Colored dissolved organic matter dynamics across the shelf-basin interface in the western Arctic Ocean

Authors

  • Céline Guéguen,

    1. International Arctic Research Center, University of Alaska Fairbanks, Fairbanks, Alaska, USA
    2. Now at Chemistry Department, Trent University, Peterborough, Ontario, Canada.
    Search for more papers by this author
  • Laodong Guo,

    1. International Arctic Research Center, University of Alaska Fairbanks, Fairbanks, Alaska, USA
    2. Now at Department of Marine Science, University of Southern Mississippi, Stennis Space Center, Mississippi, USA.
    Search for more papers by this author
  • Michiyo Yamamoto-Kawai,

    1. International Arctic Research Center, University of Alaska Fairbanks, Fairbanks, Alaska, USA
    2. Now at Institute of Ocean Sciences, Sidney, British Columbia, Canada.
    Search for more papers by this author
  • Noriyuki Tanaka

    1. International Arctic Research Center, University of Alaska Fairbanks, Fairbanks, Alaska, USA
    2. Now at Sustainability Governance Project, Creative Research Initiatives, SOUSEI, Hokkaido University, Sapporo, Japan.
    Search for more papers by this author

Abstract

[1] Spatial variations in the concentration and nature of colored dissolved organic matter (CDOM) in the western Arctic Ocean were examined by three-dimensional excitation/emission matrix (3-D EEM) spectroscopy. CDOM profiles showed distinctive features well correlated with hydrographic characteristics. CDOM fluorescence was particularly high at depths between 40 and 200 m (up to 3 fluorescence units (Fl.U.)) in both Chukchi Sea and Beaufort Sea transects. Penetration of the high CDOM signal, formed on the shelves, into the Canada Basin was confined to the upper halocline layer (salinity of ∼33.1). This layer had distinctive 3-D EEM fingerprints in fluorescence spectra, showing a marked terrestrial humic signature. The presence of CDOM in the halocline layer likely resulted from two main processes: the brine rejection during sea ice formation and transport across the sediment-water interface during early diagenesis. Despite the high primary productivity in the Chukchi shelf, CDOM contribution from in situ production seemed to have little influence on the overall CDOM distributions in the study area.

1. Introduction

[2] Although the Arctic Ocean only comprises ∼1% of the global ocean volume, it receives about 10% of the global river discharge [Shiklomanov, 1998]. Therefore a greater influence of terrigenous organic matter from the river discharge is expected in the Arctic Ocean than in other ocean basins. Indeed, a strong terrestrial origin of dissolved organic matter (DOM) has been reported in the surface water of the Arctic Ocean [Opsahl et al., 1999; Amon et al., 2003; Amon and Meon, 2004; Guéguen et al., 2005].

[3] Colored dissolved organic matter (CDOM), which absorbs light in the UV–visible range, represents one fraction of the DOM pool (up to 25% [Benner, 2002]). Changes in CDOM fluorescence reflect the effects of physical and chemical processes that occur in the water column as well as variations in CDOM composition from different sources [Coble et al., 1990; Mopper and Schultz, 1993; Coble, 1996; Del Castillo et al., 2000; Guéguen et al., 2005].

[4] An important characteristic of the Arctic Ocean is its large continental shelf, which represents approximately 30% of its surface area. During sea ice formation in winter seasons, dense water cascading over the continental shelf around the Arctic Ocean could be an important transport pathway for DOM and other chemical species [Aagaard et al., 1985; Macdonald et al., 2002; Amon et al., 2003]. For example, sea ice formation could induce expulsion of sea salts, nutrients, and DOM from the shelf to the arctic basin [Giannelli et al., 2001; Thomas et al., 2001]. The Arctic Ocean is thus a region well suited to the study of the fate and transport of terrestrial organic matter and thus the shelf-basin interactions.

[5] Another important feature of the upper surface water column in the Arctic Ocean is the dominance of a strong halocline that separates the surface water from the underlying Atlantic-origin waters. In the Canada Basin, Pacific-origin waters [Roach et al., 1995] make up a substantial part of the upper halocline layer. This Pacific-origin water entering via the Bering Strait can be highly modified on the shelves by runoff, sea ice formation, and primary production. Associated with the upper halocline is a prominent nutrient maximum [Jones and Anderson, 1986] due to sea ice formation [Aagaard et al., 1981; Melling and Lewis, 1982] and interactions with sediments [Moore et al., 1983] on the shelves. Since these processes could be sources of DOM, a transport of organic matter into the halocline layer from the shelves to the arctic basin could be expected [Walsh, 1995; Walsh et al., 1997].

