Responses of near-surface permafrost and glacial ice to climate change are of particular significance for understanding long-term effects on global carbon cycling and carbon export by high-latitude northern rivers. Here we report Δ14C-dissolved organic carbon (DOC) values and dissolved organic matter optical data for the Yukon River, 15 tributaries of the Yukon River, glacial meltwater, and groundwater and soil water end-member sources draining to the Yukon River, with the goal of assessing mobilization of aged DOC within the watershed. Ancient DOC was associated with glacial meltwater and groundwater sources. In contrast, DOC from watersheds dominated by peat soils and underlain by permafrost was typically enriched in Δ14C indicating that degradation of ancient carbon stores is currently not occurring at large enough scales to quantitatively influence bulk DOC exports from those landscapes. On an annual basis, DOC exported was predominantly modern during the spring period throughout the Yukon River basin and became older through summer-fall and winter periods, suggesting that contributions of older DOC from soils, glacial meltwaters, and groundwater are significant during these months. Our data indicate that rapidly receding glaciers and increasing groundwater inputs will likely result in greater contributions of older DOC in the Yukon River and its tributaries in coming decades.
An important question in global carbon cycling is how climate change will alter the flux, chemical nature, reactivity, and fate of dissolved organic matter (DOM) released from northern high-latitude watersheds and transported in rivers to estuaries and coastal margins [Dittmar and Kattner, 2003; Benner et al., 2004; Striegl et al., 2005; Holmes et al., 2012]. Riverine DOM is known to play important roles in many ecological and geochemical processes associated with aquatic systems [Aiken, 2014], however, its fate in coastal oceans, especially with regard to carbon cycling, remains poorly understood [Fichot and Benner, 2014]. Arctic and sub-Arctic river basins yield disproportionately large amounts of water and terrigenous DOM to northern seas and the Arctic Ocean [Opsahl et al., 1999; Benner et al., 2004; Raymond et al., 2007] and, as northern high latitudes warm, the amount and chemical nature of DOM exported from these basins is expected to change with important ramifications for ecosystem biogeochemistry [Guo et al., 2007; Walvoord and Striegl, 2007; Frey and McClelland, 2009]. It is commonly assumed that as peat and soil organic matter are increasingly degraded, older and compositionally different DOM will be transported in waters draining these soils [Neff et al., 2006]. While the production and release of DOM in northern high-latitude and temperate systems is sensitive to climate variability [Freeman et al., 2001; Frey and Smith, 2005; Davidson and Janssens, 2006], quantitative understanding of these processes is limited, and, to date, little evidence for mobilization of permafrost-derived DOM has been observed in major rivers [Benner et al., 2004; Raymond et al., 2007; Striegl et al., 2007].
The 3340 km long Yukon River drains 854,700 km2 of northwest Canada and Alaska and discharges into the Bering Sea, ultimately contributing DOM to the Arctic Ocean. It is the longest free-flowing river in the world and, having little development, is relatively pristine [Nilsson et al., 2005]. The export of DOM by the Yukon River to the Bering Sea is strongly seasonally dependent [Striegl et al., 2007; Spencer et al., 2009]. As observed in other high-latitude northern rivers, the Yukon has maximum dissolved organic carbon (DOC) concentrations in conjunction with the period of maximum discharge during the spring flush, resulting in a significant proportion of the annual DOC export occurring during this relatively short time period [Raymond et al., 2007; Holmes et al., 2012; Spencer et al., 2009]. Also, DOC radiocarbon age (Δ14C) and composition are seasonally dependent and tied to discharge in high-latitude northern rivers. The oldest DOC is transported under ice in the winter, and the youngest DOC is exported during the spring flush when discharge is greatest, and there is a strong contribution of terrigenous organic matter [Neff et al., 2006; Raymond et al., 2007; Striegl et al., 2007; Spencer et al., 2008, 2009].
