The Laptev Sea is the shallowest of the Arctic shelf seas with an average depth of 48 m (Jakobsson et al., 2004). It spans over 498 000 km2 with a volume of 24 000 km3 and is located between the Kara Sea and Severnaya Zemlya in the west and the East Siberian Sea and the New Siberian Islands in the east (Fig. 1). The main sources of POC for the Laptev Sea are terrigenous, both from coastal erosion and river runoff (Sánchez-García et al., 2011; Stein and Macdonald, 2004). Marine primary production is limited to on average 2 ice-free months per year and therefore generally low. Nutrient-poor waters on the Siberian shelves resulting from a strong stratification further impede phytoplankton growth (Sakshaug, 2004).

The destabilization of Pleistocene ice-complex deposits along the coastline is a main sediment source for the Laptev Sea (Rachold et al., 2000). The total POC input from coastal erosion to Laptev and East Siberian Sea is estimated to be between 4.0 Tg yr-1 (Semiletov, 1999; Stein and Fahl, 2000) and 22 ± 8 Tg yr-1 (including net sub-sea permafrost–carbon erosion; Vonk et al., 2012).

The Lena River is estimated to provide 20.7 Tg of sediment per year (Holmes et al., 2002), i.e., > 70 % of the total riverine input to the Laptev Sea (Gordeev, 2006) with an average water discharge of 588 km3 yr-1 (Holmes et al., 2012). It drains a watershed of ∼ 2.46 × 106 km2 (Holmes et al., 2012), of which 77 % is underlain by continuous permafrost (Amon et al., 2012). Water discharge peaks in June, during the spring flood, when about 75 % of total organic carbon (TOC) is delivered (Rachold et al., 2004). Total POC discharge by the Lena River can be up to 0.38 Tg yr-1 (Semiletov et al., 2011).

Sediment transport pathways are largely influenced by the prevailing atmospheric conditions: during cyclonic summers (i.e., positive phase of the Arctic Oscillation), northerly winds predominate, strengthening the Siberian Coastal Current, which transports Lena River water masses along the coast towards the East Siberian Sea; in contrast, during anticyclonic summers (i.e., negative phase of the Arctic Oscillation and mainly southerly winds) the Lena River plume is exported onto the mid-shelf and towards the deep part of the Arctic Ocean (Charkin et al., 2011; Dmitrenko et al., 2008; Guay et al., 2001; Wegner et al., 2013; Weingartner et al., 1999). Sediment transport in the Laptev Sea is strongly seasonal. The main transport of Lena River water with high concentrations of suspended particulate matter (SPM) towards the mid-shelf takes place shortly after river-ice breakup (Wegner et al., 2005). During the ice-free summer, SPM circulates between inner and mid-shelf with very little material escaping over the shelf break to the deeper parts of the Arctic Ocean. Significant sediment export is suggested to happen during freeze-up through SPM that is incorporated in sea ice and then transported across the continental margin (Dethleff, 2005; Eicken et al., 1997) or through the formation of dense bottom water resulting from brine ejection (Dethleff, 2010; Ivanov and Golovin, 2007). Hardly any sediment transport occurs beneath the ice cover.

Holocene-scale linear sedimentation rates for the Laptev Sea vary between 0.12 and 0.59 mm yr-1 according to 14C dating of marine bivalves (Stein and Fahl, 2004, and references therein), whereas centennial-scale 210Pb-derived rates for the more recent Laptev Sea can be up to 1.3 mm yr-1 (Vonk et al., 2012). These rates do not seem to be correlated with water depth on the shelf, but values for the continental slope and rise tend to be on the lower end (0.12–0.38 mm yr-1) (Stein and Fahl, 2004, and references therein).

