Characterization of particulate organic matter in the Lena River delta and adjacent nearshore zone , NE Siberia – Part I : Radiocarbon inventories

Particulate organic matter (POM) derived from permafrost soils and transported by the Lena River represents a quantitatively important terrestrial carbon pool exported to Laptev Sea sediments (next to POM derived from coastal erosion). Its fate in a future warming Arctic, i.e., its remobilization and remineralization after permafrost thawing as well as its transport pathways to and sequestration in marine sediments, is currently under debate. We present one of the first radiocarbon (C) data sets for surface water POM within the Lena Delta sampled in the summers of 2009–2010 and spring 2011 (n= 30 samples). The bulk 1C values varied from −55 to −391 ‰ translating into C ages of 395 to 3920 years BP. We further estimated the fraction of soil-derived POM to our samples based on (1) particulate organic carbon to particulate nitrogen ratios (POC : PN) and (2) on the stable carbon isotope (δC) composition of our samples. Assuming that this phytoplankton POM has a modern C concentration, we inferred the C concentrations of the soil-derived POM fractions. The results ranged from −322 to −884 ‰ (i.e., 3060 to 17 250 C years BP) for the POC : PN-based scenario and from −261 to −944 ‰ (i.e., 2370 to 23 100 C years BP) for the δC-based scenario. Despite the limitations of our approach, the estimated 1C values of the soil-derived POM fractions seem to reflect the heterogeneous C concentrations of the Lena River catchment soils covering a range from Holocene to Pleistocene ages better than the bulk POM 1C values. We further used a dual-carbonisotope three-end-member mixing model to distinguish between POM contributions from Holocene soils and Pleistocene Ice Complex (IC) deposits to our soil-derived POM fraction. IC contributions are comparatively low (mean of 0.14) compared to Holocene soils (mean of 0.32) and riverine phytoplankton (mean of 0.55), which could be explained with the restricted spatial distribution of IC deposits within the Lena catchment. Based on our newly calculated soilderived POM 1C values, we propose an isotopic range for the riverine soil-derived POM end member with 1C of −495± 153 ‰ deduced from our δC-based binary mixing model and δC of −26.6± 1 ‰ deduced from our data of Lena Delta soils and literature values. These estimates can help to improve the dual-carbon-isotope simulations used to quantify contributions from riverine soil POM, Pleistocene IC POM from coastal erosion, and marine POM in Siberian shelf sediments.

−55 to −391 ‰ translating into 14 C ages of 395 to 3920 years BP.We further estimated the fraction of soil-derived POM to our samples based on (1) particulate organic carbon to particulate nitrogen ratios (POC : PN) and (2) on the stable carbon isotope (δ 13 C) composition of our samples.Assuming that this phytoplankton POM has a modern 14 C concentration, we inferred the 14 C concentrations of the soil-derived POM fractions.The results ranged from −322 to −884 ‰ (i.e., 3060 to 17 250 14 C years BP) for the POC : PN-based scenario and from −261 to −944 ‰ (i.e., 2370 to 23 100 14 C years BP) for the δ 13 C-based scenario.Despite the limitations of our approach, the estimated 14 C values of the soil-derived POM fractions seem to reflect the heterogeneous 14 C concentrations of the Lena River catchment soils covering a range from Holocene to Pleistocene ages better than the bulk POM 14 C values.We further used a dual-carbonisotope three-end-member mixing model to distinguish be-tween POM contributions from Holocene soils and Pleistocene Ice Complex (IC) deposits to our soil-derived POM fraction.IC contributions are comparatively low (mean of 0.14) compared to Holocene soils (mean of 0.32) and riverine phytoplankton (mean of 0.55), which could be explained with the restricted spatial distribution of IC deposits within the Lena catchment.Based on our newly calculated soilderived POM 14 C values, we propose an isotopic range for the riverine soil-derived POM end member with 14 C of −495 ± 153 ‰ deduced from our δ 13 C-based binary mixing model and δ 13 C of −26.6 ± 1 ‰ deduced from our data of Lena Delta soils and literature values.These estimates can help to improve the dual-carbon-isotope simulations used to quantify contributions from riverine soil POM, Pleistocene IC POM from coastal erosion, and marine POM in Siberian shelf sediments.

Introduction
Huge amounts of soil organic carbon are currently stored frozen in permafrost soils of the high northern latitudes (e.g., Tarnocai et al., 2009;Zimov et al., 2009) and excluded from biogeochemical cycling.Due to recent observed and projected amplified warming of the Arctic (ACIA, 2005;Serreze et al., 2000), carbon cycling and the fate of organic carbon released from permafrost soils have received growing attention Published by Copernicus Publications on behalf of the European Geosciences Union.
Increasing permafrost temperatures, increasing thaw depth in summer (active layer depth), increasing river runoff (Boike et al., 2013;McClelland et al., 2012;Peterson et al., 2002), and increasing length of the open water season (Markus et al., 2009) affecting coastal erosion of permafrost deposits will likely lead to enhanced mobilization and export of old, previously frozen organic matter (OM) to the Arctic shelf seas.The understanding of the different terrestrial OM sources (e.g., fresh vegetation, surface soils, Pleistocene Ice Complex (IC)) and their age and quality has significantly improved over the last decade (e.g., Guo et al., 2007;Vonk et al., 2010).The use of carbon isotopes (δ 13 C, 14 C) of dissolved and particulate organic matter as well as individual biomarkers has helped characterizing and distinguishing these different carbon pools, e.g., old, yet little degraded Pleistocene IC-derived OM and comparatively younger and more degraded fluvial OM reaching the Siberian shelf seas (Feng et al., 2013;Guo et al., 2004;Gustafsson et al., 2011;Karlsson et al., 2011;Vonk et al., 2012Vonk et al., , 2010)).
However, particularly in Siberia, 14 C data on riverine suspended particulate organic matter (POM) are only sparsely available.A recently published data set from the Arctic Great Rivers Observatory (A-GRO, www.arcticgreatrivers.org, published 10 January 2015) provides POM 14 C data for the Lena River from Zhigansk, approximately 900 km upstream of the delta.To our knowledge the only additional POM 14 C concentrations sampled directly in Siberian rivers are from the Lena River (unpublished in Vonk et al., 2010) and the Kolyma River (Supplement of Vonk et al., 2012).Available POM 14 C concentrations for Siberia are from offshore of the deltas of the Lena (Karlsson et al., 2011) and Kolyma rivers (Vonk et al., 2010) and inferred from a sediment core taken in a floodplain lake of the Ob' River (Dickens et al., 2011).Due to settling of POM and marine primary production fueled by the riverine nutrients, the 14 C concentrations of samples taken off the river mouths are already altered from the "original" river signal.In contrast to Siberia, there are detailed studies on riverine POM 14 C concentrations in the Yukon and Mackenzie rivers in the North American Arctic (Goñi et al., 2005;Guo et al., 2007;Guo and Macdonald, 2006).Here, the POM was found to be significantly older than the dissolved organic matter (DOM) of these rivers and interpreted to be derived from riverbank erosion and thawing permafrost soils compared to a modern vegetation source of DOM (Guo et al., 2006(Guo et al., , 2007)).Against the backdrop of a warming Arctic (IPCC, 2013;ACIA, 2005) and the projected release of old OM in the future presumably as POM rather than DOM (Guo et al., 2007), it is important to assess the age heterogeneity carried by riverine POM and sequestered in the nearshore zone today.This will be an important benchmark to distinguish catchment-related changes caused by increasing temperatures from the natural variability of the river system.
In contrast to sedimentary OM, POM provides limited spatial and temporal snapshots of its OM properties.However, it has been shown that riverine POM in Arctic rivers carries an integrated signal from their permafrost watersheds representing the watershed environmental characteristics (e.g., Goñi et al., 2000;Lobbes et al., 2000;Vonk and Gustafsson, 2009;Winterfeld et al., 2015).Despite the fact that Lena River POM flux is an order of magnitude smaller than its DOM flux, it is more likely to transports the climate change signal from permafrost soils of the river catchment (Guo et al., 2007).Also, it has been proposed that the POM pool could be as important as the DOM pool for Arctic carbon cycling, because of its possibly high degradation rates in the water column compared to DOM (Sanchéz-García et al., 2011;van Dongen et al., 2008).
Here we present the first part of a study on particulate OM in the Lena Delta, Siberia (Winterfeld et al., 2015).Our POM samples taken in three consecutive years (2009−2011) in the spring and summer seasons add up to the existing data on elemental composition, stable carbon isotopes (δ 13 C), and radiocarbon ( 14 C) values as well as provide a first POM 14 C data set covering three consecutive years in the Lena Delta.Because riverine and marine POM concentrations are usually too low for source specific biomarker 14 C analysis, the available 14 C data are from bulk OM.This could result in a considerable age bias depending on the contribution of phytoplankton OM with a rather enriched (modern) 14 C concentration to the individual samples.We used different approaches to estimate the fraction of soil-and plankton-derived OM in our POM samples and estimated the 14 C concentration for the soil-derived fraction of POM transported by the Lena River in summer and spring.Further, we used a dual-carbonisotope three-end-member mixing model according to Vonk et al. (2010Vonk et al. ( , 2012) ) to distinguish not only between planktonand soil-derived OM but also between OM from Holocene soils and Pleistocene IC deposits.The estimated 14 C concentrations of soil-derived POM will further help to define the typical isotopic signature of river POM more accurately for modeling riverine OM contributions to Laptev Sea shelf sediments.