[6] In the Canada Basin, the occurrence of the nutrient-rich Pacific-origin waters coupled with large river inflows supports a substantial primary productivity on the western arctic shelves (360 and 40 g C m−2 yr−1 in Chukchi and Beaufort shelves, respectively [Fasham, 2003]). Surface layer biological productivity leads to the production of DOM, which can eventually be exported to depth [Melnikov and Pavlov, 1978; Bertilsson and Jones, 2003]. Because the arctic shelves are usually relatively shallow, the interactions with sediments and pore waters may also be an important source of CDOM [Skoog et al., 1996], particularly during the resuspension events caused by storms, ice rafting and bottom currents [Reimnitz et al., 1992; Eicken et al., 1997]. These different processes occurring on the arctic shelves may constitute substantial sources of CDOM, and therefore may play an important role in the cycling of DOM in the Arctic Ocean.

[7] In the present study, the nature and origin of CDOM in the western Arctic Ocean was investigated with a special emphasis on the influence of terrestrial DOM and implications for shelf-basin interactions. The Chukchi shelves with high in situ production, and Beaufort shelves with a strong influence of large river discharge, were compared in term of origin and behavior of CDOM. Specifically, three-dimensional excitation/emission matrices (3-D EEMs) were used to distinguish terrestrial from marine CDOM.

2. Methods

2.1. Study Area

[8] Samples were collected during the R/V Mirai cruise (MR02-K05) between 2 September and 10 October 2002 at 25 stations along 3 transects across the shelf-basin interface in the western Arctic Ocean (Figure 1). These transects included (1) transect A in the Chukchi Sea, (2) transect B in the NW region offshore of Barrow, Alaska, and (3) transect C in the Beaufort Sea (Figure 1).

Figure 1.

Map of the study area showing three transects with bathymetry and sampling locations.

[9] Transect C in the Beaufort Sea off the Alaska-Canada border is in the proximity to the Mackenzie River plume and was therefore greatly influenced by terrestrial DOM from the river. The Mackenzie River, the fourth largest arctic river, is estimated to deliver an annual load of 2.65 × 106 t dissolved organic carbon into the Beaufort Sea [Telang et al., 1991]. Peak discharge occurs between May and July. Mackenzie River waters flow out onto the Beaufort shelves where they mix with seawater and are eventually transported to the western Canada Basin. In contrast, transects A and B in the Chukchi Sea are within the region with the highest primary production in the Arctic Ocean [Walsh et al., 1989; Sakshaug, 2004] but with less direct influence of riverine inputs.

[10] The sample collection was confined to the first 400 m of the water column and particularly to the surface layer and the upper halocline. Unfortunately, the lower halocline, characterized by a local oxygen minimum near S = 34 at 220–240 m [McLaughlin et al., 2004], was not sampled at any stations.

2.2. Sampling

[11] Seawaters were collected using a CTD/Rosette system with 36 Niskin bottles for measurements of CDOM, stable oxygen isotope ratio of seawater (in terms of δ18O), nutrients (N, P, Si), Chlorophyll-a (Chl a), dissolved oxygen and other hydrographic parameters. Immediately after collection, CDOM samples were filtered on a precombusted (6 h at 500°C) glass fiber filter (Whatman, GF/F, 0.7 μm). All samples were stored in the dark at 4°C in previously combusted amber glass vials until analysis.

2.3. Measurements of Nutrients, Chl a, δ18O, and CDOM

[12] Water samples for nutrient determinations were collected in 30 mL polyethylene bottles and analyzed onboard for nitrate, phosphate, silicate, nitrite and ammonium on an autoanalyzer (TRACSS-800) using standard methods [Grasshoff et al., 1983]. Dissolved oxygen was determined by the Winkler method. These determinations were performed on board immediately after collection.

[13] Chlorophyll a was determined on board using a fluorescence spectrophotometer (Turner Design) for samples collected from the first 200 m of the water column.

