Pore water geochemistry along continental slopes north of the East Siberian Sea: inference of low methane concentrations

Continental slopes north of the East Siberian Sea potentially hold large amounts of methane (CH4) in sediments as gas hydrate and free gas. Although release of this CH4 to the ocean and atmosphere has become a topic of discussion, the region remains sparingly explored. Here we present pore water chemistry results from 32 sediment cores taken during Leg 2 of the 2014 joint Swedish–Russian–US Arctic Ocean Investigation of Climate–Cryosphere–Carbon Interactions (SWERUS-C3) expedition. The cores come from depth transects across the slope and rise extending between the Mendeleev and the Lomonosov ridges, north of Wrangel Island and the New Siberian Islands, respectively. Upward CH4 flux towards the seafloor, as inferred from profiles of dissolved sulfate (SO2− 4 ), alkalinity, and the δ13C of dissolved inorganic carbon (DIC), is negligible at all stations east of 143 E longitude. In the upper 8 m of these cores, downward SO2− 4 flux never exceeds 6.2 mol m−2 kyr−1, the upward alkalinity flux never exceeds 6.8 mol m−2 kyr−1, and δ13C composition of DIC (δ13CDIC) only moderately decreases with depth (−3.6 ‰ m−1 on average). Moreover, upon addition of Zn acetate to pore water samples, ZnS did not precipitate, indicating a lack of dissolved H2S. Phosphate, ammonium, and metal profiles reveal that metal oxide reduction by organic carbon dominates the geochemical environment and supports very low organic carbon turnover rates. A single core on the Lomonosov Ridge differs, as diffusive fluxes for SO2− 4 and alkalinity were 13.9 and 11.3 mol m−2 kyr−1, respectively, the δ13C-DIC gradient was 5.6 ‰ m−1, and Mn2+ reduction terminated within 1.3 m of the seafloor. These are among the first pore water results generated from this vast climatically sensitive region, and they imply that abundant CH4, including gas hydrates, do not characterize the East Siberian Sea slope or rise along the investigated depth transects. This contradicts previous modeling and discussions, which due to the lack of data are almost entirely based on assumption.

Regional assessments for abundant CH 4 in marine sediment along continental slopes can be acquired through two general approaches.The first includes geophysical applications, primarily seismic reflection profiling and the recognition of BSRs (Kvenvolden, 1993;Carcione and Tinivella, 2000;Haacke et al., 2008), which are a common, but not ubiquitous, feature of hydrate-bearing sediments.The second utilizes chemical analyses of pore waters obtained from sediment cores (Borowski et al., 1999;D'Hondt et al., 2003).In marine sediments with abundant CH 4 , a general process occurs near the seafloor.Microbes utilize upward migrating CH 4 and downward diffusing sulfate (SO 2− 4 ) via anaerobic oxidation of methane (AOM; Barnes and Goldberg, 1976;Boetius et al., 2000): The reaction leads to characteristic pore water chemistry profiles, with a clearly recognizable sulfate-methane transition (SMT; Fig. 3).The depth of the SMT inversely relates to the flux of CH 4 , which in turns relates to the distribution of CH 4 beneath the seafloor (Borowski et al., 1999;Dickens, 2001;Bhatnagar et al., 2011).Where CH 4 fluxes toward the seafloor are high, the SMT is located at shallow depth.For example, in cores from the continental shelf and slope of the Beaufort Sea, where seismic profiles indicate gas hydrate, Coffin et al. (2008Coffin et al. ( , 2013) ) have documented SMTs in shallow sediment (< 10 mbsf).
The joint Swedish-Russian-US Arctic Ocean Investigation of Climate-Cryosphere-Carbon Interactions (SWERUS-C3) project is aimed at understanding spatial changes in carbon cycling across the continental margin north of Siberia.A central theme concerns the amount, distribution, and fluxes of CH 4 .The overall project included a two-leg expedition in the boreal summer of 2014 using the Swedish icebreaker IB Oden.Between 21 August and 5 October, Leg 2 sailed between Barrow, Alaska, and Tromsø, Norway, including surveys of the continental slope of the East Siberian Sea.SWERUS Leg 2 included geophysical mapping and retrieval of numerous sediment cores, of which 446 pore water samples from 8 piston, 7 gravity, and 17 multicores (Fig. 2) were studied to ascertain potential fluxes of CH 4 toward the seafloor.