Study area
A detailed description of the Lena River watershed and Lena Delta can, for example, be found in the companion paper of this study, which deals with the lignin composition of OM in the Lena Delta (Winterfeld et al., 2015).
In short, the Lena River is one of the largest Russian Arctic rivers, draining a watershed of ∼ 2.46 × 10 6 km 2 in central Siberia into the Laptev Sea.Continuous permafrost makes up about 72−80 % of the drainage area (Amon et al., 2012;Zhang et al., 2005), storing huge amounts of old OM.The    permafrost acts as a water impermeable layer and thus affects the regional hydrology and hydrochemistry.The Lena River water discharge and related dissolved and particulate load discharge are highest during spring ice breakup and snowmelt in late May to June, while summer and winter discharges are lower (Rachold et al., 2004).The mean annual water discharge is ∼ 588 km 3 for the years 1999 to 2008 (Holmes et al., 2012).Corresponding annual sediment, dissolved organic carbon (DOC), and particulate organic carbon (POC) fluxes are 20.7 (Holmes et al., 2002), 5.7 (Holmes et al., 2012), and 1.2 Tg yr −1 , respectively (Rachold and Hubberten, 1999).A second major source for terrigenous OM delivered to the Laptev Sea is the thermal erosion of iceand OM-rich Pleistocene IC deposits along the coast (see Gustafsson et al., 2011;Mueller-Lupp et al., 2000;Rachold and Hubberten, 1999;Rachold et al., 2004).Recently, it has been shown that the annual supply of total organic carbon from IC deposits along the Laptev Sea coast by erosion is ∼ 0.66 Tg yr −1 (Günther et al., 2013).
The Lena River Delta is the largest Arctic delta (∼ 32 000 km 2 ).It is characterized by a polygonal tundra landscape with active floodplains, and it can be divided into three geomorphological terraces (Grigoriev, 1993;Schwamborn et al., 2002).The first terrace includes active floodplains that were formed during the Holocene and makes up about 55 % of the total delta area (Morgenstern et al., 2008), covering the central and eastern part.Within the first delta terrace remains of a Pleistocene accumulation plain, also called Ice Complex or yedoma deposits, form the third terrace, covering about 6 % of the total delta area (Morgenstern et al., 2008).Sandy islands forming the second terrace cover the rest of the delta in the west.Water and sed-iment discharge are not equally distributed through the several delta channels.Approximately 80−90 % of the total water and up to 85 % of the sediment discharge are delivered through the three main eastern channels to the Buor Khaya Bay east of the delta (Fig. 1b), i.e., through the Sardakhsko-Trofimovskaya channel system (60−75 % water, 70 % sediment) and the Bykovskaya Channel (20−25 water, 15 % sediment).Only a minor portion is discharged to the north and west through the Tumatskaya and Olenyokskaya channels (5−10 water, 10 % sediment) (Charkin et al., 2011, and references therein;Ivanov and Piskun, 1999).

Sampling
The sampling sites presented in this study are located in the eastern part of the Lena Delta and adjacent Buor Khaya Bay (Table 1, Fig. 1b) in and along the channels of highest discharge.Permafrost soil samples from the first delta and surface water total suspended matter (TSM) were collected during three expeditions in August 2009, July/August 2010, and in late May and late June/early July 2011.All soil samples were taken from riverbank bluffs of the first delta terrace (Holocene formation), which is elevated (5−16 m) over the active floodplains.The soil profiles vary strongly in sediment composition and OM content.Within the profiles, sandy layers derived from extreme flooding events (Schwamborn et al., 2002) and aeolian input (Kutzbach et al., 2004;Sanders, 2011) alternate with buried surface soil layers and peat layers rich in fibrous plant and root detritus in different stages of decomposition.The peat layers are either of autochthonous or of allochthonous origin.Allochthonous material is eroded from riverbanks further upstream and redeposited in the delta.In order to obtain samples that reflect the original state of the frozen permafrost soils, we removed the thawed soil material from each riverbank bluff for its total height with a spade.Frozen pieces of soil were then excavated at different depths using a hatchet and a hammer.
Suspended particulate matter of Lena River surface water was sampled at different stations in the main river channels of the delta on the Russian vessel Puteyski 405 (Fig. 1b, Table 1) from the upper 0.5 m of the water column.In the smaller delta channels the samples were taken approximately in the middle of the channel.Because the main delta channels can be > 1 km in width, we sampled a few hundred meters off the shore with water depth of > 10 m, but not in the middle of the channel.Three samples (19,35,36;Fig. 1b) were taken outside the delta in shallow water depth.We discuss these samples as "delta samples", because the surface water at these sampling locations is outflowing Lena River water as shown by temperature and salinity profiles by Semiletov et al. (2011) and Kraberg et al. (2013).Between 1 and 30 L of water was filtered on pre-combusted (4.5 h at 450 • C) and pre-weighed glass fiber filters (GF / F Whatman, 0.7 µm membrane) for particulate organic carbon (POC) and nitrogen (PN) analysis as well as carbon isotope analysis.Additionally, one water sample of 20 L from the spring freshet in 2011 was stored cooled in opaque canisters for several days to allow for the suspended matter to settle.Before the supernatant water was decanted, it was filtered on pre-combusted and pre-weighed GF / F filters to check for the TSM remaining in suspension.For this sample (sample ID 37) the SPM of the supernatant water represented 0.1 % of the settled material on a dry weight basis, and therefore the loss of material in suspension can be neglected.
The soil samples were stored in pre-combusted glass jars (4.5 h at 450 • C) and GF / F filters were either stored in pre-combusted Petri dishes (∅ 47 mm) or wrapped in precombusted aluminum foil.All samples were kept frozen at −20 • C during storage and transport until analysis.

