Surface waters of the Arctic Ocean have the highest concentrations of dissolved organic carbon (DOC) and terrigenous dissolved organic matter (DOM) of all ocean basins. Concentrations of dissolved lignin phenols in polar surface waters are 7-fold to 16-fold higher than those in the Atlantic and Pacific oceans, and stable carbon isotopic compositions of DOM are depleted in 13C by 1–2‰ relative to those in the Atlantic and Pacific. The large contribution of terrigenous DOM from Arctic rivers is responsible for the elevated concentrations of DOC in polar surface waters. The distribution of terrigenous DOM in polar surface waters is very heterogeneous, but on average we estimate 14–24% of the DOC is of terrestrial origin. Stable nitrogen isotopic compositions were useful for distinguishing DOM of Pacific and Atlantic origins as well as terrigenous and marine origins. The size distribution and composition of lignin phenols provide some evidence of photochemical transformations of terrigenous DOM, but it appears this process is not extensive in polar surface waters. The extent to which terrigenous DOM is removed from the Arctic Ocean by microbial degradation is less clear and warrants further study. Physical transport of terrigenous DOC to the North Atlantic is a major mechanism for its removal from the Arctic. The East Greenland Current alone exports 4.4–6.6 Tg of terrigenous DOC annually to the North Atlantic. Terrigenous DOC of Arctic origin was identified for the first time in components of North Atlantic Deep Water. Preliminary estimates indicate that ∼1 Tg of terrigenous DOC is exported from the Arctic in Denmark Strait Overflow Water with an additional ∼0.7 Tg in Classical Labrador Sea Water. Together, these exports compose approximately 25–33% of the terrigenous DOC discharged annually to the Arctic via rivers.
 The Arctic Ocean is a relatively small basin surrounded by continents that have a major impact on its hydrography and biogeochemical cycles. The surface area of the Arctic Ocean (9.5 × 106 km2) covers a little over half of the drainage basin area of rivers (15.5 × 106 km2) discharging into it [Jakobsson et al., 2004; Rachold et al., 2004]. The freshwater discharged by Arctic rivers plays an important role in the maintenance of stratified polar surface waters, and the dissolved and particulate terrigenous materials delivered by rivers are a source of bioactive elements to Arctic waters and sediments. Riverborne sediments and dissolved organic matter (DOM) impart strong terrigenous signatures throughout the Arctic [Amon, 2004; Stein et al., 2004].
 The stratified surface waters of the Arctic Ocean share many characteristics of an estuary. The Arctic Ocean receives ∼10% of global riverine discharge, and the cold and relatively fresh polar surface waters are isolated from warmer waters below by a halocline [Aagaard and Carmack, 1989]. The distributions and concentrations of dissolved and particulate matter are spatially and temporally heterogeneous in polar surface waters [Wheeler et al., 1997; Guay et al., 1999; Bussman and Kattner, 2000; Amon and Benner, 2003]. High concentrations of dissolved organic carbon (DOC) and biomarkers of terrigenous DOM are common in polar surface waters [Opsahl et al., 1999; Benner et al., 2004]. Residence times for surface waters in the Arctic are relatively short and variable, ranging from a few to 20 years [Schlosser et al., 1994; Bauch et al., 1995]. The East Greenland Current (EGC) is the primary conduit for the export of polar surface waters to the North Atlantic. The EGC carries relatively high concentrations of DOC with a strong terrigenous signal, and it has been estimated that 20–50% of the terrigenous DOC discharged to the Arctic Ocean by rivers is exported to the North Atlantic by this current alone [Opsahl and Benner, 1998; Amon et al., 2003].