East Siberian margin geology
Extensive continental shelves and their associated slopes encircle the Arctic Ocean (Fig. 1).Although only 2.6 % of the world's ocean by area (Jakobsson, 2002), the present Arctic Ocean receives ∼ 10 % of global freshwater input (Stein, 2008) as well as a massive discharge of terrigenous material (> 249 Mt yr −1 ; Holmes et al., 2002).Only Fram Strait (Fig. 1), with a modern sill depth of about 2540 m taken from the International Bathymetric Chart of the Arctic Ocean (Jakobsson et al., 2012), allows deep-water flow to and from the Arctic Ocean.It opened during the early to middle Miocene (Jakobsson et al., 2007;Engen et al., 2008;Hustoft et al., 2009).Prior to this, the Arctic Ocean was connected to other oceans only through shallow seaways (e.g., Turgay Strait), such that deep waters may have been anoxic for long intervals of the Cretaceous and Paleogene (Moran et al., 2006;Sluijs et al., 2006;Jakobsson et al., 2007;O'Regan et al., 2011).
The East Siberian Sea stretches between Wrangel Island to the east and the New Siberian Islands to the west (Fig. 2).The continental shelf within this region is the widest in the world, extending 1500 km north of the coast.North of this expansive shelf lies the continental slope, which connects to the Mendeleev Ridge to the east and the Lomonosov Ridge to the west (Jakobsson et al., 2012).As these slopes lie north of the East Siberian Sea proper, we hereafter refer to them as SNESS for convenience.Ryan et al., 2009).Observed sulfatemethane transitions during the MITAS 1 expedition shown in black diamonds (Coffin et al., 2013) and Arctic Coring Expedition (ACEX) shown as red squares (Backman and Moran, 2009).
Despite the paucity of ground-truth data, many researchers have predicted widespread and abundant CH 4 within SNESS, as clearly shown by maps of inferred Arctic gas hydrate distribution (Fig. 1).This inference has arisen for two main reasons.First, the integrated input of particulate organic carbon (POC) over time provides the ultimate source of CH 4 in marine sediments (Kvenvolden and Grantz, 1990).Arctic slopes may contain high POC contents, which accumulated (a) in shallow platform environments prior to the opening of the Amerasian Basin, (Spencer et al., 2011) (b) during periods of high surface water productivity and oxygen poor bottom water conditions that persisted across much of the Arctic un- til the opening of the Fram Strait in the Neogene (Jakobsson et al., 2007;Stein et al., 2006;O'Regan et al., 2011;Jokat and Ickrath, 2015), or (c) as terrigenous material carried to or deposited along the slopes during interglacial intervals of the Quaternary (Danyushevskaya et al., 1980;Darby et al., 1989;Archer, 2015).Certainly, organic-rich Cretaceous and Eocene sediments have been documented on other Arctic margins and in the ACEX cores on the Lomonosov Ridge (Moran et al., 2006;Backman and Moran, 2009;O'Regan et al., 2011).The second reason is that the thickness of the gas hydrate stability zone depends on bottom water temperature and the geothermal gradient (Dickens, 2001).Because of very low bottom water temperatures along the slope, and generally low regional geothermal gradients (O'Regan et al., 2016), an extensive volume of sediment can host gas hydrate (Miles, 1995;Makogon, 2010).

Pore water chemistry above methane-charged sediment
Pore water chemistry provides powerful means to constrain CH 4 abundance and fluxes in marine sediment (Borowski et al., 1996;Berg et al., 1998;Jørgensen et al., 2001;Torres and Kastner, 2009;Treude et al., 2014).At locations without significant fluid advection, pore water profiles relate to Fick's law of diffusion and chemical reactions (e.g., Berner, 1977;Froelich et al., 1979;Klump and Martens, 1981;Boudreau, 1997;and Iverson and Jorgensen, 1993).The flux (J ) of a dissolved species through porous marine sediment can be calculated from the concentration gradient by (Li and Gregory, 1974;Berner, 1975;Lerman, 1977) where ϕ is porosity, D s is the diffusivity of an ion in sediment at a specified temperature, C is concentration, and Z is depth.Note that, as generally written, J is positive for upward fluxes and negative for downward fluxes relative to the seafloor.At many locations, ϕ and D s change only moderately (< 20 %) in the upper tens of meters below the seafloor.However, abundant CH 4 in sediment leads to a large concentration gradient toward the seafloor and an upward flux of CH 4 .The consequent reaction with SO 2− 4 via AOM (Eq. 1) leads to a series of flux changes in dissolved components (addition or removal) and predictable variations in concentration profiles across an SMT (Alperin et al., 1988;Borowski et al., 1996;Niewohner et al., 1998;Ussler and Paull, 2008;Dickens and Snyder, 2009;Chatterjee et al., 2011).Furthermore, the depth of the SMT directly relates to the flux of CH 4 from below (Jørgensen et al., 1990;Dickens, 2001;D'Hondt et al., 2002;  Seifert and Michaelis, 1991;D'Hondt et al., 2002) and 1230 (offshore Peru; Donohue et al., 2006) are given as reference.Hensen et al., 2003), largely because SO 2− 4 concentrations at the seafloor are nearly constant throughout the oceans.
Large areas of continental slopes across the world host CH 4 in sediment and consequently have a prominent SMT (D'Hondt et al., 2002).This feature is generally within the upper 30 m beneath the seafloor and is characterized as a thin (< 3 m) horizon with major inflections in both CH 4 and SO 2− 4 profiles (Fig. 3).Sulfate concentrations decrease from seawater values at the seafloor to near zero at the SMT; by contrast, CH 4 concentrations rise from zero at the SMT to elevated values at depth.Importantly, though, as one can infer from Eqs. ( 1) and (2), AOM affects additional species dissolved in pore water (Alperin et al., 1988;Jørgensen et al., 1990;Dickens, 2001;Hensen et al., 2003;Snyder et al., 2007).Dissolved HS − and HCO − 3 concentrations necessarily increase across the SMT, so an inflection occurs in their concentration profiles.These two species contribute to total alkalinity of marine waters (Gieskes and Rogers, 1973;Haraldsson et al., 1997), which can be defined as where X refers to several minor species.However, in shallow sediments found above almost all CH 4 -charged systems, this can be expressed as Therefore, because of the production of HS − and HCO − 3 , an inflection in Alk T occurs across the SMT (Luff and Wallmann, 2003;Dickens and Snyder, 2009;Jørgensen and Parkes, 2010;Chatterjee et al., 2011;Smith and Coffin, 2014;Ye et al., 2016).
It should be noted that at seafloor locations with significant upward advection of fluids, such as at seeps and vents, the aforementioned reactions occur, but the pore water profiles become more complicated to model (Berner, 1980;Torres et al., 2002;Chatterjee et al., 2011).This is because the advecting fluids typically have different chemistry than surrounding sediment (even if charged with CH 4 ) and because advection often involves multiphase fluid flow (free gas and liquid) that may be episodic.Nonetheless, at least on continental slopes, if the upward advecting fluids contain significant CH 4 (even as gas bubbles), a prominent SMT occurs but is shoaled toward the seafloor with respect to predictions based on CH 4 diffusion alone (Luff and Wallmann, 2003;Kastner et al., 2008a).Indeed, at locations where CH 4 gas bubbles escape the seafloor, the SMT lies at the seafloor (e.g., Aharon and Fu, 2000;Joye et al., 2004;Hu et al., 2010).