Laboratory analyses
Soil and sediment samples were freeze-dried, homogenized, and subsampled for elemental analysis.All filters were ovendried at 40 • C for 24 h.

Elemental analyses
Weight percent organic carbon (OC) and total nitrogen (TN) content of soil samples were determined by high-temperature combustion after removal of carbonates as described by Goñi et al. (2003).TSM samples were analyzed for OC and TN at the Alfred Wegener Institute in Bremerhaven, Germany, using the same protocol.Every 10 samples, control standards were analyzed to constrain the analytical uncertainty of 0.1 %.

Carbon isotope analysis ( 14 C, δ 13 C)
Samples were radiocarbon-dated at the National Ocean Sciences Accelerator Mass Spectrometry (NOSAMS) facility at Woods Hole Oceanographic Institution, USA.Bulk sediment samples and filters with TSM were submitted unprocessed, and inorganic carbon was removed during sample preparation at NOSAMS.The radiocarbon analyses at NOSAMS were carried out using standard methods (McNichol et al., 1994).Results are reported as 14 C, conventional radiocarbon ages (years BP), and fraction modern carbon (f MC) including the correction for isotope fractionation (Stuiver and Polach, 1977).
The stable carbon isotope composition (δ 13 C) was measured on splits of the CO 2 gas of the samples generated prior to graphite reduction at NOSAMS using a VG Optima IRMS.Results are reported in per mil (‰) relative to VPDB.

End-member modeling
The relative source contributions (phytoplankton and soil) to the POM samples are estimated by solving linear twoend-member (using POC : PN and δ 13 C as parameters) and dual-carbon-isotope (δ 13 C, 14 C) three-end-member mixing models with end members determined from own data and the literature.The end-member values used in the different modeling approaches are given in Table 4. Source data for the calculation of the POC : PN soil end member as well as the δ 13 C and 14 C end members for Holocene soil are given in Tables S1-S3 in the Supplement.
To account for the uncertainty of the measured values, we assume independent, normally distributed uncertainties for the observed values with a standard deviation characterizing the measurement uncertainty.For the end members we assume a normal distribution using the standard deviations from the observed samples on which the end members are based.
Uncertainty in the resulting contributions of phytoplankton and soil are estimated in a Monte Carlo approach similar to Vonk et al. (2010Vonk et al. ( , 2012) ) and Karlsson et al. (2011).The source fractions are estimated 100 000 times, each time drawing the end members as well as the observed values from their distributions.Solutions with negative fractions are omitted and the mean and the standard deviation are reported.

Elemental composition
The organic carbon (OC) and total nitrogen (TN) concentrations of the first terrace soil samples can be found in Tables 3  and S3 in the companion paper, Winterfeld et al. (2015).
Our POC data are well within the range of values reported for the Lena Delta before (Cauwet and Sidorov, 1996;Rachold and Hubberten, 1999;Semiletov et al., 2011).The two samples from the Buor Khaya Bay surface waters showed the lowest POC concentrations per liter, i.e., 0.37 mg L −1 (sample 35) and 0.15 mg L −1 (sample 36) with corresponding 4.4 and 1.5 wt % POC, respectively (Tables 2 and S4).The considerable drop of POC (and TSM) concentrations offshore of the delta is due to flocculation and settling of TSM in the zone of fresh and salt water mixing (e.g., Lisitsyn, 1995) and was also observed during other years of sampling (e.g., Cauwet and Sidorov, 1996;Semiletov et al., 2011).
The atomic particulate organic carbon (POC) to particulate nitrogen (PN) ratios (POC : PN) of samples taken in summer 2009 were slightly higher than for summer 2010 samples, with mean values of 9.6 (n = 20, 6.8−19.3)and 8.0 (n = 13, 5.0−10.3,Tables 2 and S4), respectively.The POC : PN ra-tios of samples taken in June/July 2011 were rather similar to the July/August samples with a mean of 7.8 (n = 9, 5.9−9.7).The sample from late May 2011 had a POC : PN ratio of 7.5.

Carbon isotope inventories
The stable carbon (δ 13 C) and radiocarbon ( 14 C) results are shown in Table 3.The radiocarbon data presented here will predominantly be discussed in terms of 14 C in per mil (‰).Additionally, the fraction modern carbon (fMC) and 14 C ages in years before present (years BP) are given in Table 3.

Stable carbon isotope composition (δ 13 C)
The two soil profiles from the first delta terrace showed δ 13 C values between −27.0 and −25.1 ‰ with a mean of −26.1 ‰, (n = 7, Table 3), which is within the range observed for Holocene soils in the delta (e.g., Schirrmeister et al., 2011, and references therein).Similar to the POC contents, the Lena Delta surface water POM δ 13 C values varied strongly spatially and annually (Fig. S1    May 2011 (Table 3).The δ 13 C of Buor Khaya Bay surface water POM in August 2010 could only be determined for one of the samples, i.e., sample 36, with −30.4 ‰ .

Radiocarbon ( 14 C) concentration
The 14 C values of the two soil profiles decreased with depth from −197 to −377 ‰ for the riverbank profile L09-12 and from −204 to −466 ‰ for profile L09-28 (Table 3, Fig. 3a).The corresponding 14 C ages for both profiles together ranged from 1710 to 4900 years BP.Within profile L09-12 on Samoylov Island an age reversal was observed for the two oldest samples (Table 3), which is most likely due to allochthonous material that was transported to the delta from an upstream location.The same age reversal on Samoylov Island was also observed by Kuptsov and Lisitsin (1996).Overall, our 14 C values reflect the late Holocene formation of these soils and fit within the range of ages determined for the Lena Delta first terrace (Bolshiyanov et al., 2015;Kuptsov and Lisitsin, 1996;Schwamborn et al., 2002).
As with other TSM parameters, the POM 14 C values showed strong spatial and interannual variability (Fig. 2q- 3, Fig. 2r-s).Overall these 14 C concentrations covered a range of 14 C ages from 395 to 3920 years BP.The samples with the lowest 14 C values of −262 (sample 1) and −391 ‰ (sample 31) were taken close to the Pleistocene IC deposits, which likely contributed to the local POM in the Lena River surface water.