In combination with DOM chemistry data, information about Δ14C content is relevant for identifying sources of DOM and understanding watershed processes [Butman et al., 2012; Spencer et al., 2012b]. The majority of Δ14C and DOM export data available for major rivers has been reported for samples collected at the mouths of the rivers [e.g., Raymond and Bauer, 2001; Raymond et al., 2007; Butman et al., 2012; Spencer et al., 2012a; Hossler and Bauer, 2013]. However, more detailed analyses of watersheds are required to resolve uncertainties in understanding the drivers responsible for this export and to anticipate future changes in DOM composition and flux. Here we present Δ14C-DOC data obtained from the major tributaries and the main stem of the Yukon River. The Yukon River basin is diverse with respect to tributaries draining different source areas common to northern high latitudes (e.g., glaciers, mountainous uplands, forested lowlands, and peatlands). As such, it is well suited to address the variability of contributions from these sources on the age and composition of DOM. Our goals, therefore, were to assess the age and chemical quality of current contributions of DOM from tributaries to the main stem of the Yukon River, to improve understanding of how future changes in climate might impact the nature of DOM exported from these and other similar high-latitude watersheds draining to the Arctic Ocean, and to establish baseline data against which future changes may be assessed.
2 Materials and Methods
2.1 Sample Sites and Collection
The Yukon River and its tributaries have been the subject of intensive U.S. Geological Survey (USGS) investigations since 2001 [Schuster et al., 2011]. Water samples were collected between August 2004 and July 2006 for DOC concentration, DOC optical properties, and DOC isotopic composition (δ13C and Δ14C) throughout the Yukon River basin at different locations on the main stem of the Yukon River (from near the headwaters to just above the head of tidal influence at Pilot Station), from a range of tributaries incorporating groundwater-dominated clearwaters, peat-dominated blackwaters, and glacial meltwaters (Table 1 and Figure 1). Additional samples were collected through September 2008 for DOC concentration and DOC optical property analyses. Samples were filtered in the field (0.45 µm) and shipped on ice to the USGS laboratory in Boulder, Colorado for DOC, UV-visible, and fluorescence analyses, and to Yale University for isotope analyses. For a subset of samples, particulate organic carbon (POC) samples were obtained by filtering 300 mL of water through prebaked glass fiber filters (0.7 µm). Filters were frozen and shipped to Yale University.
Table 1. DOC Concentration, Specific Ultraviolet Absorbance (SUVA254), δ13C-DOC, and Δ14C-DOC Measurements for Samples Collected in the Yukon River and its Tributariesa
DOC measurements were performed on an OI Analytical Model 700 total organic carbon analyzer [Aiken, 1992]. UV-visible absorbance was measured at room temperature on a Hewlett-Packard photodiode array spectrophotometer (model 8453) between 200 and 800 nm. Absorbance data are expressed as decadal (linear) absorption coefficients determined at λ = 254 nm (a254), in units of cm−1 and were determined by dividing the absorbance (A(254)) by the cell path length (l) in centimeters [Braslavsky, 2007]. Specific UV absorbance (SUVA254) values, a measure of DOC aromaticity, were determined by dividing the UV absorbance measured at λ = 254 nm by the DOC concentration [Weishaar et al., 2003]. Fluorescence data were obtained on whole water samples diluted to an A(254) of < 0.2 absorbance units (AU) using a Horiba-JY Fluoromax-3 spectrofluorometer with DataMax software. Fluorescence excitation emission matrices (EEMs) were measured at room temperature on aqueous samples by measuring fluorescence intensity across excitation wavelengths ranging from 240 to 450 nm (5 nm intervals) and emission wavelengths ranging from 300 to 600 nm (2 nm intervals). Excitation and emission slit widths were 5 nm, and the instrument was configured to collect fluorescence scans in ratio mode. Both fluorescence EEMs and absorbance scans were blank-corrected with Milli-Q water, and EEMs were Raman normalized and corrected for any inner-filter effects [McKnight et al., 2001; Murphy et al., 2010].