Sediment sampling locations span from close to the Lena River mouth and in the Buor-Khaya Bay, across the shelf, to the continental slope and rise, covering a transect of > 800 km with water depths increasing by more than 2 orders of magnitude (Fig. 1). Samples SW-1, SW-2, SW-3, SW-4, SW-6, SW-14, SW-23 and SW-24 were collected during the SWERUS-C3 expedition on IB ODEN during summer 2014 using an Oktopus multicorer (eight Plexiglas tubes, 10 cm diameter). All other samples were collected during the International Siberian Shelf Study (ISSS-08) expedition onboard the R/V Yacob Smirnitskyi during summer 2008. The YS-4, YS-6, YS-13 and YS-14 samples were taken with a GEMAX gravity corer (two Plexiglas tubes, 9 cm diameter); YS-9 and TB-46 were collected with a Van Veen grab sampler. For the grab samples only surface sediments (uppermost cm) were subsampled and used in this study. Sediment cores were cut into 1 cm slices within 24 h after sampling. To account for lower sediment accumulation rates on the rise, for SW-1, SW-2, SW-3 and SW-4 a higher resolution of 0.5 cm for the top 10 cm was chosen. The depositional age for all samples is thus between ∼ 8 and ∼ 70 years (depending on which sedimentation rates are employed). All samples were kept frozen throughout the expedition and freeze-dried upon arrival to Stockholm University laboratories. See Semiletov and Gustafsson (2009) for more information on the ISSS-08 expedition. For exact sampling locations see Table 1.

All SA analyses have been performed on a Micromeritics Gemini VII Surface Area and Porosity analyzer. Freeze-dried subsamples of ∼ 0.7 g were heated at 400 ∘C for 12 h and gently cooled down to room temperature to remove all organic material. Keil and Cowie (1999) have shown that this method yields statistically similar results to the method using removal with sodium pyrophosphate/hydrogen peroxide (Mayer, 1994). The samples were then desalted by repeated mixing with 50 mL of Milli-Q water and centrifugation (20 min at 8000 rpm), followed by further freeze-drying. Directly prior to analysis they were degassed in a Micromeritics FlowPrep 060 Sample Degas System for 2 h at 200 ∘C under a constant nitrogen flow. Each analysis was initiated by measuring the free space in the vial. The specific SAs were derived from six pressure-point measurements (relative pressure p/p0= 0.05–0.3, equilibration time 5 s) with nitrogen as adsorbing gas (Brunauer et al., 1938). The instrumental precision was 0.1–0.3 m2 g-1, which corresponds to a relative uncertainty of about 1 %. The performance of the instrument was monitored with the SA reference material carbon black (21.0 ± 0.75 m2 g-1) provided by Micromeritics.

The mineral composition of ∼ 1 g freeze-dried, homogenized subsamples was also characterized with a wavelength dispersive sequential Philips PW2400 XRF spectrometer. Prior to the analysis, sediment samples were combusted for 12 h at 450 ∘C to remove the organic fraction. The XRF was operated under vacuum conditions on samples prepared as glass beads using lithium tetraborate and melted with a Claisse fluxer (∼ 1150∘C) (Mercone et al., 2001). The relative error was less than 0.6 % for major elements and less than 3 % for trace elements. In this study only SiO2, Al2O3 and CaO are reported.

Concentration and 13C isotopic composition of TOC were determined at the Stable Isotope Laboratory, Department of Geological Sciences, Stockholm University. Homogenized subsamples of ∼ 10 mg were repeatedly acidified (HCl, 1.5 M, Ag capsules) to remove carbonates (Nieuwenhuize et al., 1994). TOC concentrations and 13C isotopic composition were measured simultaneously with a Carlo Erba NC2500 elemental analyzer connected via a split interface to a Finnigan MAT Delta V mass spectrometer. TOC concentrations were blank corrected and the relative error was < 1 %. Stable isotope data are reported relative to VPDB using the δ13C notation.

Radiocarbon analyses of acidified samples were conducted at the US National Ocean Sciences Accelerator Mass Spectrometry (NOSAMS) facility of the Woods Hole Oceanographic Institution, USA, according to their standard routines (Pearson et al., 1998). The relative error of the measurements was < 0.5 %. Radiocarbon data are reported using the Δ14C notation following Stuvier and Polach (1977).