Origin of organic matter in the Lena Delta
Riverine particulate organic matter consists of a heterogeneous mixture derived from two major sources, i.e., terrestrial (e.g., fresh vegetation and litter; surface and deep soil horizons) and aquatic (phytoplankton/bacterial primary production).The terrestrial OM in the Lena River catchment can further be differentiated into two pools of different age: the late Pleistocene organic-rich IC deposits, particularly in the lowlands (0-400 m elevation; e.g., Grosse et al., 2013) and the Holocene permafrost soils.POC : PN ratios as well as δ 13 C and 14 C values of bulk OM can be used to estimate terrestrial and aquatic contributions (e.g., Hedges and Oades, 1997).However, due to overlaps in soil/plant and algal/bacterial signatures, it might be difficult to unambiguously differentiate between terrestrial and aquatic sources.
Our POC : PN ratios from the summers 2009 and 2010 as well as spring 2011 vary largely throughout the delta (Tables 2 and S4) similar to the TSM and lignin phenol concentrations sampled during the same field trips (Winterfeld et al., 2015).The ratios range from 3.7 to 19.3 for all samples and seasons with mean atomic POC : PN ratios of 9.6 (2009), 7.6 (2010), and 7.8 (2011).Low POC : PN ratios, i.e., ∼ 6, generally indicate high contributions from phytoplankton and/or bacterial primary production, while ratios > 20 are indicative of soil and plant contributions (e.g., Hedges et al., 1997;Meyers and Lallier-Vergés, 1999).Based on these end members, our data from the summer and spring seasons would suggest a considerable fraction of primary production OM to be present in our samples.
However, due to possible sorption of inorganic nitrogen to clay minerals (Schubert and Calvert, 2001), our calculated POC : PN ratios might be too low, underestimating the soilderived OM fraction.This will be discussed in more detail in Sect.4.3.1.Another possibility for lower but mainly soilderived POC : PN ratios could be the selective degradation of labile OM compared to total nitrogen (Kuhry and Vitt, 1996).In general, the more enriched δ 13 C POM values (around −27 ‰) reflect the dominant contribution from C3 plants and soils (e.g., Hedges et al., 1997) from the river catchment.
The more depleted δ 13 C POM values (≤29 ‰) point to mixing with riverine plankton utilizing dissolved inorganic carbon (DIC) with depleted δ 13 C signatures as suggested for the Lena River (Alling et al., 2012;Rachold and Hubberten, 1999).The total flux of primary production OM in the Lena River is thought to be negligible due to low light penetration in the turbid waters (Cauwet and Sidorov, 1996;Sorokin and Sorokin, 1996).Yet, the surface water layer from which we took our TSM samples is characterized by an abundance of riverine plankton (Kraberg et al., 2013;Sorokin and Sorokin, 1996), which could explain the small soil-derived OM fraction.
In Sect.4.3 we present the results of two binary and one three-end-member mixing models to estimate the soil OM contribution to our POM samples and to further calculate the 14 C concentration of the soil fraction.

14 C and age heterogeneity of POM ( 14 C POM ) in the Lena Delta
Terrestrial POM enters a river predominantly by physical weathering of adjacent soils (Raymond and Bauer, 2001, and references therein).Compared to vegetation utilizing modern atmospheric 14 CO 2 , the bulk POM of soil is pre-aged.The specific residence time of POM in the soil before entering the river depends on various environmental factors like humidity, temperature, topography, soil type, size and topography of the catchment area (e.g., Kusch et al., 2010;Oades, 1988;Raymond and Bauer, 2001;Trumbore, 2009Trumbore, , 1993)).Furthermore, the fluvial transport of POM (and TSM) is governed by hydrological characteristics like runoff, discharge and flow velocity, sedimentation along riverbanks and floodplains, and resuspension of deposited material.Therefore, the age of ter-restrial POM is a combination of its residence time within the soil plus its residence time within the river basin, which can differ substantially for different terrestrial POM fractions (lipids versus lignin) from Arctic watersheds (e.g., Feng et al., 2013;Gustafsson et al., 2011).The Lena Delta and Buor Khaya Bay surface water 14 C POM concentrations presented here are depleted with respect to the current atmospheric 14 CO 2 (Table 3).The 14 C values range from −391 to −55 ‰ for 2009 to 2011, translating into 14 C ages > 400 years and up to 3920 years BP for samples taken close to the Pleistocene IC deposits of Kurungnakh Island (Fig. 2q-t, Table 3).These results are within the range of reported 14 C values for surface water POM offshore of the Lena Delta and influenced by Lena River outflow (Karlsson et al., 2011).Based on these results it seems reasonable to assume that a large fraction of Lena Delta POM originates from physical weathering of relatively young Holocene soils (active layer) of the first delta terrace and south of the delta.
However, soils outcropping along the Lena River south of the delta can be substantially older than late Holocene ages (Kuptsov and Lisitsin, 1996), including some areas of Pleistocene IC deposits (Grosse et al., 2013).Also, OM within the active layer and shallow permafrost table can be as old as 3000 years BP in the Lena Delta (30 cm below surface, Höfle et al., 2013) or more than > 10 000 years BP south of the delta (Kuptsov and Lisitsin, 1996).As shown by the lignin phenol composition of Lena Delta POM (Winterfeld et al., 2015), approximately half of the surface water POM is derived from the catchment south of the delta with considerable contributions from delta soils, particularly in the summer season, when riverbank erosion is strongest.Considering that, we would expect generally older POM 14 C ages.Additionally, riverbank erosion contributes POM covering the 14 C age range of the whole soil profile, not only from the active layer.
One possible explanation for our relatively young POM 14 C ages could be the contribution of plankton-derived OM with a rather modern 14 C concentration concealing the "true" age of soil-derived OM.Phytoplankton-derived OM contribution was also suggested to be the reason for relatively young POM from the Ob' River inferred from a core of a floodplain lake (Dickens et al., 2011).Although the overall contribution and flux of autochthonous phytoplankton OM in the Lena River is rather small or even negligible due to the high turbidity and low light penetration (Cauwet and Sidorov, 1996;Sorokin and Sorokin, 1996), phytoplankton can be quite abundant in the surface water (Kraberg et al., 2013;Sorokin and Sorokin, 1996), which we sampled for our POM analyses.In the following Sect.4.3 we present results from two different approaches to quantify the soil OM fraction in our samples and further estimate its 14 C concentration based on the assumption of plankton-derived OM with modern 14 C values.