[14] The oxygen isotope ratio (δ18O) was measured with a Finnigan MAT-252 mass spectrometer connected to a CO2-H2O equilibration unit [Yamamoto-Kawai et al., 2005]. The precision of measurement was around 0.03‰ at the 95% confidence level. Some samples were measured in duplicate with two working standards calibrated against VSMOW. The working standards were also measured along with samples (four of each standard in each series of measurements). The uncertainty of δ18O values of each water sample, calculated from the standard error of standard measurements, was about 0.03‰.

[15] Values of δ18O were used to estimate the relative contribution of meteoric water (runoff + precipitation) and sea ice meltwater [e.g., Östlund and Hut, 1984; Macdonald et al., 1999, 2002; Yamamoto-Kawai et al., 2005]. Mass balance equations for the calculation are as follows:

equation image

where f, S and δ are the fraction, salinity and δ18O values and SW, MW and SIM designate Atlantic-originated seawater, meteoric water, and sea ice meltwater, respectively. The end-member oxygen isotope compositions used for the calculation are listed in Table 1. The respective fractions calculated for each water component gave a range of fractional composition of MW and SIM waters. The error in the calculated fractions due to uncertainty in end-member values was about 0.03 [Yamamoto-Kawai et al., 2005].

Table 1. Values of Salinity and δ18O for End-Members Used in the Mass Balance Equations
 Seawater (SW)Meteoric water (MW)Sea Ice Melt (SIM)
Salinity34.806
δ18O0.2−21−2.4

[16] Absorbance of DOM was measured on an UV–visible spectrophotometer (Agilent 8453) with a 5 cm quartz cuvette using Milli-Q water as a reference [Guéguen et al., 2006]. Values of average absorbance at 700 nm were set to zero to correct the spectra for refractive index effects [Green and Blough, 1994]. The measured absorbance was converted to absorption coefficient (m−1) according to the equation: a(λ) = 2.303 A(λ)/L, where A(λ) is the absorbance and L the path length of the optical cell in meters. The absorption coefficient at 300 nm has been reported as an index of CDOM abundance [Blough et al., 1993; Battin, 1998; Del Castillo et al., 1999].

[17] Fluorescence was measured using a Fluoromax3 Jobin Yvon spectrofluorometer equipped with two monochromators for both the excitation (Ex) light source and the emission (Em) detector [Guéguen et al., 2006]. The recorded spectra were corrected for the instrumental response characteristics [Ewald et al., 1993; De Souza Sierra et al., 1994]. Milli-Q water was used as a blank and subtracted from sample spectra. Since absorbance values were always lower than 0.1 at 370 nm using a 5 cm path length cell in this study, both reabsorption and inner filter effects were thus minimized [Ewald et al., 1984] allowing us to work on the linear domain of the fluorescence intensity variation with concentration. No samples were diluted before optical measurements. Three dimensional EEMs were generated by concatenating the emission spectra from 260 to 700 nm at forty excitation wavelengths ranging from 255 to 455 nm. Final EEMs were interpolated (4Ex/2Em) and smoothed (Savitsky-Golay, second-order polynomial over 5 points) using GRAMS/32 software (version 4.14, Galactic Industries) [Coble et al., 1993; Coble, 1996]. Spectral intensities were screened for maxima within defined excitation/emission pairs [Coble, 1996]. To determine the Exmax/Emmax of the visible humic-like region of spectra, maximum slope at each grid node of the EEM was calculated using the Terrain Slope function of Surfer (version 7, Golden Software) [Komada et al., 2002]. CDOM fluorescence intensities were converted to fluorescence units (Fl.U.) on the basis of measurements of corrected fluorescence intensity of 1 ppb quinine sulfate dihydrate (NIST 936a) in 0.105 M perchloric acid. As the region of interest for protein-like DOM (Ex/Em 270/327–348 nm) yielded essentially no fluorescence for quinine sulfate, we reported all our measurements as fluorescence intensity in counts per second (cps). However, comparison of our results with those determined using other instruments can still be achieved using the following intensity obtained with Milli-Q water. At the peak for protein-like fluorescence (Ex = 270 nm), Milli-Q water yielded 50,000 cps at 331 nm emission wavelength, which was at least 10 times lower than the protein-like fluorescence measured in the samples. The measurements of corrected protein-like fluorescence intensity were reported in 105 cps. In this study, we report the intensities at Ex/Em 270/327–348 and 370/457 nm, which was referred to protein-like fluorescence [Coble et al., 1990; Mopper and Schultz, 1993] and terrestrially derived CDOM fluorescence [Donard et al., 1989; De Souza Sierra et al., 1994], respectively.