Bulk TOC concentrations decreased across the shelf with highest values (∼ 2 %) at shallow water depths and lowest values on the shelf edge (∼ 0.8 %); at high water depths (> 2000 m) concentrations were slightly higher (∼ 1 %) (Table 1). TOC values and the general pattern were in accordance with previous data from the Laptev Sea (Semiletov et al., 2005; Shakhova et al., 2015; Stein and Fahl, 2004; Vonk et al., 2012) and within the same range of those measured for the North American Arctic margin (Goñi et al., 2013).

Normalizing TOC concentrations to the mineral-specific SA helps to understand the influence of physical sorting and preferential deposition on the observed TOC trends since SA is correlated to the sediment grain size to a first-order approximation. To test if the mineral SA is altered by the input of autochthonous organisms with siliceous or carbonaceous skeleton (e.g., silicoflagellates/diatoms or foraminifera/shells respectively), the mineral composition of the sediments was examined by XRF analysis. There were no apparent trends with water depth for either SiO2 / Al2O3 or CaO / Al2O3 (Table 1); therefore, marine production is not expected to have a measurable effect and SA can thus be regarded as a conservative parameter. This was also confirmed by low biogenic silica concentrations for the Laptev Sea reported earlier (< 1.4 %, Mammone, 1998).

The relationship between TOC and SA has been widely studied on continental margins (e.g., Blair and Aller, 2012; Keil et al., 1994; Mayer, 1994). The TOC / SA ratios of typical river suspended sediments range between 0.4 and 1 mg m-2 (Mayer, 1994). TOC / SA ratios > 1 mg m-2 have been found in areas with high TOC supply (e.g., river outlets) and where the deposited organic matter had spent little time under oxic conditions (short OET) (Mayer et al., 2002). Ratios < 0.4 mg m-2 generally correspond to sediments from deeper parts of the ocean and long OETs (e.g., Aller and Blair, 2006). Accordingly, the TOC / SA values along the Laptev Sea transect displayed a strong decrease from 2.2 and 1.7 mg m-2 close to the Lena River delta (water depths of 11 and 7 m, respectively) to about 0.3 mg m-2 at water depths greater than 2000 m (Fig. 2a), proposing extensive TOC loss during cross-shelf transport.

Bulk TOC isotopes have been broadly used to distinguish between organic matter sources. Radiocarbon isotopes (14C) convey information about the age of organic material, with younger OC having higher Δ14C values. Marine organic matter produced primarily from CO2 is expected to have modern 14C signatures, whereas permafrost-derived TerrOC has aged both on land and during transport and has thus more depleted 14C values. The Δ14C values for our Laptev Sea transect were generally low (< -280 ‰, Fig. 2b, Table 1), suggesting a significant input of pre-aged TerrOC (as in Vonk et al., 2012). Bulk TOC showed less depleted Δ14C signatures with increasing distance from land on the shelf (from about -500 to about -340 ‰ on the outer shelf), reflecting a dilution of older TerrOC with younger marine material. On the slope and rise, however, Δ14C values decreased again to about -410 ‰. This difference may be a result of ageing during lateral transport and/or after deposition due to lower accumulation rates on slope and rise. The range between -340 and -410 ‰ corresponds to a Δ14C age difference of about 900 years; however, the depositional age differences between shelf and slope samples were estimated to be less than 80 years (see Sect. 2.2). Ageing after burial alone does therefore not explain the difference in Δ14C. Keil et al. (2004) estimated a lateral transport time of 1800 years across the Washington margin (158 km) from Δ14C data of bulk OC in surface sediments. For the > 200 km distance between mid-shelf and rise a bulk ageing of 900 years does therefore not seem unreasonable, particularly since the Washington margin, as opposed to the Laptev Sea shelf, is an active margin. The value from Keil et al. (2004) may therefore be regarded as a lower boundary. It has to be taken into account, however, that mainly the TerrOC fraction of the bulk OC is subject to such protracted lateral transport. Transport times would thus have to be significantly higher in order to explain this age difference for the entire bulk OC. One indication supporting this hypothesis of protracted lateral transport of TerrOC is the degradation status of TerrOC at the deep stations. All molecular degradation proxies point towards highly reworked material (see Sect. 3.3), suggesting that only the most refractory TerrOC fraction is found at great water depths off the continental margin. Alternatively, the lower Δ14C values at high water depths may be the consequence of more effective degradation of marine organic matter throughout the water column, resulting in a comparatively lower input of young autochthonous material. However, this latter scenario is not supported by the stable carbon isotopic signature, as values for δ13C increase from about -24.3 ‰ on the mid-shelf to about -22.5 ‰, suggesting a higher fraction of marine organic matter for the deep stations.