2.3 Organic Carbon Isotopic Analyses
Carbon isotopic analyses were performed using methods described previously [Raymond and Bauer, 2001]. Briefly, river water (120 mL) was placed into a clean quartz tube and acidified with 0.2 mL of ultrahigh purity (UHP) 40% phosphoric acid. The samples were then sparged with UHP nitrogen to remove any inorganic carbon. Pure UHP O2 was subsequently sparged through the system to provide an oxidant for the UV oxidation of DOC. The sample was then oxidized with UV. The resulting CO2 was transferred to a vacuum line and cryogenically purified. For POC samples, filters were acidified overnight with 1% concentrated sulfurous acid to remove carbonates, thoroughly dried, and the POC oxidized to CO2 by dry combustion with CuO and Ag metal at 850°C in 9 mm quartz tubes [Raymond and Bauer, 2001]. For both DOC and POC samples, purified CO2 gas samples were converted to graphite targets by reducing CO2 with an iron catalyst under 1 atm H2 at 550°C, and the targets were subsequently analyzed for carbon isotopes (δ13C in ‰ and 14C as fraction modern carbon) at the National Ocean Sciences Accelerator Mass Spectrometry (NOSAMS) facility at the Woods Hole Oceanographic Institution, or the University of Arizona's AMS facility. Δ14C data (in ‰) were corrected for isotopic fraction using measured δ13C values; for those cases where a δ13C measurement was unavailable, a value of −25‰ was used. Δ14C and radiocarbon age were determined from percent modern carbon using the year of sample analysis according to Stuiver and Polach . Ages are presented as Modern when the fraction modern exceeded 1.
3 Results and Discussion
3.1 Age and Composition of Dissolved Organic Matter in the Yukon River Basin
The most depleted Δ14C-DOC values (i.e., oldest DOC radiocarbon ages) in the Yukon River basin were measured in headwater streams and tributaries strongly influenced by glacial meltwater and groundwater discharge (Table 1). In the headwaters of the Yukon River, old DOC was measured in Atlin River (Δ14C = −453‰), Nares River (Δ14C = −223‰), and in the main stem of the Yukon River at Whitehorse (Δ14C = −203‰) (Table 1). These rivers drain water predominantly from glaciers and ice fields in the Yukon Territory and British Columbia and also have a substantial amount of groundwater contribution to flow (31–38% of annual discharge) [Walvoord and Striegl, 2007]. Ancient DOC (age range 850–3235 yBP) was also measured in rivers receiving groundwater that originates from glacial and snowpack meltwater in the Alaska Range, including the Clearwater River (Δ14C = −340‰), the largest spring-fed tributary of the Tanana River, and the Tanana River (Δ14C = −112 to −312‰). DOC in meltwaters collected directly from Gulkana Glacier (Alaska Range; (Δ14C = −335 to −650‰; Table 1), which drains to the Yukon River via the Tanana River, and Llewellyn Glacier (Juneau Icefield, British Columbia) meltwaters in Atlin Lake (Δ14C = −677 to −718‰; Table 1) were the oldest samples found in our study, ranging in age from about 3200–10,000 yBP. The Δ14C-DOC values for these glacier-derived meltwaters are comparable to other reported values for a range of watersheds in southeast Alaska having varying contributions of glacier meltwater to downstream stream water [Hood et al., 2009; Stubbins et al., 2012]. In the most heavily glaciated watershed examined by Hood et al.  (Sheridan; 64% glacier cover) the Δ14C-DOC value was −386‰, which is similar to that observed in glaciated catchments and waters in this study within the Yukon River basin (Table 1).
The concentrations of DOC in these glacially dominated samples were low, (typically less than 1.2 mg C L−1), and DOM optical property data show little terrigenous signature, as has been noted for other glacial systems [Sharp et al., 1999; Hodson et al., 2008; Hood and Scott, 2008]. Fluorescence EEMs (Figure 2) indicate the predominance of a fluorophore (excitation270/emission305) commonly associated with protein-like material (tyrosine) and simple phenols, and a marked absence of humic and fulvic-like fluorophores associated with vascular plant and upper soil horizon sources of DOM [Maie et al., 2007; Hernes et al., 2009; Fellman et al., 2010a] (e.g., excitation λ 260 nm; emission λ 448–480 nm and excitation λ 320–360 nm; emission λ 420–460 nm). This observation is further supported by low SUVA254 values (in general < 2.0 L mg C−1 m−1) in the glacially dominated samples [Weishaar et al., 2003]. Our results are consistent with those reported for other glacial systems from a range of locations [LaFreniere and Sharp, 2004; Barker et al., 2006; Hood et al., 2009; Stubbins et al., 2012; Singer et al., 2012].