 Several studies have addressed various aspects of the abundance and distribution of terrigenous DOM in the marginal seas and central basins of the Arctic Ocean during the past 5 years [Kattner et al., 1999; Guay et al., 1999; Opsahl et al., 1999; Dittmar and Kattner, 2003; Benner et al., 2004]. These studies have clearly noted that terrigenous DOC is an abundant and widely distributed component of polar surface waters. The present study summarizes data on DOM and its terrigenous component from six expeditions to the region during 1996–1999. Lignin phenols and stable carbon and nitrogen isotopic compositions of 67 samples of ultrafiltered DOM (UDOM) are used to distinguish sources of DOM and to determine its abundance and diagenetic history in surface and deep waters. This paper provides novel insights about terrigenous DOM sources, the mechanisms of terrigenous DOM distribution and removal, and the presence of terrigenous DOM from Arctic rivers in North Atlantic Deep Water.
2. Materials and Methods
 Water samples were collected from various locations and depths in the Kara, Greenland, Norwegian, and Irminger seas, Fram Strait, and Arctic Ocean during cruises on the U.S. Navy submarines USS Pogy (1996), USS Archerfish (1997), and USS Hawkbill (1998), the German ice breaker FS Polarstern (ARK XIII/3, 1997, and ARK XIV/2b, 1998) and on the Russian research vessel Akademik Boris Petrov, in 1997 and 1999 (Figure 1). Stations in Fram Strait and the Greenland, Norwegian, and Irminger seas were chosen to represent different water masses flowing in and out of the Arctic Ocean. The East Greenland Current (EGC) was sampled at several locations from 81°N to 66°N to trace the export of Arctic DOM and polar surface water to the North Atlantic (Figure 1). Additionally, we collected samples from throughout the water column in the central Arctic Ocean and the Greenland, Norwegian, and Irminger seas. Surface samples in the southern Kara Sea were collected across salinity gradients from the mouths of two major Russian rivers (Yenisey and Ob'). Most of the samples from the central Arctic were collected from submarines between 35 and 236 m, with the exception of one depth profile between 10 and 1600 m over the Mendeleev Ridge between the Makarov and Canada basins and two discrete deep water samples, sampled during a Polarstern cruise in the Nansen and Makarov Basins. Additional information about sampling locations and hydrographic characteristics is given by Amon and Benner .
 Water samples (30–200 L) were collected from surface ships using Niskin bottles on a rosette CTD system, except in the Kara Sea where a stainless steel water sampler was used. On the submarine cruises, samples were collected from a through-hull water inlet system or a rosette CTD system lowered from the surface. Water samples were filtered through 0.2-μm-pore-size polycarbonate filters (Nuclepore) immediately after sampling or preserved with HgCl2 (65 μM) in the case of the submarine samples. Kara Sea samples and samples collected on submarines were stored until processing at the home laboratory. All other samples were ultrafiltered immediately after sampling. DOC concentrations in the Kara Sea and submarine samples did not indicate DOC losses or contamination during storage. The ultrafiltrate from samples collected during the submarine expeditions in 1997 and 1998 was acidified (pH 2.5) and passed through solid-phase extraction columns (Megabond Elut) to recover low-molecular-weight lignin phenols [Louchouarn et al., 2000]. Following extraction the columns were rinsed with acidified Milli-Q water and eluted with methanol. The methanol extract was dried and analyzed for lignin phenols as described below.
 Ultrafiltration was performed with Amicon DC-10 or Amicon Proflux M 30 systems with two spiral-wound polysulfone filter cartridges (S10N1; 1000 Dalton cutoff). Prior to ultrafiltration the water samples were filtered through 0.2-μm-pore-size polycarbonate filters. After concentration of the initial volume (30–200 L) to ∼1 L, the sample was diafiltered with 12–18 L of Milli-Q water to remove inorganic salts. The diafiltered sample was stored frozen in a polycarbonate bottle until further processing. The volume of the diafiltered concentrate was reduced to ∼75 mL by rotary evaporation and dried using a Savant SVC200 SpeedVac concentrator [Benner et al., 1997].
 Dissolved organic carbon (DOC) was measured using the high-temperature combustion method and either a Shimadzu TOC 5000 analyzer [Benner and Strom, 1993] or MQ-Scientific Inc. 1001 TOC analyzer [Qian and Mopper, 1996]. Filtered water samples were stored frozen until analysis at the home laboratory.
