Geophysical Research Letters

Utilizing chromophoric dissolved organic matter measurements to derive export and reactivity of dissolved organic carbon exported to the Arctic Ocean: A case study of the Yukon River, Alaska



[1] The quality and quantity of dissolved organic matter (DOM) exported by Arctic rivers is known to vary with hydrology and this exported material plays a fundamental role in the biogeochemical cycling of carbon at high latitudes. We highlight the potential of optical measurements to examine DOM quality across the hydrograph in Arctic rivers. Furthermore, we establish chromophoric DOM (CDOM) relationships to dissolved organic carbon (DOC) and lignin phenols in the Yukon River and model DOC and lignin loads from CDOM measurements, the former in excellent agreement with long-term DOC monitoring data. Intensive sampling across the historically under-sampled spring flush period highlights the importance of this time for total export of DOC and particularly lignin. Calculated riverine DOC loads to the Arctic Ocean show an increase from previous estimates, especially when new higher discharge data are incorporated. Increased DOC loads indicate decreased residence times for terrigenous DOM in the Arctic Ocean with important implications for the reactivity and export of this material to the Atlantic Ocean.

1. Introduction

[2] In tropical to temperate ocean basins, riverine dissolved organic carbon (DOC) is rapidly remineralized by microbial and photochemical processes, such that terrigenous DOC constitutes ∼1–2% of total DOC in these basins [Hernes and Benner, 2006]. However, in the Arctic Ocean, diminished sunlight and colder temperatures in conjunction with the greatest riverine input on a per volume basis of any ocean basin results in terrigenous DOC constituting up to one-third of the total DOC in Arctic surface waters [Opsahl et al., 1999]. Rivers draining into the Arctic Ocean exhibit high concentrations of allochthonous DOC [Lobbes et al., 2000; Raymond et al., 2007] and recent measurements indicate that DOC export is changing due to climatic warming and alteration in permafrost condition [Striegl et al., 2005; Guo et al., 2007]. Arctic catchments are estimated to contain over half of the global carbon stored in soils [Dittmar and Kattner, 2003]. Therefore, transformations resulting in the mobilization of this soil organic carbon pool due to global climatic change will likely have significant ramifications for the biogeochemical cycling of carbon, particularly the land-ocean carbon flux.

[3] Sampling of Arctic rivers has largely been constrained to late summer after break-up of sea ice in the Arctic Ocean. The highly degraded, unreactive nature of riverine DOM during this season has been assumed to be representative of year-round discharge [Dittmar and Kattner, 2003]. More recent work demonstrates seasonal variation in DOM composition [Neff et al., 2006; Spencer et al., 2008], lability [Holmes et al., 2008] and age [Raymond et al., 2007]. Arctic rivers exhibit a highly seasonal discharge with increased DOC concentrations at high discharge during the spring flush that results in a large proportion of DOC export occurring in a short time period. During the spring flush Arctic riverine DOM is younger, markedly more labile and terrigenous in nature (e.g. increased lignin carbon-normalized yields) than at any other time of year due to high surface runoff and leaching of surface litter and shallow soil layers [Holmes et al., 2008; Spencer et al., 2008].

[4] This study investigates the potential for using chromophoric DOM (CDOM) measurements to describe DOM dynamics across the hydrograph in the Yukon River at Pilot Station (YRP) as a model for all Arctic rivers, including simple proxies for both DOM composition and reactivity in these systems. CDOM measurements allow for rapid and relatively inexpensive monitoring of DOM and can also be measured in situ allowing high spatial and temporal resolution data to be collected [Amon et al., 2003; Spencer et al., 2007]. Furthermore, we examine the relationships between CDOM and DOC and CDOM and lignin phenols (unique tracers of vascular plant material) and utilize these relationships to derive the export of DOC and lignin phenols from YRP.