Given that ice residence times for glaciers in the Juneau Icefield are estimated to be approximately 300 years (e.g., Mendenhall Glacier, [Stubbins et al., 2012]), it is unlikely that simple storage effects of DOC within the glaciers account for the age of the DOC (3200–10,000 yBP) reported here. DOM associated with precipitation has been shown to be of sufficient concentrations and to contain a number of atmospherically transported, anthropogenic compounds depleted in 14C to strongly influence the signals we are observing in our glacier samples [Raymond, 2005; Jurado et al., 2008; Yan and Kim, 2012; Mead et al., 2013]. In particular, glaciers are known to accumulate anthropogenically derived compounds and act as sources of these compounds to systems receiving glacial meltwaters [Blais et al., 2001; Jenk et al., 2006; Grannas et al., 2006; Stubbins et al., 2012]. Recent analyses of glacial DOM associated with the coastal, maritime glaciers of the Juneau Icefield and streams influenced directly by glacial meltwaters suggest that anthropogenic aerosols derived from fossil fuel burning may be important sources of ancient DOM in our samples [Stubbins et al., 2012].
Evidence that glacial meltwaters directly discharge into some rivers and streams in the Yukon River basin are to be found in the river discharge data. Water discharge is highly seasonal throughout the Yukon River basin. Discharge for rivers directly influenced by glacial meltwaters (e.g., Tanana River, White River, and Atlin River) typically peak during the summer (July), whereas discharges for nonglaciated rivers (e.g., Yukon River and Porcupine River) peak in May–June [Striegl et al., 2007]. The situation for DOM delivered in ground water (e.g., Tanana River and Clearwater River winter samples) is more complicated since this pool of organic matter can also be influenced by overlying soils and geologic materials such as formations containing shales and coal, which may also supply ancient carbon to the DOC pool. Groundwater-bearing deposits in the Tanana River basin, however, are primarily thick silt, sand, and gravel deposits [Pewe and Reger, 1983]. While the influence of geologic materials is not likely in this scenario, they cannot be discounted at this time.
In contrast, blackwater rivers draining low-lying areas influenced by permafrost and containing organic matter rich soils (e.g., Hess Creek, Black River, and Fortymile River; Table 1) all contained modern DOC with Δ14C ranging from −7 to 109‰ (Table 1). DOC concentrations for these samples were much greater (6.8–32.0 mg C L−1) than those in rivers having large percentages of water from glacial and groundwater sources and contained substantially more aromatic DOM as evidenced by high SUVA254 values (2.7–3.8 L mg C−1 m1; Table 1) and strong fluorescence signals associated with terrestrially derived organic matter (i.e., EEMs dominated by humic and fulvic-like fluorophores) (Figure 2). The fluorescence data and the high SUVA254 values associated with these samples indicate the presence of relatively unaltered organic matter derived from leaf litter and upper soil horizons [Wickland et al., 2007; Fellman et al., 2010a]. DOC associated with these rivers also contains elevated carbon-normalized yields of lignin phenols, a biomarker for vascular plant sources [Spencer et al., 2008].
Older DOC has been reported for rivers draining regions containing managed agricultural soils and degraded peatlands supporting the hypothesis that the age of DOC can reflect the relative degree of oxidation of older soil organic matter [Evans et al., 2007; Kalbitz and Geyer, 2002; Moore et al., 2013]. Therefore, our results indicate that older DOM from peat or soils in permafrost-rich landscapes is not being mobilized and transported from blackwater catchments within the Yukon River basin at this time in sufficient quantities to be reflected strongly in the 14C-DOC signature. This observation is perhaps surprising given that degradation of permafrost soils is known to be occurring across northern latitudes. For instance, Vonk et al.  report very old (>21,000 years) highly biolabile DOC in thaw streams draining Siberian yedoma deposits, although the influence of these small streams on the 14C-DOC signal in the Kolyma River, a major river, was not explored. However, as pointed out by Vonk et al. , DOC from these small meltwater streams is subject to both high in-stream turnover and dilution by substantially greater flows in the Kolyma River, which has an average discharge of 136 km3/year compared to 208 km3/year for the Yukon River [Holmes et al., 2012]. Each of these processes acts to mask the presence of DOC from Yedoma sources when considering bulk 14C-DOC data.