 Lignin phenols were analyzed following CuO oxidation of dried powders of UDOM or methanol extracts from solid phase extraction [Opsahl et al., 1999; Louchouarn et al., 2000]. Lignin phenols were separated and quantified using a Hewlett Packard 5890A gas chromatograph with a DB5-MS capillary column and a Hewlett Packard 5972 mass selective detector. Quantification was achieved using selected ion monitoring with cinnamic acid as an internal standard.
 Stable carbon and nitrogen isotope ratios of dried powders of UDOM were measured using a Finigan Delta Plus system with in-line combustion. Isotope ratios were calculated using the δ(‰) notation: [(Rsample/Rstandard) − 1] × 1000, where R is the ratio of 13C/12C or 15N/14N in the sample and standard (Pee Dee Belemnite carbonate and atmospheric dinitrogen).
 The Yenisey and Ob' rivers and plumes were sampled in the southern Kara Sea during early September and late August of 1997 and 1999. Concentrations of DOC were somewhat higher in the Yenisey than the Ob', and DOC concentrations were higher in 1999 than 1997 in both rivers (Table 1). Discharge from the Yenisey was higher in August 1997 (19,500 m3 s−1) compared to August 1999 (13,000 m3 s−1), whereas discharge from the Ob' was higher in August 1999 (31,700 m3 s−1) compared to August 1997 (13,800 m3 s−1 [Water Systems Analysis Group, University of New Hampshire, 2000]). The average DOC concentration in both rivers during the two sampling periods was 568 ± 75 μM. About 62% of the DOC was recovered as high-molecular-weight DOM by tangential-flow ultrafiltration using a membrane with a 1000 Dalton molecular-weight cutoff. The C/N of this ultrafiltered dissolved organic matter (UDOM) was high in both rivers, with an average value of 44±3. These values are similar to those for total DOM in these and other Siberian rivers [Lara et al., 1998; Lobbes et al., 2000; Köhler et al., 2003]. Concentrations and carbon-normalized yields of lignin phenols in UDOM were also higher in the Yenisey than the Ob'. There was a distinct compositional difference in lignin phenols from the two rivers, with the ratio of syringyl:vanilyl phenols being higher in the Ob' indicating a greater contribution of angiosperm lignin than the Yenisey. The stable carbon (δ13C) and nitrogen (δ15N) isotopic compositions of UDOM were similar in both rivers and averaged −27.5 ± 0.3‰ and 2.4 ± 0.4‰, respectively.
Table 1. Average (±SD) Characteristics of Dissolved Organic Matter in Arctic Ocean Water Massesa
Lignin Phenols, ng L−1
Dissolved organic carbon (DOC), ultrafiltered DOC (UDOC), sum of six lignin phenols (ng L−1), carbon-normalized sum of six lignin phenols (Λ 6; μg 100 mg OC−1), ratio of syringyl to vanilyl phenols (S/V), and stable carbon and nitrogen isotopic ratios (δ13C, δ15N) of UDOM.
Yenisey plume (Sept. 1997; n = 4)
Yenisey plume (Aug. 1999; n = 3)
Ob' plume (Sept. 1997; n = 5)
Ob plume (Aug. 1999; n = 5)
Yenisey and Ob' rivers (1997, 1999; n = 4)
East Greenland Current (n = 6)
Nordic seas surface (<250 m; n = 6)
Nordic seas deep (>1000 m; n = 4)
Polar Surface Water (<200 m; n = 24)
Arctic Ocean mid depth (200–600 m; n = 5)
Arctic Ocean deep (>1000 m; n = 6)
 Concentrations of DOC in surface waters of the Arctic Ocean were higher than those in the adjacent Greenland and Norwegian (Nordic seas) and Irminger Sea (Figure 2a). The average DOC concentration in polar surface waters was 80 ± 15 μM, and the average concentration in surface waters of the Nordic seas was 61 ± 4 μM (Table 1). Deep-water concentrations of DOC were similar (∼53 ± 3 μM) at all locations. Concentrations of lignin phenols in UDOM were also considerably higher (282 ± 251 ng L−1) in polar surface waters compared to deep Arctic waters (45 ± 21 ng L−1) and waters in the Nordic seas (37 ± 5 ng L−1; Figure 2b). Elevated lignin phenol concentrations (98 ± 12 ng L−1; Table 2) were observed in three deeper water samples from Denmark Strait and the Irminger Sea and in three samples from a station over the Mendeleev Ridge in the Arctic Ocean. Stable carbon isotopic signatures (δ13C) of UDOM were relatively depleted in polar surface waters (−22.5 ± 0.6‰) and relatively enriched in Arctic deep water (−21.1 ± 0.5‰; Figure 2c). Stable carbon isotopic signatures (δ13C) of UDOM in the Nordic seas were similar to those in Arctic deep waters, with values of −21.1 ± 0.1‰ in surface waters and −20.8 ± 0.1‰ in deep water (Table 1).