3. Results

3.1. Hydrographic Data

[18] The temperature-salinity (T-S) diagram for the three sections showed a mixture of several water masses in the western Arctic Ocean (Figure 2). Three main water masses composed the vertical structure [Aagaard et al., 1985]: a surface layer, Arctic Surface Water (ASW), influenced by runoff and surface exchange processes; a warm intermediate layer (T > 0°C), Arctic Intermediate Water (AIW), derived from Atlantic-origin water; and a high-density deep layer (σ ∼ 28.09, S ∼ 34.96, T ∼ −0.5°C), Canadian Basin Deep Water (CBDW). A pronounced cold halocline (−1.7 < θ < −1.2°C) occurring at depths of about 40–200 m separated ASW from AIW. The lowest salinity in ASW was found in the Beaufort transect (S < 26.5) due to the proximity of direct inputs from the Mackenzie River [Guéguen et al., 2005].

Figure 2.

T-S diagram for transects A (solid circles), B (solid triangles), and C (open squares). Abbreviations: ASW, Arctic Surface Water; AIW, Atlantic Intermediate Water; CBDW, Canadian Basin Deep Water.

3.2. Distributions of Inorganic Nutrients and Chl a

[19] Overall, phosphate concentrations ranged from 0.45 to 2.38 μM, with a general increase from the surface to the bottom of the halocline (Figures 3 and 4) . Highest concentrations were found between depths of about 40–250 m in the basin. The phosphate maximum (up to 2.38 μM) overlaid the sediments at the shelf-slope break in transect A, indicating potential sedimentary sources. Concentrations of nitrate and silicate (Figures 3 and 4) ranged from 0.01 to 16.6 μM and 2.5 to 53.6 μM, respectively. The highest concentrations were found at the shelf-slope break and between depths of about 40–150 m in the basin. Nitrite concentration ranged from 0.01 to 0.26 μM, and the maximum concentrations were measured just above the sediments on the shelf and at the shelf-slope break (not shown). Ammonium concentrations ranged from 0.01 to 6.58 μM, with the highest concentrations in bottom waters over the shelf and at the shelf-slope break (not shown). Ammonium concentrations decreased northward along transects A and C, showing little variation with depth at the northern stations. Transport of nutrients along the halocline from shelf to basin was consistent with the distribution pattern of water temperature discussed in the next section.

Figure 3.

Contour plots of (left) phosphate (μM) and (right) silicate (μM) in the upper 400 m water column extending from the Chukchi (sections A and B) and Beaufort (section C) seas to the Canada Basin. Isohaline 33.1‰ is superimposed on map contours to highlight the cold halocline layer.

Figure 4.

Vertical profiles of potential temperature, salinity, nutrients (NO3, PO4, and SiO2), and CDOM in the basin for transects A (solid circles, 162.5°W, 74.2°N), B (solid triangles, 157°W, 73.5°N), and C (open squares, 162.5°W, 74.2°N). The shaded region depicts the halocline layer characterized by a temperature minimum (see section 3.1) and nutrient maximum.

[20] High nutrient concentrations in the Chukchi Sea support one of the most productive areas in the world's oceans [Walsh et al., 1989]. Indeed, Chl a concentrations on the Chukchi shelves (sections A and B) were as high as 2.6 μg/L, 7 times higher than concentrations measured on the Beaufort shelves. In the basin, Chl a concentrations were lower than concentrations on the shelves but remained high for transects A and B (as high as 2.03 and 1.13 μg/L, respectively, compared to 0.55 μg/L for transect C).

3.3. Oxygen Isotopic Composition (δ18O) and Freshwater Sources

[21] Distributions of MW (including freshwater runoff and precipitation) and SIM fractions are shown in Figure 5 as isopleths of fractional composition, fMW and fSIM, respectively. As expected, MW was mostly found in the surface layers. The higher fraction of MW was found in transect C with about 7.5% to 15% in the upper 100 m water column. The 10% isoline was found at 5–10 m in the shelf and 50–75 m in the basin.