2. Materials and Methods

[5] The Yukon River (YR) drains a relatively pristine catchment of 853,300 km2 that includes large areas underlain by permafrost and is a major source of terrigenous DOM to the eastern Bering Sea and Arctic Ocean. Samples were collected from YRP (N 61°56′04″: W 162°52′50″), which is the furthest downstream location above the head of tides contained in a single channel. Samples were collected between late May 2004 and late October 2005, filtered in the field (0.45 μm) and shipped on ice to the USGS laboratory in Boulder (Colorado, U.S.A.).

[6] Dissolved organic carbon measurements were performed on an OI Analytical Model 700 TOC analyzer [Aiken, 1992]. UV-visible absorbance was measured at room temperature on a Hewlett-Packard photo-diode array spectrophotometer (model 8453) between 200 and 800 nm [Spencer et al., 2008]. The CDOM absorption ratio at 250 to 365nm was calculated (a250:a365) and SUVA254 values were determined by dividing the UV absorbance measured at λ = 254 nm by the DOC concentration. Spectral slope (S) was calculated using a non-linear fit of an exponential function to the absorption spectrum in the ranges of 275–295, 290–350 and 350–400 nm [Hernes et al., 2008; Helms et al., 2008].

[7] Daily riverine DOC, CDOM derived DOC, and lignin loads (mass d−1) were calculated from continuous water discharge (Q) for water years (October 1–September 30) 2004–2005 using the FORTRAN Load Estimator (LOADEST) program [Runkel et al., 2004]. The model requires a minimum of twelve measurements (n = 39) of the chemical concentration under investigation over a range of flow conditions and calculates loads by applying the method of adjusted maximum likelihood estimation (AMLE). LOADEST centers Q and chemical concentration data to eliminate colinearity and can automatically select one of nine predefined regression models to fit the data based on the AIC (Akaike Information Criterion) [Dornblaser and Striegl, 2007].

3. Results and Discussion

[8] On the basis of seasonal patterns of Q in the YR, the annual hydrograph can be split into three distinct periods: spring flush (May 1–June 30; n = 19), summer-autumn (July 1–October 31; n = 18) and winter (November 1–April 30; n = 2) [Striegl et al., 2005; Dornblaser and Striegl, 2007]. Concentrations of DOC at YRP were greatest during the spring flush and showed the largest range (6.0–17.0 mgL−1, equation image = 10.8 mgL−1) in comparison to the summer-autumn period (2.8–7.7 mgL−1, equation image = 4.0 mgL−1) and the winter (2.6–2.8 mgL−1, equation image = 2.7 mgL−1) (Figure 1a). This highlights the importance of the spring flush in terms of magnitude of the flux with respect to the annual DOC load, while the large range in spring shows the rapid transition to peak Q and DOC export. Specific ultraviolet absorbance at 254 nm (SUVA254) has been shown to be positively correlated to percent aromaticity of DOM [Weishaar et al., 2003]. SUVA254 values were greatest at high DOC concentrations during the spring flush (2.4–3.7 LmgC−1 m−1, equation image = 3.4 LmgC−1 m−1), declined at intermediary DOC concentrations throughout the summer-autumn period (2.5–3.4 LmgC−1 m−1, equation image = 2.9 LmgC−1 m−1) to lowest values at low DOC concentrations under-ice during the winter (2.0–2.4 LmgC−1 m−1, equation image = 2.2 LmgC−1 m−1) (Figure 1b). The a250:a365 ratio has also been previously related to changes in the aromaticity and molecular size of DOM with decreasing values relating to increasing aromaticity and molecular size [Peuravuori and Pihlaja, 1997] and a250:a365 mirrored the trends in SUVA254. During the spring flush the mean a250:a365 value was 5.08 (4.03–5.40), increasing during the summer-autumn to a mean value of 6.10 (5.45–7.04) and further increasing to a mean winter value under-ice of 7.22 (6.99–7.45) (Figure 1c). These simple measurements thus appear to track trends shown by other studies in Arctic rivers relating to DOM composition, age and lability [Raymond et al., 2007; Holmes et al., 2008; Spencer et al., 2008]. As such they have clear potential to be developed as simple proxy measurements for high resolution monitoring in these systems.

Figure 1.