Quantitative POM source determination and soil
14 C concentrations

Binary mixing model scenario 1
POC : PN ratios around 6 are usually associated with OM derived from algal/bacterial primary production and higher ratios of > 20 with OM derived from soils (Hedges et al., 1997;Meyers and Lallier-vergés, 1999 and references therein).However, as noted above (Sect.4.1) adsorption of inorganic nitrogen derived from OM decomposition (e.g., ammonium) to clay minerals (Schubert and Calvert, 2001), which is not accounted for when determining the total PN content, may additionally affect the POC : PN ratio in TSM, leading to lower values than would be expected from mixing of the two end members alone.Likewise, selective degradation of labile organic carbon (Kuhry and Vitt, 1996) would also result in lower POC : PN ratios.Both processes would lead to low estimates of the soil-derived OM contribution and suggest a higher contribution from plankton.Both explanations were also suggested by Sanchéz-García et al. (2011) for POM in the Laptev Sea offshore of the Lena Delta with unusually low POC : PN ratios.
In order to estimate the inorganic nitrogen content for our sample sets from each year, we used the intercept of the POC versus PN content regression line at POC = 0 (Fig. S2).By subtracting these amounts from the analyzed total PN content of the respective samples, we could calculate new soil POC : PN ratios (POC : PN NEW ) for our samples (Table S4 Supplement).Based on these POC : PN NEW ratios (and their uncertainty, derived from the standard error of the POC versus PN regressions), we calculated the soil-derived fraction within our POM samples using a simple two-end-member mixing model as follows: where POC : PN NEW is the corrected value of the POM sample (Table S4).As the end member for POC : PN soil we chose 23.7 ± 11 (Tables 4 and S1), which is an average calculated from Lena Delta first terrace soils presented in Winterfeld et al. (2015) and delta soils as well as soil from the Lena River watershed covering the taiga to tundra transition from Höfle et al. (2013), Zubrzycki (2013), and Zubrzycki e al. (2012, individual values in Table S1).The POC : PN plankton endmember value is 6 ± 1 (Table 4; e.g., Meyers, 1994).
The calculated soil-derived OM fractions of Lena Delta POM varied from 0.26 to 0.70 (mean of 0.35, n = 21, Table 5) for summer 2009, from 0.26 to 0.44 (mean of 0.34, n = 11) for summer 2010, and from 0.39 to 0.50 (mean of 0.43, n = 9) for late June/early July 2011 (Table 5).We used the POC versus PN (wt %) regression line from the summer 2010 samples within the delta to correct the two samples taken outside of the delta in the Buor Khaya Bay.  had calculated soil fractions of 0.39 (sample 35) and 0.12 (sample 36).The same was done for the single sample from late May 2011.We used the regression line from the samples taken 4 weeks later in June/July, and the soil fraction of sample 37 was 0.13 (Table 5).Note that these soil-and plankton-derived OM fractions can only serve as rough estimates.Without determining the particulate organic nitrogen directly for every sample, our POC : PN NEW ratios (Table S4) might be highly over-and underestimating OM fractions in individual samples.
The calculated soil and plankton OM fractions (f soil and f plankton , Table 5) were further used in an isotopic mass balance to determine the 14 C concentration of the soil fraction assuming the plankton-derived OM is modern ( 14 C plankton = 41.9 ± 4.2 ‰, Table 4): wheref POM , f soil , and f plankton are the fractions of POM and soil-and plankton-derived OM, and 14 C POM , 14 C soil , and 14 C plankton are the 14 C values of these sources.As noted above, 41.9 ± 4.2 ‰ is assumed as a maximum estimate for 14 C plankton based on atmospheric CO 2 14 C values for May to August 2009-2011 from the Schauinsland observatory, Germany.
The results of the newly calculated soil 14 C values using POC : PN NEW ratios ( 14 C POC : PN ) to partition between soil-and plankton-derived OM are shown in Table 5.The soil 14 C POC : PN values range from −884 to −322 ‰ for the sampling period of 2009-2011, translating into soil 14 C ages > 3000 years BP with an average of 6700 years BP (Table 5).Analogous to the two comparatively old bulk 14 C POM ages off Kurungnakh Island taken in 2009 and 2010 (Table 3), the calculated 14 C POC : PN age for sample 1 was 17 250 years BP and yielded a fossil age (> 50 000 years BP) for sample 31.Considering that the riverbank outcrops of Kurungnakh Island cover an age range of approximately 100 kyr with organic-rich IC deposits of about 50 kyr (Wetterich et al., 2008), these 14 C ages seem to be realistic.
Again, the uncertainties associated with the contribution of inorganic nitrogen to our total PN contents are rather high, resulting in a relatively rough estimation of soil and plankton OM fractions.These uncertainties are further affecting the calculation of the soil 14 C POC : PN concentrations.That means the calculated 14 C ages are estimates.Yet, they demonstrate that a possible underestimation of soil-derived OM ages can be considerable.