SWERUS-C3 expedition, Leg 2
Leg 2 of SWERUS-C3 included four transects across the SNESS (Fig. 2).These transects were along Arlis Spur (Tr1), the north of central east Siberia (Tr2), from close to Henrietta Island to the Makarov Basin (Tr3), and on the Amerasian side of the Lomonosov Ridge (Tr4).Along each transect, scientific operations involved bathymetric mapping as well as sediment coring at stations.An additional coring station was located on the Lomonosov Ridge, near its intersection with the Siberian margin.
An array of coring techniques were used along each transect.In total, 50 sediment cores were collected at 34 stations.These included multicore sets ( 22), gravity cores (23), piston cores (11), and kasten cores (2).The multicorer was an eighttube corer built by Oktopus GmbH.The polycarbonate liners were 60 cm long with a 10 cm diameter.The piston and gravity coring system was built by Stockholm University with an inner diameter of 10 cm.Trigger weight cores were also collected during piston coring.The different coring systems enabled sediment and pore water collection from the seafloor to upwards of 9 m below the seafloor (mbsf).

Core material
For gravity and piston cores, physical properties were analyzed on the ship using a Geotek Multi-Sensor Core Logger (MSCL).These included measurements of the gamma-ray derived bulk density, compressional wave velocity (P wave), and magnetic susceptibility at 1 cm resolution.Discrete samples (2-3 per section) were taken for sediment index property measurements (bulk density, porosity, water content, and grain density).Grain density was measured using a helium displacement pycnometer on oven-dried samples.Porosity profiles were generated using the smoothed (3 pt) MSCLderived bulk density (ρ b ) and the average grain density (ρ g ) from each core, where and the pore fluid density (ρ f ) was assumed to be 1.024 g cm −3 .

Interstitial water collection
Cores were cut into ∼ 1.5 m long sections immediately on the ship deck, brought to the geochemistry laboratory, and placed on precut racks.Laboratory temperature was a near constant 22 • C. Pore waters were collected using Rhizon samplers Notes: PE = percent error; PD = percent difference; RSD = relative standard deviation; BDL = below detection limit.(Seeberg-Elverfeldt et al., 2005;Dickens et al., 2007).Sampling involved drilling holes through the core liner, inserting Rhizons into the sediment core, and obtaining small volumes of pore water via vacuum and "microfiltration."The Rhizons used were 5 cm porous flat tip male Luer locks (19.21.23) with 12 cm tubing, purchased from Rhizosphere Research Products (www.rhizosphere.com).
In total, 529 pore water samples were collected from 32 cores, which ranged from 0.16 to 8.43 m in length (Table S2 in the Supplement).Rhizons in gravity and piston cores typically were spaced every 20 to 30 cm.Because the use of Rhizon sampling for collecting pore waters of deep-sea sediments remains a relatively novel and engaging topic (Dickens, 2007;Xu et al., 2012;Miller et al., 2014), we discuss the procedure in detail, as well as several experiments regarding our sampling, in the Supplement.
While in the shipboard laboratory, Rhizon samples were divided into six aliquots when sufficient water was available.This sample splitting led to 2465 aliquots of pore water in total, which then could be examined for different species at different laboratories.Aliquots 1, 3, and 6 (below) were collected for all 32 cores.