Table 2. Locations and Hydrographic Characteristics of Water Samples Collected in Denmark Strait and the Irminger Seaa
Dissolved organic carbon (DOC), ultrafiltered DOC (UDOC), sum of six lignin phenols (ng L−1), carbon-normalized sum of six lignin phenols (Λ 6; μg 100 mg OC−1), ratio of syringyl to vanilyl phenols (S/V), and stable carbon and nitrogen isotopic ratios (δ13C, δ15N) of UDOM. Potential temperatures (°C) are reported.
Lignin phenols, ng L−1
Λ 6, μg 100 mg OC−1
 A vertical profile of water samples was collected over the Mendeleev Ridge from the USS Hawkbill in August 1998. Concentrations of DOC ranged from 54 μM at 1600 m to 137 μM at 10 m (Figure 3a), and concentrations of total dissolved lignin phenols ranged from 93 to 1489 ng L−1 (Figure 3b). Elevated lignin phenol concentrations were observed throughout the water column at this location. Stable carbon isotopic signatures (δ13C) of UDOM ranged from −21.2‰ at 1000 m to −24.3‰ at 10 m (Figure 3c). The size distribution of DOC and lignin phenols was determined using tangential-flow ultrafiltration. An average of 22 ± 2% of the DOC and 76 ± 4% of dissolved lignin phenols was recovered in the high-molecular-weight fraction (i.e., UDOM) of DOM.
 Concentrations of DOC in polar surface waters were significantly correlated (r = 0.911; p < 0.001) with concentrations of lignin phenols (Figure 4a). Similarly, there was a significant correlation (r = 0.855; p < 0.001) between concentrations of DOC and δ13C values of UDOM in polar surface waters (Figure 4b). These strong correlations indicated that elevated concentrations of DOC in polar surface water were largely due to the occurrence of terrigenous DOC.
 Carbon-normalized lignin phenol values (Λ6) and δ13C values of UDOM in all marine samples collected in this study were significantly correlated (r = 0.868; p < 0.001; Figure 5). Values of Λ6 and δ13C ranged from 10 to 287 and from −24.3 to −20.7‰, respectively. The high Λ6 values and depleted δ13C values indicate a large fraction of the DOM is of terrestrial origin, whereas the low Λ6 values and enriched δ13C values are similar to those reported in other ocean basins [Benner et al., 1997; Opsahl and Benner, 1997; Hernes and Benner, 2002].
 The stable nitrogen isotopic compositions (δ15N) of all marine and riverine UDOM samples are presented in Figure 6. The δ15N values of river and Kara Sea samples ranged from 1.8 to 3.5‰ and are indicative of terrigenous DOM (Figure 6a). Most marine δ15N values fell in the range of 3.8–5.0‰. There was a single deep marine sample with a low δ15N (3.0‰). All five marine samples from polar surface waters of the Canada Basin had higher δ15N values (5.7–6.1‰). A single UDOM sample from 236 m in the Canada Basin, which has a substantial Atlantic water component (salinity 34.564), had a δ15N value of 4.8‰. A plot of δ15N versus δ13C values provides greater distinction between the riverine and marine samples (Figure 6b). Samples from the Nordic seas, which had low concentrations of lignin phenols, had enriched δ13C values (−20.7 to −21.3‰) and δ15N values in the range of 3.8 to 4.8‰. These samples are most representative of marine UDOM because they have very low terrigenous contributions and are of Atlantic origin. Samples from the Arctic Ocean, particularly those from polar surface waters, have isotopic compositions between those of the riverine and marine values for this region.