Figure 5.

Contour plots of fractional composition for (left) meteoric water (fMW) and (right) sea ice meltwater (fSIM) in transects A, B, and C. Isohaline 33.1‰ is superimposed on map contours to highlight the cold halocline layer. Positive fSIM numbers mean addition of freshwater by melting of sea ice, while negative fSIM numbers indicate brine rejection during sea ice formation.

[22] SIM was predominantly distributed in the upper 25 m water column and it tended to be more concentrated on the shelf. At a depth of about 30–40 m, fSIM dropped below zero because of brine injection, indicating a lens of remnant brine from sea ice production during the previous winter. This lens of water was usually located between 50 and 200 m. In transect C, thick and highly concentrated brine (values of fSIM up to −0.075) spread out from the Beaufort shelf to the basin. In contrast to transect C, the brine lenses in transects A and B were concentrated toward the basin region at 74.2°N and 73°–73.4°N, respectively. The importance of the brine lens in the basin area was comparable between transects A and B (f ∼ −0.04).

3.4. CDOM Distributions

[23] Absorption coefficient at 300 nm (a300) relative to CDOM abundance [Blough et al., 1993; Battin, 1998; Del Castillo et al., 1999] ranged from 0.31 to 4.15 m−1 (Figure 6). The highest value for a300 was found on the shelf break in transect C. CDOM fluorescence (Ex/Em 370/457 nm) ranged from 0.5 to 3.1 Fl.U. (Figure 7) and showed a maximum between 40 and 200 m water depth (Figures 47). In transect C, the CDOM maximum was found on the shelf break, whereas in transects A and B in the Chukchi Sea, the CDOM maximum was found in the basin around 74.2°N and 73°–73.4°N, respectively. These CDOM distributions showed a pattern similar to those of SIM fractional composition shown in Figure 5. In contrast to transect C, the CDOM-rich layer was not found on the shallower shelf in transects A and B. In addition, the prominent maximum observed in the CDOM-rich waters from the slope in transect C was not diffused uniformly but was rapidly mixed into the basin, likely influenced by alongshore currents. Similar to CDOM maximum distributions, the protein-like maximum (Ex/Em 270–275/327–348) was also measured between depths of 40 to 200 m (Figure 7). In transect C, a protein-like maximum was found on the shelf extending to the slope. In both transects A and B, the protein-like maximum was found in the basin at depth of about 100–200 m.

Figure 6.

Contour plots of a300 (m−1) in the upper 400 m water column in transects A, B, and C. Isohaline 33.1‰ is superimposed to highlight the cold halocline.

Figure 7.

Contour plots of (left) CDOM (Ex/Em 370/457nm, Fl.U.) and (right) protein-like (Ex/Em 270–275/327–348, ×105 cps) in sections A, B, and C. Isohaline 33.1‰ is superimposed to highlight the cold halocline layer.

4. Discussion

4.1. CDOM Dynamics

[24] Measured nutrient concentrations were highest in transect A (Figure 3), with bottom water levels above 30, 12 and 1.7 μM for silicate, nitrate and phosphate, respectively. Penetration into the Canada Basin of the high nutrient signal that was formed on the shelves was confined to layers with a salinity of ∼33.1, which defined the upper halocline [Jones and Anderson, 1986]. Similar to nutrients, CDOM values were particularly high at depths between 40 and 200 m (up to 3 Fl.U., Figures 47), coinciding with low temperature, and centered around S ∼ 33.1. This prominent CDOM maximum was thus associated with the upper part of the halocline. Cooper et al. [2005] also found a maximum CDOM fluorometer voltage in waters with salinity close to 33.1‰ This halocline overlays the more saline Atlantic waters (S ∼ 34.8) where CDOM values are lower but relatively homogeneous (0.87 ± 0.24 (1σ) Fl.U., n = 14) and comparable to values found in previous studies [De Souza Sierra et al., 1994]. Surface waters in the Beaufort transect had a larger fraction of freshwater (up to 15%, Figure 5) compared to the Chukchi transects probably because of the proximity of outflow from several Alaskan rivers and the Mackenzie River [Cooper et al., 2005; Guéguen et al., 2005]. That would likely contribute to the higher terrigenous DOM values found in the surface waters in transect C. This organic material might eventually be entrained and circulated westward into the Beaufort Gyre. A numerical sea ice–ocean climate model based on the Surface Heat Budget of the Arctic Ocean (SHEBA) experiment showed that surface ocean currents of 5 cm s−1 combined with the maximal peak discharge occurring in June–July were sufficient to transport fresh waters from the Mackenzie shelf to the Chukchi transects in September–October during our sampling campaign [Steele et al., 2006]. It is thus apparent that the riverine discharge is capable of contributing to the high offshore CDOM levels measured in transect A and B surface waters.