Water discharge (Q) hydrograph at YRP (grey line) for 2004–2005 versus (a) DOC, (b) SUVA254, (c) a250:a365, (d) S275–295, (e) S350–400, and (f) SR.

[9] The spectral slope parameter (S) has been used to examine changes in DOM source and composition; typically a steeper S indicates low molecular weight material or decreasing aromaticity and a shallower S indicates DOM with a higher aromatic content and higher molecular weight [Blough and Del Vecchio, 2002; Helms et al., 2008]. Helms et al. [2008] proposed calculating S over the ranges 275–295 nm and 350–400 nm as their comprehensive studies of aquatic systems and DOM sources indicated the largest variations in S over these ranges. Helms et al. [2008] also showed S275–295 and SR (S275–295:S350–400) to be correlated to DOM molecular weight and source (the more allochthonous the sample, the higher molecular weight DOM, the lower the SR value). S was also calculated over the wavelength range 290–350 nm as previous studies have shown this region to be sensitive to changes in DOM source and composition [Hernes et al., 2008; Spencer et al., 2008].

[10] The three ranges of S investigated in this study all show consistent trends across the seasons. S290–350, S275–295 and S350–400 all showed their lowest mean values during the spring flush (15.07 × 10−3 nm−1; 14.00 × 10−3 nm−1 and 17.08 × 10−3 nm−1 respectively), increasing to intermediary mean values during the summer-autumn period (16.62 × 10−3 nm−1; 16.65 × 10−3 nm−1 and 18.06 × 10−3 nm−1 respectively) to highest mean values in the winter under-ice samples (18.13 × 10−3 nm−1; 18.96 × 10−3 nm−1 and 20.50 × 10−3 nm−1 respectively). Examining S values in more detail shows variation within seasons. For example, S275–295 ranged from 12.28–15.13 × 10−3 nm−1 during the spring flush, 14.60–18.00 × 10−3 nm−1 during the summer-autumn period, increasing to 18.04–19.88 × 10−3 nm−1 in the winter. It is apparent that for S275–295, as for all S ranges, the shallowest values occur at the height of the spring flush (Figures 1d1e) in conjunction with highest SUVA254 and lowest a250:a365 (Figures 1b1c) indicative of increased vascular plant, high molecular weight aromatic carbon sources [Neff et al., 2006; Spencer et al., 2008]. Thus during the spring, leaching of surface litter and soil layers exports DOC that has undergone little microbial processing during transport and is thus younger in age [Neff et al., 2006; Raymond et al., 2007] and more labile [Holmes et al., 2008] than later in the year. After the spring flush, S values increase (Figures 1d1e) in combination with declining SUVA254 and increasing a250:a365 values (Figures 1b1c), thus highlighting increased residence time of DOC in contact with subsurface microbial communities [Striegl et al., 2005]. However, also during the summer-autumn period, autumn rainfall can be observed with elevated Q corresponding to a decrease in S values (Figures 1d1e), an increase in SUVA254 and a decrease in a250:a365 (Figures 1b1c) as again DOC contact time with microbial communities decreases in surface litter and soil layers.

[11] Although the three ranges of S show the same seasonal trends the range of variation between slopes was not consistent. This becomes apparent when examining SR (S275–295:S350–400) values. During the spring flush the mean SR was 0.82 (0.79–0.86), increasing during the summer-autumn to a mean value of 0.92 (0.80–0.98) and increasing to a mean winter value of 0.93 (0.91–0.94) (Figure 1f). Lowest SR values are associated with the spring flush and then increase with decreasing Q further indicating the decreased molecular weight of the source material and its degradation in subsurface microbial communities throughout the summer-autumn and winter. Clearly, the optical proxies described here show potential for examining DOM composition and reactivity in Arctic rivers and thus the fundamental role this material plays in ecosystem biogeochemistry.

