Miocene (?) “Zebra” Subunit 1/5 not further discussed in this paper due to its questionable stratigraphy.
 C/N ratios of the OC-rich middle to upper early Eocene section characterized by hydrogen-rich aquatic-type organic matter, are unusually high (12 to >30; Figure 2). Often maximum C/N ratios coincide with TOC and HI maxima. Similar high C/N ratios atypical for algal-source OC were also determined in Mediterranean sapropels, late Neogene OC-rich sediments from upwelling areas as well as Cenomanian-Turonian black shales, all characterized by the dominance of marine OC [Meyers, 1997; Twichell et al., 2002]. In these sediments, the high C/N ratios are explained by the fact (1) that algae are able to synthesize lipid-rich OC during times of abundant nutrient supply, i.e., high primary productivity, and/or (2) that during sinking, partial degradation of algal OC may selectively diminish N-rich proteinaceous components, and raise the C/N ratio [Meyers, 1997]. Both factors, i.e., increased preservation of algae-type OC and increased primary production, probably also have caused the high C/N ratios in our ACEX record (see below). Maximum C/N ratios occur in the interval representing the “Azolla Event” (Figure 2), where enormous quantities of the free-floating fern Azolla were found, implying nearly fresh surface water conditions [Brinkhuis et al., 2006]. Such high C/N ratios were also determined in lake sediments and related to significant contribution of freshwater Botryococcus-type algae [Fuhrmann et al., 2003], which may support the low-salinity (almost freshwater) conditions in the Arctic Ocean at that time. The distinct C/N maximum at 396–400 mcd (late Paleocene), on the other hand, which coincides with maximum TOC values and very low HI values, probably indicates a terrigenous OC origin deposited under oxic conditions (see below).
3.2. Depositional Environment: Anoxia and Primary Productivity
 The organic carbon/sulfur (C/S) ratio [Leventhal, 1983], occurrence of small-sized pyrite framboids [Wilkin et al., 1996], and sedimentary structure (fine lamination) were used to get information about the oxygenation of bottom water. In euxinic environments such as the modern Black Sea, H2S already occurs in the water column and framboidal pyrite can be initially formed, resulting in an excess of sulfur (indicated by a positive S intercept in the C/S diagram; see Auxiliary Material, Figure S3) and very low (typically <1) C/S ratios [Leventhal, 1983]. The interval with very low C/S in the ACEX record coincides with significant occurrence of pyrite (see Figure 2), supporting the interpretation of the C/S ration in terms of oxygenation of water mass.
 Based on the high C/S ratios, oxic conditions occurred during the late Paleocene (Figures 2 and S3). Terrigenous OC partly composed of reworked older material, was predominant (see Auxiliary Material, Figure S1). Benthic foraminifers indicate a deposition in shallow-marine neritic environments [Backman et al., 2006]. With the PETM event, a major change to euxinic conditions is obvious, as indicated by a drastic decrease in the C/S values (Figure 2) and the occurrence of fine lamination [Backman et al., 2006]. During the PETM, euxinic conditions expanded even into the photic zone as suggested from the occurrence of the biomarker isorenieratene related to photosynthetic green sulfur bacteria which requires euxinic conditions to thrive [Sinninghe-Damsté et al., 1993; Sluijs et al., 2006]. Possible cause for the euxinic conditions was on one hand a salinity stratification suggested from the high abundance of low-salinity dinocysts [Sluijs et al., 2006]. Because these dinocysts were already abundant prior to the PETM event, i.e., during times of oxic water-mass conditions, increased flux of algae-type OC (indicated by increased HI values; Figure 2) due to enhanced primary production was needed as additional factor causing the euxinic conditions. The increased primary production was probably related to increased fluvial nutrient supply at that time [Pagani et al., 2006]. Based on our records (Figures 2 and 3), similar conditions occurred during the early Eocene isotope event (Elmo Event [see Lourens et al., 2005]) at 368 mcd.
 Based on the continuously low C/S ratios and the abundance of pyrite (Figure 2) as well as the occurrence of fine lamination [Backman et al., 2006], euxinic conditions favorable for the preservation of labile algae-type OC, probably occurred throughout the early–middle Eocene Arctic Ocean, except for a short interval of oxic conditions directly following the PETM. In the still limited set of Eocene and PETM samples used for kerogen microscopy we found finely dispersed small-sized (Ø: 5μm) pyrite framboids (see Auxiliary Material, Table S1), which supports in-situ formation of the framboids in an euxinic water column [Wilkin et al., 1996]. The environment may have been similar to the modern Black Sea [Leventhal, 1983]. These euxinic conditions were most probably caused by widespread salinity stratification in the isolated Arctic Ocean at that time. A low surface-water salinity (brackish) environment during middle Eocene times is also suggested from the rare and sporadic occurrence of radiolarians [Backman et al., 2006]. Runoff-related low salinity might be indicated as well by the abundance of terrestrial palynomorphs and green algae such as Tasmanites and Botryococcus [Backman et al., 2006]. Warm surface-water temperatures [Brinkhuis et al., 2006; Sluijs et al., 2006], and a high-productivity environment suggested from the abundance of marine diatoms and diatom resting spores [Backman et al., 2006] and causing increased flux of labile OC, acted as further processes to obtain an oxygen-deficient environment. The latter was probably the most important process to explain the high-amplitude short-term variability in amount and composition of OC (Figure 2). Taken the preliminary age model and mean sedimentation rates [Backman et al., 2006], and the TOC data from the upper, most densely sampled interval between 44.1 and 46.2 Ma (215 to 275 mcd), these short-term oscillations seems to have periods close to the Milankovitch bands. Thus, short-term changes in primary production controlled by fluvial nutrient supply in combination with variable terrigenous input may have been a superimposed process causing the short-term “Milankovitch-type” OC cycles. As additional factor nutrient supply by upwelling triggering increased primary production, has to be taken into account [Backman et al., 2006].
 Based on the C/S ratios, euxinic conditions continued into Subunit 1/6 (middle Eocene) and may have become even more intensive as indicated by maximum sulfur excess values (see Auxiliary Material, Figure S3) and maximum pyrite abundance (Figure 2). Despite intensive euxinic conditions, HI values indicate a dominance of terrigenous OC. This is explained by a major increase in supply of terrigenous OC at that time. Further detailed inorganic and organic geochemistry studies are in process to prove this interpretation.
 This research used samples and data provided by the Integrated Ocean Drilling Program (IODP). Funding by the German Research Foundation (DFG) is gratefully acknowledged (grant STE 412/22-1).