4.2. Processes Controlling DOM Distribution

[25] While the T-S diagram showed a well developed halocline in all three transects (Figure 2), the distribution of CDOM was somewhat different from that of nutrients such as silicate. The fact that CDOM distributions were more closely related to the fractional composition of SIM implied that, in addition to river inputs, lateral transport, and in situ biological production, ice formation/brine injection and alongshore transport processes were also important in controlling the distribution of CDOM and DOM in the western Arctic Ocean.

4.2.1. Terrestrial DOM Inputs

[26] The EEM spectra provide a comprehensive fingerprint of DOM which can be used to help distinguish the different origins of the CDOM [Coble et al., 1998]. The typical EEM for the halocline layer (samples collected between 40 and 200 m) showed a fluorescence maximum observed at Exmax/Emmax ∼ 314/415 nm (peak T, Figure 8), characteristic of transitional CDOM [Coble, 1996]. In addition to this dominant peak, two additional peaks were found at Ex/Em = 274/458 nm and Ex/Em = 358/452 nm, which were identified as peak A and peak C, respectively [Coble, 1996]. Such a two-peak pattern is usually found in humic material from terrestrial environments [Coble, 1996]. All EEMs measured between 50 and 200 m depth from the shelf to the basin in both transects showed a distinct peak C (Ex/Em = 338–369/435–462 nm) in the visible terrestrial humic region defined by Coble [1996], suggesting the occurrence of terrestrial humic material in the halocline.

Figure 8.

Fluorescence fingerprints representative of samples collected at 100 m depth on the shelf (141.7°W, 70.5°N, transect C) and in the basin (142°W, 72.8°N, transect C). The major peak regions for CDOM are labeled according to previous convention; peaks A, C, and T correspond to UV humic, visible terrestrial humic, and transitional humic material, respectively [Coble, 1996].

[27] Although the land signature in the CDOM pool was found in ASW as well as in the halocline in both transects, CDOM absorbance and fluorescence (Figures 6 and 7, respectively) in Chukchi transects were lower than in the Beaufort transect, which is consistent with the smaller fraction of MW found in the Chukchi transects based on δ18O results (2.5–10% in transects A and B; Figure 5). Higher CDOM fluorescence found in shelf bottom waters and at the slope/break in the Beaufort transect reflected the influence of river runoff (primarily from the Mackenzie River) [Guay and Falkner, 1997; Macdonald et al., 1999; Guéguen et al., 2005].

[28] The ratio of fluorescence to absorption at 370 nm showed that samples collected in the halocline had significantly higher relative fluorescence efficiency (p < 0.05) as well as higher silicate content than those found in ASW. For example, mean values for CDOM/a370 were 3.38 and 4.55 Fl.U./m−1 for ASW and halocline, respectively, while mean dissolved silicate concentrations were 8 and 35 μM for ASW and halocline, respectively (Figure 9). Changes in this ratio reflected differences in the photophysical properties of the CDOM and thus call into question whether the two materials were similar or not [Blough and Del Vecchio, 2002]. The difference between the ASW and the halocline indicated that these materials were different. Whether these differences arose because the CDOM came from different sources (e.g., terrestrial versus autochthonous or different terrestrial end-members) or through alteration of a single terrestrial source (e.g., photochemical alteration and/or fractionation during brine rejection) is less clear. However, all EEMs measured from the shelf to the basin in both transects in the halocline (between 50 and 200 m depth) showed a distinct peak C (Ex/Em = 338–369/435–462 nm) in the visible terrestrial humic region defined by Coble [1996], again suggesting the occurrence of terrestrial humic material in the halocline. In ASW, runoff contribution was relatively significant (up to 15%), suggesting terrigenous influence, particularly in transect C. Recent studies have suggested that photobleaching altered humic material in surface waters [Vodacek et al., 1997; Moran et al., 2000; Hernes and Benner, 2003], but additional work is necessary to evaluate the importance of photoalteration of terrigenous material in surface waters in the Arctic Ocean.