Binary mixing model scenario 2
Similar to the approach discussed in Sect.4.3.1,we used a second scenario based on δ 13 C values to distinguish between soil-and plankton-derived OM in our Lena Delta POM samples.The vegetation in the Lena River catchment (taiga and tundra) is dominated by C3 plants with a δ 13 C of around −25 to −27 ‰ (Rachold and Hubberten, 1999).This is also reflected in our δ 13 C data from the first delta terrace soils, with an average of −26.2 ‰ (n = 7, Table 3).Bird et al. (2002) determined the δ 13 C composition of taiga and tundra soils (excluding peatlands) along Yenisey River on a latitudinal transect.For the binary mixing model we used a soil OM end-member value of −26.6 ± 1 ‰ (Table 4), which is a combination of δ 13 C data from this study and the literature (Table S2; Bird et al., 2002;Pitkänen et al., 2002;Xu et al., 2009) covering tundra and taiga soils in Siberia and Alaska.
The δ 13 C composition of riverine plankton POM depends on the fractionation between phytoplankton and dissolved in-organic carbon (DIC).The distribution of the different DIC fractions (dissolved CO 2 , bicarbonate (HCO − 3 ), and carbonate ion (CO 2− 3 ) varies depending on temperature and pH.In the lower reaches of the Lena River and the Lena Delta, > 90 % of the DIC is made up of bicarbonate (Alling et al., 2012), i.e., the δ 13 C of bicarbonate represents the δ 13 C of DIC.Sources for DIC are generally the CO 2 derived from soil OM degradation, CO 2 released during the dissolution of carbonates, and CO 2 from the atmosphere.The Lena River geochemistry is mainly influenced by carbonate weathering and groundwater in the summer season (Gordeev and Sidorov, 1993).Assuming a fractionation of −22.5 ± 2.5 ‰ between phytoplankton and DIC (Mook and Tan, 1991) and a δ 13 C DIC of −8 ‰ for the Lena Delta (Alling et al., 2012), a plankton δ 13 C end-member value of −30.5 ± 2.5 ‰ would be expected.Similar or more depleted δ 13 C values of bicarbonate and phytoplankton POM were also determined in the Yenisey and Ob' rivers (Galimov et al., 2006) and in temperate estuaries (e.g., Ahad et al., 2008;Chanton and Lewis, 2002).Our most depleted δ 13 C POM of −34.2 ‰ for a sample from the summer 2009 (Table 3) is even lower than the plankton end member used here (δ 13 C plankton = −30.5 ‰).The sample location is offshore of Muostakh Island (Fig. 1b) and influenced by mixing with marine waters, which complicates the DIC composition and processes affecting δ 13 C of DIC (Alling et al., 2012).
We used the following two-end-member model to estimate the soil-and plankton-derived OM fractions (f soil , f plankton ) of our POM samples: where δ 13 C POM is the analyzed δ 13 C value of our POM samples (Table 3), δ 13 C soil is the soil end-member value of −26.6 ± 1 ‰, and δ 13 C plankton is plankton end-member value of −30.5 ± 2.5 ‰ (Tables 4 and S3).
The calculated soil OM fractions varied from 0.12 to 0.51 in summer 2009 (mean of 0.32, n = 12, Table 5), from 0.34 to 0.59 in summer 2010 (mean of 0.45, n = 12), and from 0.44 to 0.59 in spring 2011 (mean of 0.53, n = 3).The Buor Khaya Bay POM sample 21 showed a soil fraction of 0.75 and the spring freshet value (sample 37) a soil fraction of 0.80 (Table 5).As discussed above in Sect.4.3.1,these soil OM fractions are rough estimates in the absence of direct plankton determination.
Furthermore, 14 C values of the soil-derived OM fraction were calculated for the contribution of modern plankton OM as described in Sect.4.3.1 using Eqs.(3) and (4).Because the soil contributions calculated in this scenario are slightly higher than in the POC : PN-based scenario, the 14 C calculations resulted in estimates that are less radiocarbon-depleted compared to the POC : PN scenario (Table 5).Nonetheless, the 14 C values based on the δ 13 C end-member model are considerably 14 C-depleted compared www.biogeosciences.net/12/3769/2015/Biogeosciences, 12, 3769-3788, 2015 to bulk POM 14 C concentrations.The estimated 14 C values range from −944 to −495 ‰ for all samples for 2009-2011 representing 14 C ages from 23 100 to 2370 years BP (Table 5).The oldest samples are, again, the ones taken close to Pleistocene IC deposits (samples 1 and 31).

Comparison of scenario 1 and 2
Of all samples from 2009 to 2011 presented here, 28 samples have a soil fraction estimate from both scenarios.For these samples the POC : PN scenario gives a lower mean soil fraction of 0.33 compared to 0.43 from scenario 2 (based on δ 13 C, Fig. S3).The comparison demonstrates that the error estimates determined in the end-member calculation are realistic, as only 6 (2) of 28 samples are outside 1 (2) standard deviations of the estimated error.As the uncertainty of the estimates is high, both estimates are only weakly correlated (removing the clear outlier sample 37, we get R = 0.3, p = 0.1).
Because we assumed a modern 14 C value for the phytoplankton OM, the estimated soil 14 C concentrations are related to the amount of soil OM calculated in both scenarios.The lower mean fraction in scenario 1 therefore resulted in a lower mean soil 14 C value of −569 ‰ ( 14 C age = 6700 year BP, n = 26) compared to −495 ‰ ( 14 C age = 5430 years BP, n = 28) for scenario 2.Besides the spring freshet sample from 2011, the two samples with the lowest bulk 14 C values (sample 1 and 31, Table 3) also showed one of the lowest estimated soil 14 C values, namely −884 ‰ and radiocarbon-free (fossil) for scenario 1 as well as −944 and −754 ‰ for scenario 2. The two samples were taken close to IC deposits within the delta, which illustrates the locally pronounced influence of these deposits.Another sample standing out is sample 19, taken outside the delta close to Muostakh Island, which is also composed of IC deposits (Fig. 1b).Its bulk POM 14 C value was highest ( 14 C = −55 ‰, Table 3), pointing to comparatively young OM, and its δ 13 C value was lowest (−34.2 ‰), pointing to a rather high contribution of phytoplankton.This is in our line of argumentation assuming a contribution of planktonderived OM with a modern 14 C value.Because of our chosen end members, the soil fraction in sample 19 was rather small in the 13 C-based scenario, and the resulting 14 C value of the soil fraction was ( 14 C age of 11 890 years BP) the second lowest at −774 ‰.
In contrast to scenario 1, it is more obvious in scenario 2 that the POM samples taken in late June/early July 2011 are more enriched in 14 C (younger) than the POM samples taken later in the summer (August 2009 and July/August 2010).Similar observation were made for DOM 14 C concentrations of the Lena River at Zhigansk ∼ 900 km south of the delta (Raymond et al., 2007) and the Kolyma River in East Siberia (Neff et al., 2006).Their explanation was that, due to the deepening of the active layer in summer, older soil layers are accessible for meltwater and groundwater, contribut-ing an older DOM signature to the river than in spring, when most soil is still frozen.A comparable scenario could explain estimated 14 C concentration decreasing from spring freshet to summer.In addition to active layer deepening, riverbank erosion is strongest during the summer and might contribute a considerable amount of soil-derived OM with a large 14 C age range to the Lena Delta surface water.

Holocene soil versus Pleistocene IC deposits
As an extension of the two-end-member model, we applied a dual-carbon-isotope (δ 13 C, 14 C) three-end-member model (after Karlsson et al., 2011;Vonk et al., 2010Vonk et al., , 2012) ) to separate the source contributions of phytoplankton, Holocene soil and Pleistocene IC deposits (ICD) using the following equations: and where f soil_Holocene , f soil_ICD , and f plankton are the fractions of Holocene soil, IC deposits, and riverine plankton contributing to each POM sample.The end-member values chosen for Holocene soil OM were δ 13 C −26.6 ± 1 ‰ and 14 C −282 ± 133 ‰.As the two soil end members have very similar δ 13 C values (see Table 4, Holocene soil δ 13 C = −26.6 ± 1 ‰, IC δ 13 C = −26.3± 0.67 ‰), the δ 13 C mainly determines the phytoplankton vs. total (Holocene + Pleistocene) soil fraction, whereas 14 C mainly determines the fraction of Holocene to Pleistocene soil.Therefore, the results concerning the total soil fraction are very similar to those derived from the two-end-member model (R 2 = 0.92, mean two-end-member soil fraction is 0.43, mean three-endmember fraction is 0.46, Fig. S4) than derived with the twoend-member model.
As shown in Table S5 and Fig. 4, the POM source determinations resulted in relative fractions of 0.19-0.83,0.11-0.65,and 0.07-0.34for plankton, Holocene soil, and IC deposits, respectively.Again, sample 31 taken off Kurungnakh Island (IC) stands out with the highest contribution from ICD (fraction of 0.34) and sample 19 taken outside the delta with highest contribution of phytoplankton (fraction of 0.83).Overall, with the end members chosen here and despite the high spatial variability within the delta, the riverine phytoplankton fraction contributes most (mean of 0.55) to the surface water POM compared to the Holocene soil OM (0.32) and IC deposits (0.14).The rather low OM contribution from IC deposits reflects the distribution of these deposits in the Lena watershed, where they are only locally concentrated within elevations up 400 m (Grosse et al., 2013).