Interstitial water analyses
The first aliquot was used to measure Alk T using a Mettler Toledo titrator on IB Oden.Immediately after collection, pore water was diluted with Milli-Q water and auto-titrated.Fifteen spiked samples and eight duplicates were analyzed onboard for quality control.Spiked samples were created by pipetting certified reference material (Batch 135; www.cdiac.ornl.gov/oceans/Dickson_CRM)into Milli-Q water.Results for spiked samples and duplicates are reported in Table 1.
The second aliquot was used to measure the δ 13 C composition of DIC.Septum-sealed glass vials prepared with H 3 PO 4 and flushed with helium were prepared before the ex-pedition.Samples were sealed in boxes and refrigerated for the remainder of the cruise.Four field duplicates, two seawater standards, and a field blank were collected, stored, and analyzed with the samples.The δ 13 C-DIC analyses were performed on a Gasbench II coupled to a MAT 253 mass spectrometer (both Thermo Scientific) at Stockholm University.The δ 13 C-DIC is reported in conventional delta notation relative to Vienna PeeDee Belemnite (VPDB).Results for field duplicates and standards are reported in Table 1.Standard deviation for the analyses of δ 13 C-DIC was less than 0.1 ‰.
The third aliquot was used to measure dissolved sulfur and metal concentrations.Samples were acid preserved with 10 µL ultrapure HNO 3 .Additionally, 11 blind field duplicates and 2 field blanks were collected and processed in the same manner.Concentrations of Ba, Ca, Fe, Mg, Mn, S, and Sr were determined on an Agilent Vista Pro inductively coupled plasma atomic emission spectrometer (ICP-AES) in the geochemistry facilities at Rice University.Known standard solutions and pore fluid samples were diluted 1 : 20 with 18 M water.Scandium was added to both standards and samples to correct for instrumental drift (emission line 361.383nm).Wavelengths used for elemental analysis followed those indicated by Murray et al. (2000).Following initial analysis, an additional dilution, 1 : 80 with 18 M water, was analyzed for Ca, Mg, and S.After every 10 analyses, an International Association for the Physical Sciences of the Oceans (IAPSO) standard seawater spiked sample and a blank were examined for quality control.Relative standard deviations (RSD) from stock solutions are reported in Table 1.
The fourth aliquot was used to measure dissolved ammonia (NH + 4 ) via a colorimetric method similar to that presented by Gieskes et al. (1991).Set volumes of pore water were pipetted into cuvettes and diluted with Milli-Q water.Two reagents were then pipetted into the cuvettes.
www.biogeosciences.net/14/2929/2017/Biogeosciences, 14, 2929-2953, 2017 Reagent A was prepared by adding Na 3 C 6 H 5 O 7 , C 6 H 5 OH, and Na 2 (Fe(CN) 5 NO) to Milli-Q water.Reagent B was prepared by dissolving NaOH in Milli-Q water and adding Na-ClO solution.Solutions were mixed and allowed to react for at least 6 but not more than 24 h.Solutions turned various shades of blue, which to relate to NH + 4 concentration, and were measured by absorbance at 630 nm on a Hitachi U-1100 spectrophotometer.Five-point calibration curves were measured before each sample set and corrected using the VKI standard (QC RW1; www.eurofins.dk;Table 1).
The fifth aliquot was used to measure dissolved phosphate (PO 3− 4 ) following the method given by Gieskes et al. (1991).Remaining pore water (generally between 1 and 3 mL) was added to Milli-Q water to a sum of 10 mL.Two reagents were added to the solution to react with PO 3− 4 .Reagent A was prepared by making three solutions: (NH 4 ) 2 MoO 4 , H 2 SO 4 , and C 8 H 4 K 2 O 12 Sb 2 • XH 2 O were added to Milli-Q water, and the solutions were added dropwise.Reagent B was created with C 6 H 8 O 6 .After samples were prepared, reagents A and B were added, mixed, and allowed to react for 30 min.Solutions turned various shades of blue, relating to PO 3− 4 concentration, and were measured at an absorbance of 880 nm.Fivepoint calibration curves were measured before each sample set and corrected using the VKI standard.
For 352 pore water samples, a sixth aliquot of approximately 2 mL was mixed with 200 µL of a 2.5 % Zn-acetate (Zn(C 2 H 3 O 2 ) 2 ) solution.Given the extremely low solubility of ZnS, a white precipitate forms when such a solution is added to pore water samples with even trace H 2 S concentrations (Cline, 1969;Goldhaber, 1974).

General observations
With the large number of pore water measurements (Table S1 in the Supplement), we begin with some generalities regarding results.We plot pore water concentration profiles along each transect collectively , irrespective of coring device or water depth, although clear variance in pore water chemistry exists between stations for some dissolved species (e.g., Fe).Most species display "smooth" concentration profiles with respect to sediment depth .That is, concentrations of successive samples do not display a high degree of scatter.This is expected for pore water profiles in sediment where diffusion dominates (Froelich et al., 1979;Klump and Martens, 1981;Schulz, 2000;Torres and Kastner, 2009;Hu et al., 2015).However, for some dissolved species whose concentrations do not appreciably change over depth (e.g., Ba 2+ and Ca 2+ ), scatter exists beyond that predicted from analytical precision.We discuss this in the Supplement.

Alkalinity and δ 13 C of DIC
Alkalinity concentrations increase with depth in all cores (Figs.4-8).Moreover, in most cases, the rise is roughly linear.Across all stations on the four transects, alkalinity increases by an average of 0.51 mM m −1 , although variance exists between mean gradients for each transect (Tr1 = 0.46 mM m −1 , Tr2 = 0.34 mM m −1 , Tr3 = 0.91 mM m −1 , and Tr4 = 0.44 mM m −1 ) and by station along each transect.The Lomonosov Ridge station differs (Fig. 8), as alkalinity increases much more with depth (1.86 mM m −1 ).

Sulfur and sulfate
No sulfide was observed by smell, and no ZnS precipitated in any pore water sample upon addition of Znacetate solution.Molar concentrations of total dissolved sulfur should, therefore, represent those of dissolved SO 2− 4 .Along the four transects, dissolved sulfur concentrations decrease with depth at all stations (Figs.4-7).The total dissolved sulfur concentrations in the shallowest samples varied from 27.3 to 30.6 mM and averaged 28.7 mM.From these "seafloor" values, concentrations decrease by an average 0.69 mM m −1 , again with variance according to stations and transect (Tr1 = −0.58mM m −1 , Tr2 = −0.57mM m −1 , Tr3 = −1.09mM m −1 , and Tr4 = −0.60 mM m −1 ).The dissolved sulfur gradients across all stations within SNESS range from −0.41 to −1.13 mM m −1 .Dissolved sulfur at the Lomonosov Ridge station displays a significantly steeper decrease than any other station (−1.92 mM m −1 ).Importantly, decreases in dissolved sulfur are similar in magnitude to increases in alkalinity at each station examined.Indeed, the molar ratio of alkalinity change to sulfur change (-Alkalinity/ S) is 0.98 (Fig. 9a).