Figure 9.

ASW (solid circles), halocline (inverted solid triangles), and deeper waters (solid squares) characterized by CDOM/a370 ratio and dissolved silicate concentrations. The halocline is characterized by potential temperature ∼−1.5°C and S‰ ∼ 33.1.

4.2.2. Influence of Sea Ice Formation

[29] The majority of DOM can be excluded from the sea ice crystals into the brine solution [Belzile et al., 2000] where DOC concentration can be 3 orders of magnitude higher than the concentration of DOC found in sea ice [Eicken et al., 1995; Krembs et al., 2002; Eicken, 2003, and references therein]. Thus rejected brine from sea ice (negative fSIM) can be a potential source of DOM in the cold layer. Sea ice formation, brine injection and ice scouring can all result in DOM repartitioning, transformation and transport. As shown in Figures 5 and 7, CDOM distribution exhibited a pattern similar to that of the brine fraction (negative fSIM) estimated from δ18O data. In transects A and B, in particular, high CDOM values were observed in the basin where brine fractions were also high. Compared to ASW, the halocline was characterized by higher CDOM values (from 1 to 3 Fl.U.; Figure 10) and higher brine water fractions or lower SIM fractions (fSIM < −0.02; Figure 10). Although distributions of CDOM and fSIM were significantly different between ASW and the halocline (p < 0.05, ANOVA), the highest values of CDOM in the halocline were not always associated with the highest brine fractions. During winter, fresh water discharge was relatively low and CDOM levels on the shelves were thus relatively low. The level of DOM in the brine excluded from sea ice was consequently relatively low. Additional sources relatively high in CDOM must therefore be required to explain high CDOM levels in the halocline. Surface sediment samples in the Mackenzie estuary and the Alaskan shelves contained more than 1% (w/w) of organic carbon with the highest value of 1.7% near the Mackenzie River mouth [Macdonald et al., 2004, and references therein]. Results from shipboard incubation of Alaskan shelf sediments indicated that sediment could be a net source of DOM to overlying waters [Cooper et al., 2005]. Moreover, during ice formation, the brine that was generated formed high-salinity water that could be trapped along the shelf bottom. This dense water should flow over the surface of sediments, accumulating nutrients [Moore et al., 1983] and CDOM [Chen et al., 1993; Skoog et al., 1996; Seritti et al., 1997] in the process. This nutrient and CDOM accumulation may result from the leaching out of sediments by brine waters [Moore et al., 1983]. Indeed, relatively higher ammonium content and protein-like fluorescence (Figure 7) detected locally on the bottom shelf in transects A and B and at the slope/break in transect C revealed the release of DOM from sediments, likely enhanced by benthic boundary layer processes, such as resuspension [e.g., Guo and Santschi, 2000; Hamilton and Lugo-Fernandez, 2001]. The occurrence of a bottom nepheloid layer has been widely reported in other ocean margin regions [e.g., McCavem, 1986; Jahnke et al., 1990; Boss et al., 2001]. Although shelf sediments [Skoog et al., 1996; Cooper et al., 2005] and pore waters [Chen et al., 1993; Skoog et al., 1996; Seritti et al., 1997] can be a source of DOM, the amount of CDOM released into the water column has yet to be quantified.

Figure 10.

Relationship between fSIM (fraction of sea ice meltwater) and CDOM in ASW (solid circles), halocline (inverted solid triangles), and deeper waters (solid squares). The halocline is characterized by potential temperature ∼−1.5°C and S‰ ∼ 33.1.

[30] There was a significant amount of DOM within the sea ice due to ice algae production and release of algal exudates occurring at the bottom of the sea ice [Perovich et al., 1998; Belzile et al., 2000]. However, the amounts of sea ice with high DOM concentration as well as the overall contribution of sea ice melting (SIM < 10%, Figure 5) were too small to contribute significantly to DOM in ASW.