Implications of estimated soil-derived POM 14 C
The two binary mixing models (scenario 1 and 2) discussed above allow an estimate of the soil 14 C values based on a contribution of modern phytoplankton-derived OM.Both scenarios show considerably 14 C-depleted soil-derived OM compared to the bulk 14 C POM concentrations.This implies that the bulk POM 14 C age of samples taken in surface water during the summer, when the riverine primary production is high, likely underestimate the age of the soil-derived OM transported by the Lena River.The estimated soil 14 C values and 14 C ages in both scenarios give a more plausible picture for soil-derived POM in the Lena River watershed.In contrast to DOM, which is restricted in its flow path to the unfrozen soil layers, POM is not exclusively derived from surface soils.It also originates from resuspension of accumulated pre-aged material along the river channels and from riverbank erosion.The latter contributes POM with 14 C concentrations representing the whole range covered by the respective riverbank bluffs.In the Lena Delta this is predominantly OM of late Holocene age with local inputs from IC deposits of Pleistocene age (e.g., Bolshiyanov et al., 2015;Schirrmeister et al., 2011;Schwamborn et al., 2002).About half of the POM in the Lena Delta originates from the boreal forest hinterland south of the delta (Winterfeld et al., 2015).In the hinterland the soils along the Lena River and its tributaries can be older than the delta soils, i.e., covering the whole age range from Holocene soils to Pleistocene IC deposits (e.g., Grosse et al., 2013;Kuptsov and Lisitsin, 1996).Our estimated soil 14 C ages of about 2370 to 23 100 years (Table 5) for both scenarios therefore better reflect these hinterland deposits contributing a heterogeneous 14 C age mix to riverine POM than the bulk POM 14 C ages.
The soil POM 14 C estimates as well as the Lena Delta first terrace soil data (δ 13 C and 14 C) presented here improve our knowledge of the stable and radiocarbon isotopic range characteristic for soil-derived OM exported by the Lena River to the Laptev Sea.This information is critical for modeling the OM contribution from different terrestrial (fluvial vs. coastal erosion) and marine sources to Laptev Sea sediments and thus help in characterizing and quantifying the OM pools released from permafrost thawing.Recent studies suggest that OM exported by Arctic rivers and OM derived from erosion of IC coasts differ in their mineral and OM composition and thus show different potential for remineralization by microorganisms after thawing as well as different modes of transport and burial (e.g., Feng et al., 2013;Gustafsson et al., 2011;Knoblauch et al., 2013;Vonk et al., 2012).This has a direct impact on how to assess the possibility of a positive carbon-climate feedback from permafrost degradation, which has the potential to enhance global greenhouse warming by releasing huge amounts of previously frozen OM to the atmosphere.
A promising approach to distinguish OM sequestered in Arctic sediments is the dual-carbon-isotope end-member  4. Individual values comprising the Holocene soil δ 13 C and 14 C end member are given in Tables S2  and S3 in the Supplement.Note that the bulk POM 14 C values are ordered from the lowest to the highest and the respective sample IDs are given on the top axis.simulation applied by Karlsson et al. (2011) and Vonk et al. (2012Vonk et al. ( , 2010) ) to Laptev and East Siberian Sea sediments.The authors use the δ 13 C and 14 C values of surface water suspended matter and surface sediments to quantify OM derived from Pleistocene IC deposits, soil/topsoil OM exported by Siberian rivers, and marine phytoplankton OM.Their end-member definitions for IC deposits and marine primary production are rather well constrained.In contrast, the soil/topsoil end member is more difficult to define, particularly when using indirect parameters such as riverine DOM and POM.Here the δ 13 C end member chosen for surface soil including our first terrace soil data (Table S2) and our estimates for soil-derived POM (Table 5) provide new δ 13 C and 14 C value ranges (Fig. 3b) for fluvially exported soil POM.Together with published δ 13 C values from tundra and taiga soils in Siberia and Alaska (Bird et al., 2002;Pitkänen et al., 2002;Xu et al., 2009), we defined a Lena River soil OM end member with a δ 13 C value of −26.6 ± 1 ‰.Based on the δ 13 C binary mixing model (scenario 2), the corresponding 14 C value is −495 ± 153 ‰ (Table 5).We favor scenario 2 over the POC : PN-based scenario, because it allows for the calculation of the soil fraction of each POM sample based on its bulk δ 13 C value, which is a mixture of soil and phytoplankton OM.In contrast, in the POC : PN-based scenario a constant value for particulate inorganic nitrogen is subtracted, which does not necessarily represent the actual inorganic nitrogen content of the particular sample.Moreover, it does not account for selective degradation of labile carbon, which could result in overestimating the phytoplankton contribution.
The above-proposed δ 13 C and 14 C values for the soil OM end member of the Lena River catchment make it more complicated to distinguish between soil/topsoil-derived OM from the river and IC deposits from coastal erosion (see Fig. 3b).The δ 13 C values of both end members are almost indistinguishable.The 14 C range of our soil-derived POM estimates is lower than the end members used by Karlsson et al. (2011) and Vonk et al. (2012Vonk et al. ( , 2010)).Furthermore, Höfle et al. (2013) have shown that OM within the first 30cm of a polygon rim of the Lena Delta first terrace can be 3000 14 C years old, which makes the end members chosen for fluvial exported soil/topsoil OM by Karlsson et al. (2011) and Vonk et al. (2012Vonk et al. ( , 2010) ) appear too young.Using the bulk surface water POM and DOM δ 13 C and 14 C values as end members could therefore highly over-or underestimate the soil OM contribution from permafrost watersheds and in turn highly over-or underestimate OM contribution from IC to marine sediments.
However, we are aware of the limitations and uncertainties associated with the soil 14 C estimates discussed above (and in Sects. 4.3.1,4.3.2).Without determining the phytoplankton biomass of each sample by microscopic counting or from chlorophyll a analysis, the plankton OM fraction calculated based on POC : PN ratios (Sect.4.3.1)and δ 13 C values (Sect.4.3.2) can only be regarded as rough estimates.These estimates give orders of magnitude of OM contribution from the individual sources rather than exact values, which in turn provide a possible range of 14 C values for soil-derived POM.An additional source of uncertainty is our assumption of a modern 14 C concentration of plankton OM.Without determining the 14 C concentration of the Lena River DIC, which is utilized by phytoplankton, we cannot be sure of a modern 14 C concentration.Lena River DIC is derived from several sources providing carbon with different 14 C concentrations.DIC derived from carbonate and silicate weathering is 14 C-depleted, DIC derived from soil respiration has a broad range of 14 C values due decomposition of soil OM pools of varying age, and DIC derived from exchange with the atmosphere has modern 14 C values.The resulting DIC 14 C value is a mixture of these sources depending on the varying contribution from each source.Furthermore, Tank et al. (2012) found that DIC yields are negatively correlated with continuous permafrost extent in the watersheds of the six largest Arctic rivers, including the Lena.This would imply that carbonate weathering is to some extent hampered by continuous permafrost making up 77 % of the Lena catchment area (Tank et al., 2012).In summary, because there is no DIC 14 C value available for the Lena River and there are too many factors influencing the DIC 14 C value, we made the simplified assumption of a modern 14 C of 41.9 ± 4.2 (average value of atmospheric CO 2 14 C May-August 2009-2011 from Levin et al., 2013).Consequently, if true 14 C values of DIC in the Lena were depleted relative to the modern atmosphere, the soil-derived POM would be less 14 C-depleted than estimated here.Thus, 14 C values for soils have to be considered minimum estimates or, in other words, the estimated 14 C soil ages have to be considered maximum ages.
The best way to determine the 14 C value of riverine soil-derived OM would be a biomarker-specific radiocarbon analysis using source-specific compounds, e.g., shortand long-chain alkanoic acids for plankton-and terrestrialderived OM, respectively.However, for these analyses large samples of POM are needed.The samples analyzed during our study were too small to allow for compound-specific dating.