For stable carbon isotopes (13C), terrigenous sources are generally more depleted than marine organic matter (Fry and Sherr, 1984). In this study, values for δ13C of TOC ranged between -26.5 and -22.3 ‰. The trend towards more enriched TOC with increasing distance from the coast (Fig. 2b, Table 1) can be explained by a growing proportion of marine organic matter. However, the δ13C signature of the marine source appeared to be heavier than typical marine planktonic material in that region (-26.7 ± 1.2 ‰, Panova et al., 2015; -24 ± 3 ‰, Vonk et al., 2012, and references therein). One possible explanation for this discrepancy is an underestimated influence of ice algae that were reported to have highly enriched δ13C values between -15 and -18 ‰ (Schubert and Calvert, 2001). Significant seafloor deposition of ice algal biomass has been observed previously for the Arctic basins (Boetius et al., 2013). Another option would be a more refractory, isotopically enriched marine endmember (-21.2 ‰) as suggested by Magen et al. (2010). They argue that lighter isotopes are preferentially consumed by bacteria, which in turn enriches the remaining marine organic matter. Following their reasoning, the more enriched values observed for this transect may be interpreted as an increasing proportion of refractory marine organic matter.

Winterfeld et al. (2015b) analyzed surface water POC in the Lena River delta and found a mean δ13C of -29.6 ± 1.5 ‰. Karlsson et al. (2011) reported similarly depleted δ13C values for POC from the Buor-Khaya Bay (-29.0 ± 2.0 ‰), while their mean value for sedimentary OC for the same stations was significantly more enriched (-25.9 ± 0.4 ‰) and agreed well with our data for the shallow stations (-26.2 ± 0.3 ‰, stations YS-13, YS-14 and TB-46). Lena River POC δ13C values from high-discharge periods agree well with the more enriched values we found for the shallow stations (Rachold and Hubberten, 1998). Stein and Fahl (2004), Semiletov et al. (2011, 2012) and Vonk et al. (2012) presented similar δ13C ranges and trends for sediments from parts of the Laptev Sea as is reported in the current study for the entire width of the Laptev Sea shelf. For the Arctic Amerasian Continental shelf, Naidu et al. (2000) reported contrasts in absolute δ13C values comparing surface sediment samples from different regions, but all commonly displayed an increasing trend for δ13C values across the shelf, suggesting a growing fraction of marine organic matter with increasing distance from the coast.

Combining TOC / SA ratios with stable isotope signatures (δ13C) may serve to disentangle two different processes, which occur synchronously during cross-shelf transport (as in Keil et al. 1997a): (1) the net loss of TerrOC and (2) the replacement of TerrOC with autochthonous marine OC. Net loss of TerrOC, caused by either degradation or hydrodynamic sorting during transport, has been quantified previously using TOC / SA ratios (e.g., Aller and Blair, 2006; Keil et al., 1997a). The carrying capacity of inorganic particles for OC is assumed to be a function of the SA (Mayer, 1994); a decrease in TOC / SA values can therefore be regarded as TOC net loss.

Replacement of TerrOC with autochthonous marine OC does not change this ratio. However, since marine OC is known to be isotopically enriched in δ13C over TerrOC, this process is recorded by an increasing isotopic signature. Along the Laptev Sea transect, both processes seemed to play an important role (Fig. 2c). High TOC / SA values close to the Lena River decreased sharply outbound in the nearshore regime, pointing to extensive net loss, while the increase in δ13C values was minor in this area. Once TOC / SA ratios were < 0.8 mg m-2 (water depths > 20 m), the isotopic changes and thus the replacement of TerrOC with marine OC became increasingly important. Similar trends were observed for the Amazon River delta (Keil et al., 1997b).