[12] In addition to monitoring shifts in the quality of DOM due to climate change and associated permafrost degradation, it is also imperative to monitor shifts in the export and timing of the land-ocean flux of terrigenous DOC [Peterson et al., 2002; Guo et al., 2007]. With this goal in mind we examined CDOM relationships to DOC and lignin phenols as previous studies have shown the potential for utilizing CDOM measurements to derive the export of DOC and lignin phenols [Baker et al., 2008; Hernes et al., 2008]. YRP CDOM absorption was found to be strongly correlated to DOC concentration (r2 = 0.95–0.98; Figure 2). Using the equations of the linear regression lines, estimated DOC concentrations were subsequently calculated from measured CDOM absorption values at a254, a350, and a440 to estimate CDOM derived DOC (DOC254, DOC350 and DOC440) at the different wavelengths. Additionally, the linear relationship established between lignin phenols and a350 for the YR Basin (r2 = 0.97) [Spencer et al., 2008] was utilized to calculate CDOM derived lignin phenols (lignin350).

Figure 2.

The relationship at YRP between CDOM and DOC (a254, black circles, black line; a350, grey circles, grey line, and a440, white circles, black dashed line).

[13] Daily riverine DOC, DOC254, DOC350, DOC440 and lignin350 loads were estimated using the LOADEST program [Runkel et al., 2004] for water years 2004–2005 for YRP (Table 1 and auxiliary material). The model was validated with independent DOC and CDOM data from water year 2008 (n = 14) and the mean CDOM-derived DOC loads were within <2% of the measured DOC load. The mean annual Q for 2004–2005 (208.5 km3 yr−1) at YRP was close to the 25 year annual average and the mean annual DOC flux was 1.75 × 109 kg yr−1 with 61.2% of total DOC export in the spring flush period (Table 1). The intensive sampling undertaken in this study over the spring 2004 and 2005 hydrographs more tightly constrained earlier DOC flux estimates reported for the same periods by Striegl et al. [2005]. Mean CDOM-derived DOC loads for 2004–2005 were within ∼3% of the measured DOC load and were shown to estimate the same proportion of DOC within the spring flush within <3% of the measured DOC load. CDOM-derived calculated lignin phenol loads result in a mean flux at YRP for 2004–2005 of 5.27 × 106 kg yr−1 and an estimate of 86.8 % of total lignin export occurring during the spring flush (Table 1). Carbon-normalized yields of lignin phenols at YRP were observed to increase dramatically during the spring flush period [Spencer et al., 2008], thus explaining the greater proportion of lignin export during the spring flush in comparison to DOC. This highlights the critical need to intensively sample across the spring flush maxima in these systems to accurately capture the dynamic nature of different DOM moieties and to accurately develop flux estimates to the Arctic Ocean particularly for biomarkers such as lignin that can then be used to trace terrigenous organic carbon in the Arctic Ocean [Opsahl et al., 1999; Amon et al., 2003].

Table 1. Annual Water, DOC, CDOM Derived DOC (DOC254, DOC350, DOC440) and Lignin (lignin350) Fluxes and Percent Export in the Spring Flush at YRP
YearAnnual Q (km3 yr−1)DOC (109 kg yr−1)DOC254 (109 kg yr−1)DOC350 (109 kg yr−1)DOC440 (109 kg yr−1)Lignin350 (106 kg yr−1)% Spring Flush DOC Flux% Spring Flush DOC254 Flux% Spring Flush DOC350 Flux% Spring Flush DOC440 Flux% Spring Flush Lignin350 Flux