 In 1998 during the ARK XIV/2b cruise, we had the opportunity to sample two components of North Atlantic Deep Water (NADW) near their sites of formation. A sample of Denmark Strait Overflow Water (DSOW) was collected at 569 m depth in the Denmark Strait (Table 2). The concentration of DOC, C/N ratio, and stable C and N isotopic compositions of this UDOM sample were similar to those in other samples collected from the Arctic Ocean and Nordic seas (Tables 1 and 2). Likewise, the composition of lignin phenols (S/V ratio) in this DSOW sample was similar to that of other samples. However, the concentration of lignin phenols (107 ng L−1) was substantially higher, indicating an elevated concentration of terrigenous DOC (Table 2).
 Two samples were collected to the southwest of Denmark Strait at 1705 and 2413 m depths in the Irminger Sea (Table 2). The salinity and temperature of the sample collected at 1705 m were indicative of Classical Labrador Sea Water (CLSW), and those of the sample from 2413 m were indicative of DSOW in this region. As observed before with the sample collected in Denmark Strait, the concentrations of lignin phenols (85–103 ng L−1) were elevated relative to those of other samples (Table 2). It is important to observe that the chemical and isotopic characteristics of the DSOW samples from Denmark Strait and the Irminger Sea are very similar and consistent with an Arctic origin. The single sample of CLSW has similar characteristics that are also consistent with an Arctic origin.
4.1. Abundance and Distribution of Terrigenous DOM in the Arctic Ocean
 The Arctic Ocean has the highest concentrations of DOC of all ocean basins. Polar surface waters have particularly high concentrations of DOC [Bussman and Kattner, 2000; Anderson, 2002: Amon, 2004], yet they are among the least productive waters in the global ocean [Sakshaug, 2004]. Arctic rivers discharge a large volume of freshwater (3300 km3 yr−1) and terrigenous DOC (25 Tg C yr−1) to shelf and polar surface waters [Aagaard and Carmack, 1989; Rachold et al., 2004]. The freshwater influences water column structure and biogeochemical cycles, and as demonstrated in this study, terrigenous DOC maintains the high concentrations of DOC found in polar surface waters. Results of the present and several previous studies [Guay et al., 1999; Kattner et al., 1999; Opsahl et al., 1999; Benner et al., 2004] clearly establish that terrigenous DOM is an abundant component of polar surface waters. The abundance and distribution of terrigenous DOM in polar surface waters are highly variable and heterogeneous, but average concentrations of lignin phenols in polar surface waters are sixfold to sevenfold higher than in Arctic deep waters and Atlantic waters and approximately 16-fold higher than in Pacific waters (Figure 7). The DOM in polar surface waters also has the most depleted δ13C values of these ocean basins (Figure 7). These data demonstrate that surface waters in the Arctic Ocean have the highest concentrations of terrigenous DOM among ocean basins.
 On average, concentrations of terrigenous DOM in the limited number of deep-water samples from the Arctic Ocean examined in this study were not significantly higher than those in the Nordic seas and Atlantic Ocean [Opsahl and Benner, 1997]. However, all three deep-water (600–1600 m) samples collected over the Mendeleev Ridge had elevated concentrations of terrigenous DOM. Previous studies have indicated elevated concentrations of chloro-fluorocarbons in deep water over the Mendeleev and Lomonosov ridges [Carmack et al., 1997]. These authors identified topographically steered boundary currents as the primary mechanism for rapid ventilation to depths of 1500 m over the ridges. Dittmar  found evidence of terrigenous DOM in waters off the Laptev Sea shelf and slope where topographically steered boundary currents could transport it over the ridges. Therefore additional investigations are needed to determine the abundance and distribution of terrigenous DOM in deep waters of the Arctic Ocean.