However, the TOC / SA trend in the shallower sediments is likely driven by both degradation of OC bound to the mineral matrix during cross-shelf transport and sorting of vascular plant fragments that are retained in the inner shelf. A recent study (Tesi et al., 2016) showed that ∼ 50 % of the total OC pool in the inner Laptev shelf surface sediments exists in the form of large vascular plant fragments. They are trapped close to the coast due to their size and resulting settling (Stoke's law), while the OC bound to the fine mineral matrix is more buoyant and transported offshore towards deeper waters.

During degradation, syringyl and vanillyl phenol aldehydes are oxidized to carboxylic acids of the same phenol group. Increasing Sd / Sl and Vd / Vl ratios can therefore qualitatively indicate ongoing degradation of lignin phenols (Ertel and Hedges, 1984; Hedges et al., 1988). For fresh plant material typical acid-to-aldehyde ratios are around 0.1–0.2 (Hedges et al., 1988). Winterfeld et al. (2015a), however, found values as high as Sd / Sl = 0.80 and Vd / Vl = 0.67 for a moss species (Aulacomnium turgidum), Sd / Sl = 0.87 for larch (Larix) needles and Sd / Sl = 0.49 Vd / Vl = 0.41 for wild rosemary (Ledum palustre). Sedges (Carex spp.), dwarf birch (Betula nana) and willow (Salix) range between Sd / Sl = 0.13–0.24 and Vd / Vl = 0.18–0.23.

The ratio of CuO oxidation-derived 3,5-dihydroxybenzoic acid to vanillyl phenols (3,5-Bd / V) also serves as a proxy for degradation as 3,5-Bd is formed during humification likely occurring in soils (Gordon and Goñi, 2004; Hedges et al., 1988; Prahl et al., 1994; Tesi et al., 2014). For this reason, this proxy can trace mineral-rich soil organic matter in contrast to vascular plant debris (e.g., Dickens et al., 2007; Prahl et al., 1994) as well as degradation during cross-shelf transport (Tesi et al., 2016).

Sd / Sl, Vd / Vl and 3,5-Bd / V all increased along the transect, implying more degraded material with increasing residence time in the shelf system (Fig. 6a, Table 2). There appeared to be no differences between outer shelf/slope and rise, which may indicate that TerrOC on the slope is already highly reworked. In contrast, Tesi et al. (2014) found no correlation between Sd / Sl or Vd / Vl and distance from the coast, while 3,5-Bd / V significantly increased with increasing distance from the coast (Fig. S2 for comparisons with other studies). Sd / Sl values for the Buor-Khaya Bay from Winterfeld et al. (2015a) were slightly higher (1.04 ± 0.24) than samples from the corresponding water depths in this study (0.66 ± 0.15), whereas Vd / Vl values were significantly higher (1.28 ± 0.30 compared to 0.59 ± 0.14). Measurements for the Mackenzie Shelf agreed with the ones in this study (Sd / Sl = 0.81 ± 0.25 compared to 1.01 ± 0.33 for the corresponding water depths; Vd / Vl = 0.69 ± 0.14 to 0.86 ± 0.26; 3,5-Bd / V = 0.19 ± 0.04 to 0.31 ± 0.15) but did not show a trend with water depth (Goñi et al., 2000).

Tesi et al. (2016) observed lower acid/aldehyde ratios for the lignin-rich low-density fraction compared to the other fractions (high density with different grain sizes and settling velocities) in coastal surface sediments from the ESAS. With increasing distance from the coast, these values increased, whereas for the other fractions there were no apparent trends. These findings were interpreted as relatively fresh lignin in the low-density fraction (rich in large plant fragments) compared to the relatively degraded lignin that had likely experienced leaching and adsorbed to the fine mineral fractions (i.e., mineral-bound OC). Bulk 3,5-Bd / V values are potentially affected by both sorting and degradation, as they increased with decreasing particle size (fine and ultrafine fractions had the most degraded signal and are preferentially transported to the outer shelf) and across the shelf in each of the fractions.