While our blackwater DOC samples were modern, POC ages for the samples were relatively old: Forty Mile River (2213 yBP), Hess Creek (2861 yBP), Black River (1576 and 2585 yBP), and Porcupine River (1800, 3849, and 4728 yBP). These values are similar to those reported for deeper organic horizons and mineral soils in the active layers of soil profiles obtained in the Hess Creek drainage [O'Donnell et al., 2011] suggesting that deeper soil horizons are being physically eroded; however, the link between these horizons and the export of DOC from these soils is not apparent from the river samples. Organic matter below the active layer reported by O'Donnell et al.  was much older (Δ14C = −800‰; approximate age of 12,500 years).
Samples from rivers draining nonglaciated mountainous regions (e.g., Big Salmon, Pelly, and Stewart) were younger in age than samples from glaciated regions, although older than modern, with Δ14C-DOC ranging from −12 to −129‰ (Table 1). These samples, which were collected during the late summer periods of 2004 and 2005, typically had greater DOC concentrations and greater SUVA254 values than the samples from glaciated and groundwater dominated rivers, but lower than those measured in blackwater rivers (Table 1). In the late summer period, the influence of organic matter derived from organic rich upper soil horizons is much less than during the spring flush, resulting in DOM that is less aromatic, contains relatively smaller amounts of high molecular weight aromatic moieties and exhibits lower lignin carbon-normalized yields than during the spring flush period [Striegl et al., 2005, 2007; Spencer et al., 2008]. The ages of these samples were similar to the age of a sample of soil-ice meltwater collected below the organic layer at a site near Fort Yukon (Δ14C-DOC = −83‰; 593 yBP). This soil-ice meltwater sample also had a SUVA254 value (2.5 L mg C−1 m−1) similar to that in the Yukon River at that time (2.6 L mg C−1 m−1) and is representative of OM leaching from deeper soil horizons (i.e., different DOM source pools, hydrologic flow paths, and microbial mineralization rates of organic matter). We previously showed that samples in the Yukon River basin are depleted in lignin phenol concentrations and have lower lignin carbon-normalized yields under base flow conditions or with increased contributions from glacial meltwater or groundwater [Spencer et al., 2008, 2009]. Samples are enriched during the spring flush period when inputs from leached plant materials and upper soil horizons dominate the DOM pool consistent with our findings here.
3.2 Relationships Between Dissolved Organic Matter Composition and Age
For the samples in this study, the relationship between DOC concentration and the absorption coefficient at λ = 254 nm (a254) nm was linear (R2 = 0.9927), similar to results published previously for a different subset of samples in this system [Spencer et al., 2009]. There is also a strong correlation between DOC concentration and Δ14C-DOC (Figure 3a; r = 0.86, p < 0.001), and between a254 and Δ14C-DOC (Figure 3b; r = 0.83, p < 0.001. These relationships are nonlinear because Δ14C is an intensive property with an upper limit imposed by the modern carbon signal. Almost all samples with a DOC concentration > 5 mg C L−1 contain modern carbon whereas those samples with DOC concentration < 5 mg C L−1 are depleted with respect to Δ14C-DOC (Figure 3a); DOC is predominantly modern above a threshold value for a254 of 0.2 cm−1 (Figure 3b).
There is a broad relationship between SUVA254 and Δ14C-DOC (Figure 4). SUVA254 reflects the contribution of aromatic moieties to the optical signature of DOM [Butman et al., 2012; Spencer et al., 2012a; Weishaar et al., 2003]. Figure 4 reinforces the aforementioned influence of organic matter derived from leaf litter and upper soil horizons on the age of DOC in these rivers. The trend is similar to that observed in the Kolyma River in Siberia [Neff et al., 2006] for a much smaller range of SUVA254 (2.2 to 4.6 L mg C−1 m−1) and Δ14C-DOC (−88 to +150‰) values. For each river, modern, high SUVA254 DOM was associated with elevated flow periods, such as during the spring freshet. DOM becomes both less aromatic and older with declining discharge as the balance between different DOM sources shifts to favor deeper soil horizons and groundwater and, for the Yukon River system, glacial meltwaters.