 These new data on terrigenous DOM in the Arctic provide an opportunity to update previous estimates of the percentage of terrigenous UDOM in various water masses [Opsahl et al., 1999]. Two independent approaches were used based on carbon-normalized lignin phenol yields (Λ6) and stable carbon isotopic compositions (δ13C) in UDOM. The percentage of terrigenous UDOC was estimated using the following formula and average Λ6 values presented in Table 1: %UDOCter = 100 × (Λ6 water mass)/(Λ6 for Yenisey and Ob' rivers). The percentage of terrigenous UDOC also was estimated using a simple two-end-member mixing model with a δ13C of −27.5‰ for terrigenous UDOM (average for Yenisey and Ob' rivers) and −20.9‰ for marine UDOM (average for Greenland and Norwegian seas) and average δ13C values for water masses presented in Table 1. A range of values is presented representing the average estimates from the Λ6 and δ13C calculations. About 14–24% of the UDOC in polar surface water (including the East Greenland Current) is terrigenous. Deep waters in the Arctic have ∼3% terrigenous UDOC, a value that is similar to the 2–3% estimated for the Greenland and Norwegian seas. These estimates are consistent with those presented earlier [Opsahl et al., 1999; Amon et al., 2003]. Terrigenous DOM is abundant in the East Greenland Current, which carries a strong biogeochemical signal from Arctic drainage basins to the North Atlantic Ocean.
 The stable nitrogen isotopic composition of UDOM was introduced in this study as a tracer of DOM sources in the Arctic Ocean. As observed previously [Amon and Meon, 2004], the δ15N and δ13C compositions of UDOM from rivers were depleted in 15N and 13C relative to marine UDOM, and both were useful for distinguishing terrigenous and marine sources of organic matter. The δ15N and δ13C compositions of surface sediments on the Beaufort [Naidu et al., 2000] and Siberian [Guo et al., 2004] shelves also demonstrate that terrigenous organic matter is depleted in both heavy isotopes relative to marine organic matter. In addition, the δ15N of marine UDOM in Atlantic waters (∼4.8‰) was ∼1‰ depleted in 15N relative to marine UDOM of Pacific origin (∼5.9‰). It appears that the δ15N of DOM in polar surface waters could be useful for distinguishing waters of Atlantic and Pacific origin. Inorganic nutrient concentrations and ratios are currently utilized for this purpose [Jones et al., 1998].
4.2. Cycling of Terrigenous DOM in the Arctic Ocean
 The relatively high abundance of terrigenous DOM in polar surface waters of the Arctic Ocean is a clear indication of the magnitude of shelf to basin transport. The fate of terrigenous DOM in the Arctic Ocean is of considerable interest as it is a potentially important source of energy and bioactive elements for microbial food webs in polar surface waters. Photochemical and microbial oxidation are the predominant mechanisms for the removal of terrigenous DOM from other ocean basins [Kieber et al., 1990; Miller and Zepp, 1995; Opsahl and Benner, 1998; Benner and Opsahl, 2001; Hernes and Benner, 2002], but the significance of these processes in polar surface waters is uncertain. Lignin phenol compositions provide some evidence of photochemical alterations of terrigenous DOM in polar surface waters, as the S/V ratio of lignin phenols in polar surface waters (0.21 ± 0.06) is lower than the ratio (0.41 ± 0.12) observed in the Yenisey and Ob' rivers [Opsahl and Benner, 1998; Hernes and Benner, 2003]. However, most (∼75%) lignin phenols in polar surface waters were in the high-molecular-weight fraction of DOM, suggesting that photooxidation is not extensive [Opsahl and Benner, 1998; Hernes and Benner, 2003]. This observation is not surprising given the extended ice coverage and relatively low solar irradiation of polar surface waters.