Ammonia and phosphate
The C : N : P molar ratio of typical marine organic matter is approximately 106 : 16 : 1 (Redfield, 1958;Takahashi, 1985).Although this ratio differs for terrestrial organic carbon (closer to 134 : 9 : 1, Tian et al., 2010), dissolved NH + 4 and HPO 2− 4 concentrations in pore water can be used in a general sense to assess consumption of particulate organic carbon.This is because the degradation releases these species to pore water (Froelich et al., 1979).Notably, concentrations of NH + 4 and HPO 2− 4 are near or below detection in samples immediately below the seafloor (Figs.4-8).
Dissolved NH + 4 profiles increase almost linearly with depth, although with slight concave-down curvature.Similar to alkalinity profiles, NH + 4 concentrations rise with depth below the seafloor and more at stations with shallower water depth (although we note an exception for Tr2).Across stations along the four transects, pore water NH + 4 concentrations increase with depth on average by 38.69 µM m −1 , with a range from 11.3 to By contrast, concentrations of dissolved HPO 2− 4 in our cores typically increase to a subsurface maximum and then decrease (Figs.4-8).Based on the available data, a more pronounced maximum appears to occur at stations with relatively shallow water depth.For example, consider the peak in HPO 2− 4 concentrations at four stations.At the two shallow stations, S12 (384 m) and S22 (367 m), the HPO 2− 4 maxima are 73 µM (1.91 m) and 18 µM (0.66 m), respectively, but at the two deeper stations, S17 (977 m) and S14 (733 m), the HPO 2− 4 maxima are only 6.7 µM (1.76 m) and 7.1 µM (2.33 m), respectively.The station on the Lomonosov Ridge (S31) has a high in HPO 2− 4 concentration of 76 µM at 1.02 m below the seafloor.In general, stations with more pronounced HPO 2− 4 maxima also have greater increases in alkalinity with depth.
The NH + 4 , HPO 2− 4 , and alkalinity profiles relate to one another statistically, although with some distinctions.All stations have a C : N ratio in pore waters much higher than the canonical Redfield ratio of 6.625 (Fig. 10).Alternatively, the concentration relationship of alkalinity and ammonium ion can be expressed by a second order polynomial ([NH  average molar ratio (Alk / NH + 4 ) of 14.7, which is close to that expected for degradation of terrestrial organic carbon.Interestingly, this ratio deviates somewhat across transects, increasing at sites from Tr1, Tr3, and Tr2, to the Lomonosov Ridge station.Across all stations and above the subseafloor HPO 2− 4 peak, the molar ratio of alkalinity to phosphate ion (Alk / HPO 2− 4 ) averages 55.7 in pore water samples.This ratio also generally increases in cores from east to west.

Metals
At most stations, dissolved Ba 2+ concentrations increase nonlinearly from values at or below detection limit (0.01 µM) near the seafloor to generally constant values (0.6-0.7 µM) within 0.8 m below the seafloor.However, at several stations, dissolved Ba 2+ concentrations remain at or below the detection limit for all samples.
Overall, dissolved Ca 2+ , Mg 2+ , and Sr 2+ concentrations decrease with depth (Figs.4-8).For the stations along the four transects, Ca 2+ concentrations drop on average between -0.09 and −0.12 mM m −1 (Tr1), about −0.09 mM m −1 (Tr2), between −0.09 and −0.The profiles of dissolved Mn and Fe are complicated in terms of location.Generally, profiles show a broad rise in concentrations within the upper sediment and a subsequent drop in concentrations at deeper depth.Some stations have a maxima in dissolved Mn (Stations S12 (135 µM at 5 m), S28 (66 µM at 3.1 m), and the Lomonosov Ridge (86 µM at 1.3 m), where concentrations decrease below.At other stations, however, Mn concentrations still appear to be increasing at the lowest depth.Iron concentrations are generally below the detection limit at or near the seafloor, and begin increasing around 2.5-3.5 m, reaching concentrations upward of 20 µM.

Fidelity of Rhizon pore water measurements
Researchers have employed multiple methods to extract pore waters from marine sediments over the last few decades, but the Rhizon technique remains relatively novel (e.g., Seeberg-Elverfeldt et al., 2005;Dickens et al., 2007;Pohlman et al., 2008).Several studies have questioned the accuracy and precision of analyses obtained through this approach (e.g., Schrum et al., 2012;Miller et al., 2014).Two experiments conducted during the SWERUS-C3 Leg 2 expedition using the Rhizons suggest that part of the problem concerns the timing and location of sampling (Supplement).Notably, however, as clearly documented in previous works (Seeberg-Elverfeldt et al., 2005;Dickens et al., 2007;Pohlman et al., 2008), Rhizon sampling can lead to smooth concentration profiles for multiple dissolved species, including alkalinity (Figs.4-8).
Concerns about Rhizon sampling may be valid for dissolved components when concentration gradients are very low.For example, Schrum et al. (2012) stressed alkalinity differences between samples collected at similar depth using Rhizon sampling and conventional squeezing.However, the total alkalinity range in this study was between 1.6 and 2.6 mM, and typical differences were 0.06 mM.A similar finding occurs in the dissolved Ca 2+ and Ba 2+ profiles of this study, where the range in values is small and adjacent samples deviate by more than analytical precision (Table 1, Fig. S3).However, when the signal-to-noise ratio becomes high, as is true with most dissolved components at most stations , the Rhizon sampling renders pore water profiles with well-defined concentration gradients that can be interpreted in terms of chemical reactions and fluxes.