The trend observed for samples throughout the Yukon River basin is also generally similar to SUVA254 and Δ14C-DOC data obtained for samples collected across the hydrograph at the mouths of 15 large temperate North American rivers (Figure 4) [Butman et al., 2012]. These data are presented here to illustrate some common patterns regarding DOC age and optical characteristics in riverine systems. Notably, modern DOM and high-average SUVA254 values are characteristic of systems with high productivity and precipitation (e.g., Atchafalaya River) underscoring the importance of wetlands, plant litter, and surface soils in the mobilization of modern, aromatic DOM. In contrast, rivers influenced substantially by groundwater contained older DOM with low-average SUVA254 values (e.g., Colorado River). It is important to note that within basin trends between SUVA254 and Δ14C-DOC for these rivers were highly variable owing to substantial differences in hydrology, watershed dynamics, sources of DOM, and anthropogenic influences [Butman et al., 2012]. Therefore, caution is warranted in generalizing the use of DOM optical data to infer DOC age for systems that have not been adequately characterized. For example, the St. Lawrence River, sampled approximately 115 km downriver of Lake Ontario, consistently contained low SUVA254, modern DOC due to the autochthonous production of DOM within the Great Lakes.
We previously established strong relationships between UV-visible absorption coefficients and lignin phenol concentrations and subsequently used these relationships to refine the export of DOC and lignin phenols from the Yukon River basin [Spencer et al., 2008, 2009]. Optical measurements such as absorbance, that require small sample volumes, are comparatively inexpensive, and data are relatively abundant [Spencer et al., 2012b] compared to Δ14C-DOC data. Optical properties may also be measured in situ at high temporal resolution [Spencer et al., 2007; Bass et al., 2011]. Relationships between DOC absorption coefficients and lignin phenols, DOC concentration, or Δ14C-DOC are system dependent [Spencer et al., 2012a] and potentially subject to change as dynamics within river basins change. Here we utilize the relationship between a254 and Δ14C-DOC to infer information about the 14C content of DOC in the present-day Yukon River and its tributaries from the easily obtained and inexpensive a254 data with clear potential for developing a high temporal resolution proxy in the future.
The relative influences of glacial versus blackwater tributaries on the age of DOC can be seen in a comparison of a254 data for the Tanana and Porcupine Rivers. These rivers are the two largest tributaries to the Yukon River with similar drainage areas in terms of size but with different catchment and DOC export characteristics [Striegl et al., 2007]. DOC concentrations and absorption coefficients for both rivers are heavily influenced by hydrologic controls [Striegl et al., 2007; Spencer et al., 2009; Figure 5], however, a254 data indicate that for most of the year, much older DOC is transported by the glacially influenced Tanana River than by the nonglaciated, blackwater influenced Porcupine River (Figure 5a). With the exception of the low-flow winter period (presumably ground water), a254 for the Porcupine River either exceeded or approximated 0.2 cm−1 suggesting the export of young DOC for most of the hydrograph, whereas a254 was much less for the Tanana River for most of the year (Figure 5a). After peaking during spring snowmelt, a254 steadily declines during the summer-fall period in the Porcupine River. This may be due, in part, to the influence of older C stocks from thawing permafrost soils or it may represent the combination of decreased flows directly off the surface landscape coupled with increasing influence of groundwater DOM as the hydrograph falls.
Absorption coefficient data also indicate that DOC is more depleted in Δ14C-DOC in the main stem of the Yukon River at upper river locations near glacial sources and becomes progressively enriched as tributaries draining blackwater catchments contribute DOM to the main stem of the Yukon (Figure 4b). For example, absorption coefficient data for the Yukon River at Carmacks, located 500 km downstream from the headwaters of the river, are much lower throughout the year compared to a254 data measured at Eagle (1057 km from source) (Figure 5b). The relative influences of different water sources on DOC characteristics within this system are most apparent when comparing Δ14C-DOC data and distance from source for samples collected during the same time period in the upper Yukon region (Figure 6). The most depleted samples (e.g., Atlin River and Nares River) strongly influence the upper Yukon River (e.g., Yukon River at Whitehorse, Canada; Location 6, Figure 1). The Yukon main stem becomes more enriched with respect to Δ14C-DOC due to DOC additions from nonglaciated rivers (e.g., Pelly, Big Salmon, and especially the Stewart River) moving downstream from the headwaters. The Fortymile River at 975 km is the first significant blackwater tributary in the Yukon River basin and has modern DOC. Beyond this point, blackwater tributaries exert a strong influence on DOM composition in the Yukon River.