 The extent and significance of microbial degradation of terrigenous DOM in the Arctic Ocean is also poorly understood. Köhler et al.  detected minimal degradation of Yenisey River DOM during long-term decomposition studies. Several studies reported conservative behavior of riverine DOM across the broad Eurasian shelves and in the Transpolar Drift, indicating minimal degradation of terrigenous DOM [Kattner et al., 1999; Guay et al., 1999; Fransson et al., 2001; Anderson, 2002; Köhler et al., 2003; Dittmar and Kattner, 2003]. In contrast, Hansell et al.  reported substantial degradation of terrigenous DOM in the Canada Basin based on the nonconservative relationship between DOC concentrations and salinity in polar surface waters. In the present study, concentrations of lignin phenols (127 ± 39 ng L−1) in polar surface waters of the Nansen Basin were similar to those (123 ± 28 ng L−1) in polar surface waters of the Canada Basin. The residence times of polar surface water are considerably longer in the Canada Basin than in the Eurasian Basin [Schlosser et al., 1994; Bauch et al., 1995; Hansell et al., 2004], so we might expect to see lower concentrations of terrigenous DOM in the Canada Basin due to more extensive biodegradation. However, the number of samples analyzed for lignin phenols is too small for a quantitative comparison between basins. In addition, variability in lignin phenol and DOC concentrations among Eurasian and North American rivers further complicates this approach for investigating the degradation of terrigenous DOM in the Arctic Ocean.
 It is unclear whether salinity is a conservative tracer in polar surface waters. Salinities of halocline waters in the Canada Basin are often higher than those of incoming Pacific waters that form the halocline, indicating that halocline waters include a large contribution of brine water produced during winter sea-ice formation on the shelves [Coachman and Barnes, 1961; Jones and Anderson, 1986]. Salts and DOM are rejected during sea-ice formation, resulting in brine that is enriched in these constituents [Amon, 2004; Rachold et al., 2004]. Brine formation can explain the high concentrations of DOC and lignin phenols observed in halocline waters [Opsahl et al., 1999; Amon, 2004; Benner et al., 2004]. Lower salinities are found in the upper mixed layer of the water column, indicating the presence of melt water from sea ice and freshwater from rivers. The seasonal processes of brine formation and sea-ice melting produce higher salinity water with elevated DOC and lignin phenol concentrations and lower salinity water with depleted concentrations. Salinity is not likely to be a conservative tracer under these conditions, so other approaches are needed to investigate the reactivity of terrigenous DOM in polar surface waters.
4.3. Export of Terrigenous DOM From the Arctic to the North Atlantic
Opsahl et al.  used lignin phenol yields and stable C isotopic compositions of UDOM to estimate that 2.9–10.3 Tg of terrigenous DOC is exported from the Arctic annually in the East Greenland Current (EGC). Using the same approach and assumptions with the data presented herein, we revise this estimate to 4.4–6.6 Tg terrigenous DOC exported annually in the EGC. Approximately 18–26% of the terrigenous DOC discharged annually to the Arctic via rivers is exported to the North Atlantic by the EGC alone. Thus a large fraction of the terrigenous DOM entering the Arctic Ocean is removed by physical transport rather than biological or photochemical processes.
 This study presents the first evidence of the injection of Arctic-derived terrigenous DOM into two components of North Atlantic Deep Water (NADW). Denmark Strait Overflow Water (DSOW), the densest component of NADW, was sampled in Denmark Strait and the Irminger Sea. Concentrations of lignin phenols were over 2.5-fold higher in DSOW (∼105 ng L−1) than in the deep Norwegian and Greenland seas (37 ± 5 ng L−1). DSOW formation is thought to occur in the Nordic seas and includes colder and less saline components of upper Arctic Intermediate Water [Swift et al., 1980]. There is evidence that a portion of DSOW forms by isopycnic mixing in the EGC [Strass et al., 1993]. As shown in this study and by Opsahl et al. , the EGC has high concentrations of lignin phenols (364 ± 150 ng L−1) and terrigenous DOM derived from Arctic rivers. The EGC is the most likely source of the elevated concentrations of lignin phenols in DSOW, because upper waters in the central Greenland and Norwegian seas have relatively low concentrations of lignin phenols (∼50 ng L−1).