General absence of methane
Direct measurements of dissolved CH 4 in deep-sea sediment are complicated (Claypool and Kvenvolden, 1983).During core retrieval and depressurization, major CH 4 loss can occur from pore space (Dickens et al., 1997).Importantly, in sediments recovered through piston coring and where in situ CH 4 concentrations significantly exceed solubility conditions at 1 atm pressure, gas release typically generates sub-horizontal cracks ("gas voids") that span the core between the liner.No such cracks were documented in any of the cores.
Excluding station 31 on the Lomonosov Ridge, there is no indication of a shallow SMT in any of the cores.Interstitial water sulfur concentrations do not drop below 22.8 mM within the upper 8 m.In fact, calculated downward SO 2− 4 fluxes, as inferred from sulfur concentration gradients (Table 2), range from −1.8 to −6.2 mol m −2 kyr −1 for all stations except Station S31.For comparison, for a site with a near-seafloor temperature of 2 • C (Fig. S2) and porosities similar to those measured (Fig. S1), an SMT at 6.0 mbsf would imply an SO 2− 4 flux of −40 mol m −2 kyr −1 .Given the lack of HS − and the measured pH (Fig. S2  (Eq.4).Estimated HCO − 3 fluxes (J HCO − 3 ) do not exceed 6.8 mol m −2 kyr −1 at any station east of the Lomonosov Ridge (Table 2).For comparison, at sites with abundant CH 4 at depth, J HCO − 3 generally exceeds 30 mol m −2 kyr −1 above the SMT (Table 2).These extreme fluxes arise because methanogenesis in deeper sediment drives an upward flux of HCO − 3 (Fig. 3) and because AOM contributes additional HCO − 3 and HS − to pore water at the SMT (Eq.1).The δ 13 C-DIC values of pore water decrease with depth at all stations, almost in concert with the rise in alkalinity, implying no CH 4 production because methanogenesis would increase δ 13 C-DIC values (Fig. 9c; Whiticar, 1999).Other than Station S31, the lowest value of δ 13 C-DIC is −25.23 ‰ at 5.5 m at Station S22 (Fig. 6).This is interesting because a series of microbial reactions utilizing POM can lead to higher alkalinity and lower δ 13 C-DIC values in pore water.The most important of these reactions is organoclastic sulfate reduction (OSR), which can be expressed as (Berner, 1980;Boudreau and Westrich, 1984 As emphasized previously, methane-charged sediment sequences do occur on continental slopes in the Arctic.Of particular interest to this study are locations in the Beaufort Sea, where indications for gas hydrate manifest on seismic profiles (Grantz et al., 1976(Grantz et al., , 1982;;Weaver and Stewart, 1982;Hart et al., 2011;Phrampus et al., 2014), and pore water profiles have been generated using shallow piston cores (Coffin et al., 2013).Striking contrasts exist between pore water profiles of the Beaufort Sea and those of SNESS (Table 2).In the Beaufort Sea, there are moderate to high downward SO 2− 4 and upward CH 4 fluxes (1.9 to 154.8 mol m −2 kyr −1 ), shallow SMTs (6.29 to 1.06 mbsf), high DIC fluxes between the SMT and the mud line (46.3 to 242.6), and negative δ13C-DIC values at SMTs (≈ −20 ‰).

Special case: Lomonosov Ridge station
Station 31 on the Lomonosov Ridge (Fig. 8) differs from all other stations examined in this study.Here, pore water chemistry profiles hint at CH 4 in pore space within shallow sediment.Extrapolation of the dissolved sulfur profile suggests an SMT at approximately 14 mbsf This depth lies within the range common for locations with AOM (D'Hondt et al., 2002), notably including well studied sites on Blake Ridge (Borowski et al., 1999).Similar to some sites with CH 4 , the δ 13 C-DIC values become very "light"; indeed, the value at the base of the core, −43.5 ‰, almost necessitates CH4 oxidation within shallow sediment.Comparably steep alkalinity (1.6 mM m −1 ) and NH + 4 gradients (60.4 µM m −1 ) also characterize most sites with CH 4 near the seafloor.However, there is an issue concerning reduced sulfur, which is a product of AOM (Eq.1).If AOM was occurring at ∼ 13.9 mbsf, one might expect evidence for HS − migrating from below (Fig. 3).No ZnS precipitated in pore waters of this core upon addition of ZnAc.

Other chemistry
Microbial communities preferentially utilize the most energetically favorable oxidant available, leading to a characteristic reaction sequence in marine sediment (Froelich et al., 1979;Berner, 1980;D'Hondt et al., 2004).With increasing depth below the seafloor, these include aerobic respiration, denitrification, manganese oxide reduction, iron oxide reduc-tion, SO 2− 4 reduction, and finally methanogenesis.Many of the cores collected across SNESS appear to terminate in the zone of metal oxide reduction.This is because, at most stations, Mn and Fe profiles are still increasing at the bottom of the sampled interval , presumably due to dissimilatory Mn-and Fe-oxide reduction.However, Mn may be more complicated.März et al. (2011) find evidence from Mn profiles along the southern Mendeleev Ridge that suggest diagenetic remobilization of Mn at depth and diffusion toward shallow sediments.The relatively deep depths of metal oxide reduction, nevertheless, are consistent with a relatively low input of POM to the seafloor and completely contrast with most sites where high CH 4 concentrations exist in shallow sediment.A simple interpretation is that there is insufficient input of POC over time to drive methanogenesis near the seafloor.
The station on the Lomonosov Ridge again stands apart.Here, Mn and Fe concentrations reach maxima at 1.3 and 0.5 mbsf, respectively, and decrease below.This is likely due to Mn and Fe produced during dissimilatory oxide reduction, but where both metals precipitate below into carbonate (Mn and Fe) or sulfide phases (Fe; Jørgensen et al., 1990;März et al., 2011).This is common at locations with modest POC input, and the Lomonosov Ridge site appears to receive higher organic carbon burial over time than all the www.biogeosciences.net/14/2929/2017/Biogeosciences, 14, 2929-2953, 2017 East Siberian slope 1120 14 −13.9 11.3 -Atlantic New Jersey continental slope q,i 912 28.9 −3.3 3.6 * -Atlantic Blake Ridge q,p 1293 50.3 −3.4 3.8 * -Atlantic Blake Ridge q,p 1798 26.9 −6.6 4.9 * -Atlantic Blake Ridge q,x 2567 42.0 −3.8 3.5 * -Atlantic Blake Ridge q,x 2641 24.5 −7.6 6.9 * -Atlantic Blake Ridge  other locations examined.Given the relationship of alkalinity to ammonia (Fig. 10), much of the organic matter on the continental slope may derive from terrestrial rather than marine sources (Müller and Suess, 1979;Reimers et al., 1992), but a more detailed study of sedimentation rates and organic matter content and composition is required to elucidate these relationships further.