Conclusions
There are only few data available on 14 C concentrations of POM from Lena River surface water, but regarding the likely positive carbon-climate feedback to greenhouse warming, the quality and fate of this permafrost OM pool in the coastal waters of the Laptev Sea is currently under debate (e.g., Feng et al., 2013;Gustafsson et al., 2011;Karlsson et al., 2011;Vonk et al., 2012).With this study we provide one of the first data sets on surface water POM 14 C concentrations from the Lena Delta sampled during the summers of 2009 and 2010 and during spring 2011 (n = 30 samples).The contribution of modern phytoplankton POM to these samples was, for one, estimated using binary mixing models based on POC : PN ratios and δ 13 C values, which allowed us to calculate the 14 C values of the soil-derived POM fraction.These soil 14 C estimates were low compared to the bulk POM 14 C values and therefore seem to represent the heterogeneous 14 C mix of soil OM ranging from Holocene to Pleistocene age (e.g., IC deposits) in the Lena River watershed more accurately.Moreover, we applied a dual-carbon-isotope three-end-member model to further distinguish between OM contributions from Holocene soils and Pleistocene IC deposits to the soil fraction.Here, we were able to show that the overall contribution of IC deposits to surface water POM in the Lena Delta was relatively low, which reflects the restricted spatial distribution of these deposits within the Lena watershed.Only samples taken close to IC deposits exhibited higher contributions of this source in the model, implying a small, locally pronounced influence on surface water POM before it becomes mixed with other soil-derived POM during fluvial transport.
Because of the limitations of our approach, particularly the assumption of modern phytoplankton OM without determining the 14 C concentration of the Lena River DIC utilized by phytoplankton, our 14 C estimates for the soil-derived fraction have to be considered minimum 14 C concentrations and maximum 14 C ages, respectively.Nonetheless, we propose average values for the soil POM isotopic composition based on our data and published values of δ 13 C = −26.6 ± 1 and 14 C = −495 ± 153 ‰ (Tables 5 and S2), which will be useful for dual-carbon-isotope simulations focusing on un-raveling the OM contributed by different terrigenous (fluvial vs. coastal erosion) and marine sources to Arctic sediments.
The complete data set presented here can also be found in PANGAEA (www.pangaea.de).
The Supplement related to this article is available online at doi:10.5194/bg-12-3769-2015-supplement.

Figure 2 .
Figure 2. Spatial distribution of organic carbon concentrations, POC : PN ratios, and stable and radiocarbon isotopic composition in surface water total suspended matter (TSM) samples from 2009 to 2011.TSM concentrations of Aug 2009, July/Aug 2010, and late May 2011 (panels a-c) are from Winterfeld et al. (2015).

Figure 3 .
Figure 3. Stable and radiocarbon isotopic values of Lena Delta surface water POM of (a) this study and literature data as well as (b) estimated isotopic data of soil-derived POM based on POC : PN and δ 13 C ratios.Ranges for different end members for topsoil, IC, and marine are taken from the literature.

Figure 4 .
Figure 4. Results of dual-carbon-isotope (δ 13 C, 14 C) three-endmember mixing model showing the fractions for riverine phytoplankton, Holocene soils, and Pleistocene IC deposits contributing to individual surface water POM samples.The end member used in the model can be found in Table4.Individual values comprising the Holocene soil δ 13 C and 14 C end member are given in TablesS2 and S3in the Supplement.Note that the bulk POM 14 C values are ordered from the lowest to the highest and the respective sample IDs are given on the top axis.

Table 1 .
Soil samples from riverbank bluffs and total suspended matter (TSM) samples presented in this study analyzed for particulate organic carbon as well as stable and radiocarbon isotope composition.Latitude and longitude are given in decimal degrees ( • ).Bluff height is given in meters above river level (m a.r.l.) measured in August 2009 and only applicable for the soil samples.The TSM samples were taken from the surface water layer at a sampling depth of ca.0.5 m.

Table 2 .
Particulate organic carbon (POC) contents in Lena Delta surface water(2009 to 2011)given in milligrams per liter (mg L −1 ) and percent based on sediment dry weight (wt %) as well as atomic particulate organic carbon (POC) to particulate total nitrogen (PN) ratios.Data on individual samples can be found in the Supplement (TableS4).Note that for August 2009 there are only n = 20 samples for POC (wt %), because the total suspended matter concentration was not determined for sample 19.

Table 3 .
Stable carbon isotope (δ 13 C) and radiocarbon composition ( 14 C) of Lena Delta first terrace soil profiles and surface water particulate organic matter(2009)(2010)(2011).Soil profile samples are given in meters below surface (m b.s.).Not determined is denoted by n.d.

Table 4 .
Meyers (1994)lues used for the binary mixing models based on POC : PN and δ 13 C, and for dual-carbon-isotope three-end-member mixing model.If not directly taken from the literature, individual values from the literature and this study used to calculate the end members can be found in TablesS1-S3.Meyers (1994)and references therein; 2 average and standard deviation of atmospheric 14 C values from May to August 2009-2011 from Levin et al. (2013); 3 from Vonk et al. (2012). 1

Table 5 .
Soil fractions calculated based on POC : PN ratios and δ 13 C values and associated estimated 14 C concentrations of the soil-derived POM fraction.Soil fractions are given with standard deviation (1σ ). 14C values are given with lower quantile (q16) and upper quantile (q84), representing the 68 % confidence interval (1σ for normal distributions). 14C values lower than −1000 ‰ were marked as "fossil".Not determined denoted by n.d.