4.2.3. Autochthonous Production

[31] There was marked contrast in primary production between the Beaufort and Chukchi shelves. On the Chukchi shelves (transects A and B), the primary production was up to 360 g C m−2 yr−1, which was about ninefold higher than those found on the Beaufort shelves [Chen et al., 2003]. However, no significant correlation was found between CDOM and Chl a in the study areas, indicating that primary production did not make a direct, significant contribution to the overall intensity of CDOM. Chl a was also higher on shelves in transects A and B, but no parallel increases in CDOM were observed. The lack of observed relationships between CDOM (or DOM) and Chl a, similar to the results reported here, has been reported earlier [Vodacek et al., 1997; Nelson et al., 1998; Rochelle-Newall and Fisher, 2002; Guo et al., 2004].

[32] In the cold layer, high silicate, nitrate, and phosphate, together with low oxygen values, were usually associated with the decay of biogenic matter. The decomposition of particulate organic matter or the exudates of bacteria, phytoplankton and algal cells can be indicated by the presence of fluorescing amino acid (protein-like) materials [Ewald et al., 1986; Coble et al., 1990; Mopper and Schultz, 1993; Determann et al., 1998; Parlanti et al., 2000], which are also involved in CDOM production [Ferrari et al., 1996; Parlanti et al., 2000]. Our results show that the maximum abundance of protein-like materials was accompanied by high CDOM fluorescence measured in both transects (r2 = 0.55, p < 0.05), meaning that part of the CDOM could result from the degradation of recently produced organic matter. However, on the Chukchi and Beaufort shelves, the terrestrial CDOM background was so overwhelmingly large (see above) that it was difficult to unambiguously identify the contribution of in situ CDOM production.

4.3. CDOM as a Tracer for Shelf-Basin Interactions

[33] Spatial distribution of CDOM showed that ASW and the halocline were rich in CDOM. Riverine CDOM was carried away in the surface layer (ASW) on the shelf and was eventually dispersed further offshore. Physical processes appeared to be the dominant mechanism controlling the fate of CDOM in ASW [Guéguen et al., 2005]. CDOM values were significantly (p < 0.001) higher in the halocline than in ASW and deep layers.

[34] Since CDOM in the halocline was higher than in the Atlantic layer (1.93 and 0.87 Fl.U., respectively) the maintenance of the cold halocline cannot be supported by the upwelling and cooling of the Atlantic layer during winter. The significant terrestrial signal in the CDOM pool suggests the importance of lateral advection through the halocline from the continental shelves.

[35] The highest CDOM fluorescence in the halocline has been measured in the Beaufort transect, suggesting a great influence of terrestrial CDOM from the Mackenzie River. The presence of CDOM in the halocline likely results from two main processes, brine rejection during sea ice formation on the shelves and diagenetic processes in bottom sediments (and at the water-sediment interface). Eventually CDOM is incorporated into the cold layer and entrained in the Beaufort Gyre [Macdonald et al., 1999] to the western Canada Basin, thus establishing predominant supply routes of the terrestrial material to the Chukchi transects. This anticyclonic circulation feature is consistent with the presence of significant CDOM of terrestrial origin in the basin in transects A and B.

[36] The Beaufort Gyre also participates in redistributing ice-trapped organic material through cycles of melting and freezing [Pfirman et al., 1995; Schubert and Calvert, 2001]. The detection of CDOM associated with the brine lenses offered further evidence that Beaufort Gyre circulation provides an important component of the transport system which exports terrestrial material to the central Arctic Ocean. Further studies are required to assess the role of the Beaufort Gyre in CDOM cycles.

Acknowledgments

[37] We gratefully thank Tomoyuki Tanaka for his technical assistance during the field sampling, and the crew of the R/V Mirai and the support personnel from Marine Works Japan for their assistance and facilitation. We gratefully acknowledge K. Shimada and A. Murata for providing hydrographic data, Gordon Bower for measurements of δ18O and EEM concatenation, Andrew Vreugdenhil (Trent University), Candace O'Connor (UAF), and Julia Johnson (USM) for comments and English corrections on an early version of the manuscript, and three anonymous reviewers for critical comments. This work was supported, in part, by the Frontier Research System for Global Change through the International Arctic Research Center and a Swiss NSF fellowship.

Ancillary