Signatures of AOM and OSR
Some authors have used changes in DIC and SO 2− 4 concentrations between the seafloor and the SMT to infer the relative importance of AOM and OSR in marine sediments (Kastner et al., 2008b;Luo et al., 2013;Hu et al., 2015).This idea can be expressed by comparing (DIC+Ca 2+   (1230, 1426, and 1427;1230, Shipboard Scientific Party, 2003;1426and 1427, Expedition Scientists, 2014) given for comparison.
for loss of DIC via carbonate precipitation (other authors, such as Snyder et al., 2007, andWehrmann et al., 2011, use fluxes instead of concentrations).The rationale lies in the fact that the C : S ratio for AOM is 1 : 1 (Eq.1), whereas the C : S ratio for OSR is 2 : 1 (Eq.7).However, this approach neglects two considerations: (1) changes in concentration do not directly relate to fluxes, because of differences in diffusivities of various ionic species, and (2) a flux of HCO − 3 from below the SMT can augment the DIC produced from AOM or OSR at or above the SMT (Dickens and Snyder, 2009).Thus, changes in alkalinity relative to SO 2− 4 often exceed 1 : 1, even at locations completely dominated by AOM (Chatterjee et al., 2011).
Rather than comparing changes in C : S molar ratios or going through detailed flux calculations to interrogate the importance of the two reactions in shallow sediment, one might also incorporate the δ 13 C-DIC values.This is because δ 13 C-DIC values and the depth of DIC production differ considerably across many sites where either AOM or OSR dominates.We generate a figure expressing these relationships at multiple sites (Fig. 12), where the y axis is and the x axis is DIC • δ 13 C-DIC.The C : S ratios of dissolved species lie above 1 : 1 at most locations, regardless of whether CH 4 exists in shallow sediment and AOM dominates, as highlighted by Chatterjee et al. (2011).However, sites with significant CH 4 have considerably more negative DIC • δ 13 C-DIC values.Notably, pore waters from all stations examined here, except S31 on the Lomonosov Ridge, have modest DIC • δ 13 C-DIC values consistent with a dominance of OSR in shallow sediment rather than AOM.In summary, from general pore water considerations as well as from comparisons to pore water profiles at other locations, sediments across SNESS do not contain CH 4 over extensive areas of shallow sediment.Implicit in this finding is that sediment sequences in this region lack widespread gas hydrate.As models for gas hydrate occurrence in the Arctic (Fig. 1) correctly predict gas hydrate in several regions (e.g., Kvenvolden and Grantz, 1990;Max and Lowrie, 1993;Max and Johnson, 2012), our findings prompt an interesting question: why are predictions so markedly wrong for the SNESS?