 A sample from another major component of NADW, nominally referred to as Classical Labrador Sea Water (CLSW), was collected in the Irminger Sea. Recent evidence indicates that deep convection in the Irminger Sea forms a water mass with hydrographic characteristics similar to CLSW, which is formed in the central Labrador Sea [Pickart et al., 2003a, 2003b]. The concentration of lignin phenols (85 ng L−1) in this sample was relatively high, indicating an Arctic origin for most of this terrigenous DOM. If this water mass formed in the Irminger Sea, the likely source of the elevated concentrations of lignin phenols is the EGC off the southeastern coast of Greenland. If formed in the Labrador Sea, the lignin phenols are likely from Arctic Ocean surface waters that pass through the Canadian Archipelago and Davis Strait.
 Assuming that the average lignin phenol concentration (37 ng L−1) in deep waters of the Nordic seas is representative of preformed concentrations in subpolar waters, we estimate that CLSW and DSOW have Arctic-derived lignin phenol concentrations of 48 ng L−1 and 68 ng L−1, respectively. Using these concentrations of lignin phenols, we provide preliminary estimates of the fluxes of terrigenous DOM exported from the Arctic in DSOW and CLSW. Deep water formation rates in these regions are temporally variable and largely influenced by the mode of the North Atlantic Oscillation [Dickson et al., 1996; Pickart et al., 2002], so we use conservative values of 2 Sv each for water fluxes in DSOW and CLSW to derive estimates of terrigenous DOC fluxes [Pickart et al., 2003a, 2003b; Smethie et al., 2000]. The annual deep-water export of Arctic-derived terrigenous DOC is estimated to be about 1 Tg in DSOW and 0.7 Tg in CLSW. These fluxes account for about 3.8% and 2.7%, respectively, of the terrigenous DOC discharge by rivers to the Arctic Ocean. It appears that Arctic rivers are the ultimate source of most of the terrigenous DOM entering the deep ocean. The terrigenous DOM in Arctic rivers and polar surface waters, including the EGC, has a modern radiocarbon age [Benner et al., 2004], indicating that terrigenous DOM in NADW is considerably younger than the marine component.
 Concentrations of lignin phenols in components of NADW collected near Bermuda are 2- to threefold lower than those observed in Denmark Strait and the Irminger Sea [Opsahl and Benner, 1997] (also P. J. Hernes and R. Benner, Sources and fluxes of terrigenous organic matter into the North Atlantic Ocean and North Atlantic Deep Water, submitted to Organic Geochemistry, 2005) (hereinafter referred to as Hernes and Benner, submitted manuscript, 2005). The decrease in lignin phenol concentrations during southward transport could be caused by mixing with waters of lower lignin concentrations, or they could reflect temporal variability in the injection of terrigenous DOM into formation waters. Spreading rates based on CFC ages in NADW indicate that the deep waters near Bermuda were formed about 20 years ago [Smethie et al., 2000]. The intensity of convection and hydrographic characteristics of DSOW and CLSW formation waters have changed during the past 20 years [Dickson et al., 1996], and these changes could be reflected in the varying concentrations of terrigenous DOC in NADW. It is also possible that terrigenous DOM is removed from NADW by microbial decomposition during southward transport. The terrigenous DOC concentration in NADW at ∼65°N is ∼3 μM, whereas the concentration at ∼32°N is approximately half that value [Opsahl and Benner, 1997; Hernes and Benner, submitted manuscript, 2005]. The average residence time of terrigenous DOC in the ocean is <100 years [Opsahl and Benner, 1997; Hernes and Benner, 2002], so removal from deep water on decadal timescales is not unrealistic.
 We thank the scientists and crew aboard the USS Pogy (1996), USS Archerfish (1997), USS Hawkbill (1998), FS Polarstern (1997, 1998) and Akademik Boris Petrov (1997, 1999) for assistance in the collection of samples. We thank Steve Opsahl for his contributions to the analysis of samples collected during 1996 and 1997. This research was supported by grants from the U.S. National Science Foundation (OPP-9996438 and OPP-0125301), the German Federal Ministry of Education and Research (BMBF; Siberian river run-off, FKZ 03G0547A), and the EC (COMET).