The carbon preference indices for HMW n-alkanes and HMW n-alkanoic acids have also been widely applied as degradation proxies for plant waxes in marine sediments (for the ESAS, e.g., van Dongen et al., 2008; Fahl and Stein, 1997; Fernandes and Sicre, 2000; Vonk et al., 2010). It measures the ratio of odd-to-even numbers of carbon chain lengths of HMW lipids and is based on the preference of odd carbon chain lengths for HMW n-alkanes in fresh plant material (even carbon chain lengths for HMW n-alkanoic acids; Eglinton and Hamilton, 1967). With ongoing degradation this preference is lost and the CPI approaches 1 (Bray and Evans, 1961).

We observed that the HMW n-alkane CPI presented a similar pattern to that of the lignin-phenol-based degradation indices. However, the HMW n-alkanoic acid CPI did not show as much of a degradation trend (HMW n-alkane CPI: ∼ 5.7 close to the coast, ∼ 2.2 for the deep stations; HMW n-alkanoic acids: ∼ 5.4 close to the coast, ∼ 4.1 for the deep stations; Fig. 6b, Table 2). Karlsson et al. (2011) measured lipid CPIs in the Buor-Khaya Bay with 10–80 km distance to the coast and obtained similar results to this ∼ 800 km cross-shelf study, with higher values closer to the river delta (Fig. S2 for comparisons with other studies). Their data appear to have a wider spread, though, which might be due to the narrower dynamic range. Fahl and Stein (1997) also reported a large range of n-alkane CPI values (< 0.2 to > 5) for Laptev Sea sediments. Fernandes and Sicre (2000) analyzed sediments from the Kara Sea and from the major rivers discharging into this sea, the Ob and Yenisei rivers. In the marine environment and the Ob River, they observed HMW n-alkane CPI values between 4.8 and 5.3, similar to those found at shallow water depths in this study. For the Yenisei River and mixing zone, they found higher CPI values, pointing to fresher material being transported there. Vonk et al. (2010) recorded HMW n-alkane CPI values for sediments along the East Siberian Sea–Kolyma paleoriver transect (across the East Siberian Sea) shelf that decreased from > 7.5 to < 4.0 with increasing distance from the river mouth, overall higher than in this study but confirming the general trend to more degraded material on the outer shelf. Tesi et al. (2016) found HMW n-alkanoic acid CPI values to decrease with decreasing particle size with no significant trends across the shelf in all but the low-density fraction, which is largely retained close to the shore. The HMW n-alkane CPI values in that study, however, showed no systematical differences between different fractions, but an overall decreasing trend with increasing distance from the coast.

When undergoing degradation, HMW n-alkanoic acids may also lose their functional groups, turning them into HMW n-alkanes (Meyers and Ishiwatari, 1993). The slightly decreasing ratio of HMW n-alkanoic acids to n-alkanes also hints at more degraded material with increasing water depth, although, due to a rather large variability, this trend is not significant (Fig. 6b, Table 2). For the Buor-Khaya Bay surface sediments Karlsson et al. (2011) obtained similar results (0.48–10.7, here 1.1–10.9) with higher values closer to the river delta (Fig. S2 for comparisons with other studies). Along the Kolyma paleoriver transect, Vonk et al. (2010) measured HMW n-alkanoic acid to n-alkane ratios between 1 and 6 with no clear trend with increasing distance from the river mouth. Tesi et al. (2016) found decreasing values with increasing distance from the coast with no differences between the fractions. Two sediment cores from inner and outer East Siberian Sea recording about 1 century of sedimentation showed no clear trend in CPI or HMW n-alkanoic acid / n-alkane towards more degraded TerrOC with increasing sediment depth (Bröder et al., 2016), but displayed a similar difference between inner and outer shelf as seen in this study. This contrasting behavior for cross-shelf and down-core trends may be caused by significantly different timescales for the two processes: about a century in situ/after burial compared to potentially several millennia long lateral transport. Furthermore, the degradation efficiency is likely higher under the oxic conditions prevailing during cross-shelf lateral transport (Keil et al., 2004) than in the anoxic conditions that predominate below a few millimeters of sediments on the ESAS (e.g., Boetius and Damm, 1998). Comparing in situ to transport-related OETs on the wide Arctic shelves could potentially resolve the observed discrepancies.