Possible explanations for widespread absence of gas hydrate and methane
To understand the likely absence of widespread gas hydrates across SNESS, one needs to consider the generalities of their occurrence in marine sediment.There are two basic conditions for gas hydrate on continental slopes (Kvenvolden, 1993;Dickens, 2001).The first consideration is the "potential volume", or the pore space where physiochemical conditions (e.g., temperature, pressure, salinity, sediment poros-  , 994, 995, 997, 1059, 994, 995, 997, , 1225, 994, 995, 997, , 1230, 994, 995, 997, , 1426, 994, 995, 997, , 1427, 994, 995, 997, , and 1319, 994, 995, 997, (994-997, 1059, 994, 995, 997, , Borowski et al., 2000;;1225and 1230, Shipboard Scientific Party, 2003;1426and 1427, Expedition Scientists, 2014), are given for comparison.Blue marginal distribution curves show global distribution, while red gives SNESS stations (this project).SNESS pore waters have higher C : N and lower C : P than comparative sites.3 ) and sulfate (SO 2− 4 ) flux exponential relationship with SMT depth for all sites listed in Table 2. ity) are amenable to gas hydrate formation.As stressed in previous works, because of cold bottom water and a low geothermal gradient, the region has a relatively large volume of sediment with appropriate gas hydrate stability conditions (Stranne et al., 2016).The second consideration is the "occupancy", or the fraction of sediment pore space with sufficient CH 4 to precipitate gas hydrate.While environmental conditions across SNESS are highly conducive for gas hydrate formation, pore water profiles strongly indicate little to no CH 4 exists in the upper hundred meters of sediment.
This inference strongly depends on recognition as to how diffusive systems operate in marine sediment.Hundreds of pore water profiles have been generated during scientific ocean drilling expeditions, including scores into CH 4charged sediment sequences.These profiles almost universally show vertical connectivity of pore water chemistry over hundreds of meters (Fig. 3).Moreover, away from local sites of advection, pore water profiles are generally similar over extensive areas.This occurs because, given sufficient permeability and time, diffusive fluxes transport species from intervals of high concentration to intervals of low concentration.Hence, unless some impermeable layer exists in the sediment sequence, even CH 4 at depth impacts near-seafloor concen- +Mg 2+ ) and sulfate change ( SO 2− 4 ) to the product of DIC and δ 13 C-DIC value (AT13-2 and KC151, Kastner et al., 2008a;PC02-PC14, Coffin et al., 2008;994-997, 1059994-997, , Borowski et al., 2000;;Paull et al., 2000;1326and 1329, Torres and Kastner, 2009;GC233 and GB425, Hu et al., 2010;D-5-D-8 and D-F, Hu et al., 2015;C9-C19, Luo et al., 2013;PC-07, Smith and Coffin, 2014;1230, Shipboard Scientific Party, 2003;1244and 1247, Claypool et al., 2006;;1305and 1306, Party, 2005, including , including  trations.Indeed, work on the outer Blake Ridge wonderfully shows this phenomenon.The uppermost gas hydrate in sediment in this region lies at about 190 mbsf (Borowski et al., 1999).Nonetheless, its presence occurs over ∼ 26 000 km 2 and affects shallow pore water profiles across this region, because the flux of CH 4 from depth drives AOM near the seafloor (Borowski et al., 1999;Dickens, 2001).
No seafloor features indicative of seafloor CH 4 expulsion were found during the bathymetric mapping of SNESS.Nonetheless, it is possible that local CH 4 venting, perhaps related to and mediated by bubble transport, could occur away from transects and cores of SWERUS Leg 2. Certainly, the chemistry of advecting fluids toward seafloor features such as mud volcanoes and cold seeps typically differs from the much broader surrounding region (Luff and Wallmann, 2003;Coffin et al., 2007;Hiruta et al., 2009;Hu et al., 2010;Coffin et al., 2014;Hu et al., 2015).However, in such cases, even the encompassing area typically has shallow SMTs.Without invoking odd geology, such as an extensive impermeable layer, it is unlikely that significant CH 4 exists in shallow sediment across much of SNESS, including as gas hydrate or free gas.Here it is stressed that neither gas hydrate nor free gas can exist in sediment on continental slopes without high concentrations of dissolved gas in surrounding pore water (Dickens et al., 1997;Hiruta et al., 2009;Geprägs et al., 2016).The surprising lack of CH 4 across SNESS, as inferred from pore water profiles, suggests insufficient net input of POC over time, so that either methanogenesis has not occurred or the product has been lost.
The accumulation of POC within the SNESS region may be relatively low over the Plio-Pleistocene.With low POC inputs, other microbial reactions can exhaust the labile organic matter needed for methanogenesis.This may, in fact, explain why the pore water chemistry suggests that metal oxide reduction dominates the geochemical environment at most of our stations.Without further investigation, we offer four possibilities (not mutually exclusive) as to why this might occur: (1) significant sea-ice concentrations, both at present-day and during past glacial intervals, greatly diminishes primary production of marine organic carbon within the water column; (2) the extremely broad continental shelf prevents large accumulations of terrestrial organic-rich sediment from reaching the slope; (3) highly variable sediment accumulation, perhaps corresponding to glacial-interglacial os-cillations, creates a situation where POC from either source is consumed during time intervals of low deposition; and, although not directly related to POC accumulation, (4) changes in sea level during the last glacial maximum caused much of the hydrate to outgas as the stability zone moved downslope (Stranne et al., 2016).With the third explanation, large landbased glaciers in the past may have physically scoured sediment (and organic matter) from the upper slope (Jakobsson et al., 2014).Importantly, the first three explanations distinguish the SNESS region from the Beaufort Sea, where abundant CH 4 in shallow sediment unquestionably occurs (Coffin et al., 2011;Treude et al., 2014).
In earlier times, particularly the Cretaceous through early Eocene (Jenkyns et al., 2004;Sluijs et al., 2006;Backman et al., 2009), organic-rich sediment may have accumulated at high rates throughout the Arctic.In the Lomonosov Ridge in the central Arctic, lower Eocene sediments definitely contain high organic carbon and potential indicators of past methanogenesis (e.g., barium mobilization).As these cores contain no CH 4 at present day, if CH 4 was generated, it has presumably been lost in the intervening time.Should these organic-rich horizons be buried across the SNESS region and presently generating CH 4 via thermogenesis, the gas is too deeply buried to affect shallow sediment.

Conclusions
Leg 2 of the SWERUS-C3 expedition recovered sediments and pore waters from numerous stations across the continental slopes north of the East Siberian Sea.These stations extend from Wrangel Island to the New Siberian Islands and provide information from a climatically sensitive but highly inaccessible area.
In an effort to understand CH 4 cycling within the SNESS region, we generated detailed pore water profiles of multiple dissolved constituents.The pore water profiles are coherent and interpretable and give a general view: most stations have low SO 2− 4 and HCO − 3 fluxes (< 6.2 and 6.8 mol m −2 kyr −1 , respectively), a moderate decrease in δ 13 C-DIC values with depth (−3.6 ‰ m −1 average), no dissolved H 2 S, a moderate rise in HPO 2− 4 and NH + 4 concentrations, and slightly decreasing Ca 2+ , Mg 2+ , and Sr 2+ concentrations.Except for one station on the Lomonosov Ridge, metal oxide reduction appears to be the dominant geochemical environment affecting shallow sediment, and there is no evidence for upward diffusing CH 4 .These results strongly suggest that gas hydrates do not occur on any of our depth transects spread across the continental slope in this region of the Arctic Ocean.This directly conflicts with ideas in multiple publications, which generally have assumed large quantities of CH 4 and gas hydrate.However, it remains possible that significant CH 4 occurs where the Lomonosov Ridge intersects the continental margin as well as westward on the Laptev Sea continental slope.

Figure 2 .
Figure 2. Bathymetric map of Eurasian Arctic showing the overall cruise track of Leg 2 along with the four transects and coring locations.Multicores shown as yellow triangles, gravity and piston cores as white stars, and the ship track line as the gray line from Barrow, Alaska.

Figure 3 .
Figure 3. Idealized pore water concentration profiles for high and low upward methane flux.Discrete data points for sites 722 (Arabian Sea; Seifert and Michaelis, 1991; D'Hondt et al., 2002) and 1230 (offshore Peru; Donohue et al., 2006) are given as reference.

Figure 8 .
Figure 8. Lomonosov Ridge Station results.IAPSO standard seawater (black dotted line) and representative stations from the four transects shown for comparison.
calculated an SO 2− 4 -flux of −14.7 mol m −2 kyr −1 for a site with an SMT at 14 mbsf in the Argentine Basin, and Berg (2008) calculated an SO 2− 4 flux of −8.05 mol m −2 kyr −1 for a site with an SMT at 16 mbsf along the Costa Rica margin.

Table 2 .
Published and calculated fluxes.