[14] Arctic rivers have all shown similar trends between discharge and DOC export [Raymond et al., 2007]. Thus, utilizing the calculated mean 2004–2005 loads for DOC at YRP it is possible to scale up and derive an estimate of total riverine export of DOC to the Arctic Ocean. Aagaard and Carmack [1989] estimate total Arctic riverine discharge to be 3,300 km3 yr−1, and thus using the mean 2004–2005 YRP data as a basis for all Arctic riverine discharge, this results in a total load for all Arctic rivers of 27.7 Tg DOC yr−1. This value is higher than previous estimates (18–26 Tg DOC yr−1) [Opsahl et al., 1999; Dittmar and Kattner, 2003] and is likely due to more representative sampling in this study, e.g. sampling at the river mouth instead of adjacent coastal waters, and making measurements during the notoriously under-sampled spring flush period when a high percentage of the DOC export occurs [Raymond et al., 2007]. However, even our calculation of the DOC load for Arctic rivers may be an underestimate. Recent estimates of Arctic freshwater inputs indicate higher values than Aagaard and Carmack [1989]. Using the Syed et al. [2007] estimate of 3,588 ± 257 km3 yr−1 we calculate a load of 30.1 Tg DOC yr−1 (27.9–32.2 Tg DOC yr−1). Even greater water discharge to the Arctic Ocean, up to 4,816 km3 yr−1 was estimated by Lammers et al. [2001] (derived by applying region-specific annual water yields to 95% of the 22.4 × 106 km2 pan-Arctic drainage basin and average annual runoff of 212 mm yr−1 to the remainder) resulting in a load of 40.4 Tg DOC yr−1.

[15] The mean estimated lignin load from 2004–2005 at YRP was also scaled up to estimate total Arctic riverine lignin loads. Using a total Arctic freshwater input value of 3,300 km3 yr−1 [Aagaard and Carmack, 1989] results in a lignin load of 83.4 kg6 yr−1, with an estimated residence time of terrigenous DOC in the Arctic Ocean (calculated after Opsahl et al. [1999]) ranging from 1–5 years. This is within the range of a 1–6 year residence time estimated for terrigenous DOC in the Arctic Ocean by previous studies [Opsahl et al., 1999; Lobbes et al., 2000]. Using the higher freshwater discharge values of 3,588 ± 257 km3 yr−1 [Syed et al., 2007] and 4,816 km3 yr−1 [Lammers et al., 2001], increases the calculated lignin fluxes to 90.7 kg6 yr−1 (84.2–97.2 kg6 yr−1) and 121.7 kg6 yr−1, respectively. Terrigenous DOC transits through the Arctic Ocean even faster if these higher loads are more representative of current Arctic riverine export (residence times of 1–4equation image years and <1–3equation image years, respectively). As with any calculation of residence time this highlights the importance of accurately constraining fluxes and reservoirs. In this case, it is clear that higher frequency sampling of Arctic rivers is necessary to more accurately measure fluxes of dissolved constituents like DOC and lignin. Future predictions of the impact of climate change on DOC export from Arctic peatlands will require a better understanding of the fate of terrigenous DOC in the Arctic Ocean, whether via degradation [Holmes et al., 2008] or transport to the North Atlantic surface or deep water [Opsahl et al., 1999; Amon et al., 2003; Hernes and Benner, 2006], and a well-constrained residence time can greatly assist in parsing out the relative importance of these two fates.

4. Conclusions

[16] There is clear potential to utilize optical measurements to characterize DOM and derive fluxes of DOC from Arctic rivers. Amon et al. [2003] have already shown uniform spectrophotometric properties from different Arctic rivers and detailed studies of the geochemical composition of DOM show it to be strikingly similar within Arctic rivers [Lobbes et al., 2000; Dittmar and Kattner, 2003; Spencer et al., 2008]. Optical measurements allow increased spatial and temporal resolution of DOM dynamics to be garnered, e.g. in situ monitoring [Spencer et al., 2007] and thus have the potential to improve our understanding of the role of DOM in the ecosystem and also to lead to more accurate flux estimates. Future studies could also investigate terrigenous DOC dynamics in the Arctic Ocean by using in situ optical instrumentation [Amon et al., 2003]. Not only is there obvious potential to utilize these measurements in Arctic rivers, there is also an apparent need to better understand the Arctic system. This is particularly true with respect to the residence time of DOC with subsurface microbial communities and the ramifications for DOC quality as land-ocean flux of DOC in these catchments alters due to climatic change [Peterson et al., 2002; Striegl et al., 2005; Guo et al., 2007].


[17] This study was supported by the USGS National Stream Quality Accounting Network ( and National Research Program ( Brand names are for identification purposes only and do not imply endorsement by the USGS.