3.3 Export of 14C-DOC to the Ocean
The oldest DOC reported near the mouth of the Yukon River at Pilot Station was measured during the winter months when low flow is dominated by groundwater (Table 1). Compositional changes in the DOM associated with these time periods are apparent in SUVA254 and fluorescence EEM data (Table 1 and Figures 2d–2f). Of significance is the presence of fluorophores associated with vascular plants and soil organic matter in addition to fluorophores indicative of microbial sources in the Yukon River at Pilot Station winter sample, suggesting that contributions of waters from nonglaciated catchments influence the DOC in the Yukon River even in winter when there is no surface water runoff. These results are consistent with those reported by O'Donnell et al.  who observed considerable variability in DOM composition under base flow conditions across 60 rivers and streams within the Yukon River basin. Differences among systems were attributable to the relative contributions of suprapermafrost and subpermafrost aquifers to individual rivers.
Based on 5 year averages (2001–2005) of DOC exported by the Yukon River at Pilot Station, 13% was estimated to occur in the winter (31 October to 30 April), 52% in spring (1 May to 30 June), and 35% in the summer-fall (1 July to 31 October) [Striegl et al., 2007]. Average a254 data for these periods based on 11 years of data (2001–2011) are 0.07 ± 0.02 cm−1 (n = 37) in winter, 0.40 ± 0.13 cm−1 (n = 132) in spring, and 0.16 ± 0.052 cm−1 (n = 60) in summer-fall periods. Applying the relationship between a254 and Δ14C-DOC (Figure 3b) for samples from only Yukon River at Pilot Station to these average absorbance values permits a tightening of estimates of seasonal Δ14C-DOC export that have previously been based on only a few data points. This approach yields estimated values for average Δ14C of −89‰ (winter), 55‰ (spring), and 7‰ (summer-fall). The export of older DOC, especially during the summer-fall period, may indicate increased metabolism of soil-derived organic matter, increased inputs from permafrost thaw at the time of year when active layers are deepest, and/or the influence of greater glacial contributions in summer when glacial melting is at its annual peak.
3.4 Implications of DOC Export to the Ocean From Glacial-Dissolved Organic Matter
Whereas the seasonal DOC export trend is similar to those for other major Arctic rivers (e.g., Lena, Ob, Yenisey, Mackenzie, and Kolyma), DOC transported by the Yukon River is, on average, more depleted in Δ14C than samples from the other major Arctic rivers [Raymond et al., 2007]. A possible reason for this difference is that the Yukon River is more heavily influenced by glacial meltwater than the other major Arctic rivers. The watershed of the Yukon River has an estimated glacial surface area of 10,500 km2 which is substantially greater than that of the major Siberian Arctic Rivers (e.g., Lena River 18.8 km2, Ob' River 1750 km2, and Yenisey River 50.7 km2) [Dyurgerov and Carter, 2004]. In recent decades a number of the glacial tributaries in the Yukon River basin have exhibited increases in annual discharge due to increased glacier melt [Brabets and Walvoord, 2009]. In addition, the Yukon River flows east to west through the discontinuous permafrost zone, whereas the other major Arctic rivers flow predominantly south to north. A consequence of the orientation of the Yukon River is that permafrost underlies much of the Yukon River basin [Striegl et al., 2007], and it is particularly vulnerable to permafrost degradation. The Yukon River and its tributaries, therefore, represent a major sentinel river system for observing the effects of climatic change and any associated increased glacier melt and permafrost degradation signature in North America.
Our work identifies glacially derived organic matter as an important source of old DOC in those catchments originating in ice and snowfields in the south central and southeastern portions of the Yukon River basin. Globally, DOC currently frozen in glacial ice is a pool of organic matter that represents a potential source of relict carbon if mobilized by rapid glacial melting. Glacier-derived DOC has been shown to be highly biolabile and to represent an important source of reduced carbon to downstream freshwater and marine ecosystems [Hood et al., 2009; Fellman et al., 2010b]. In addition, Dyurgerov and Carter  demonstrated that freshwater contributions from pan-Arctic glaciers exceeded those of rivers flowing into the Arctic Ocean, including the Yukon, for the 1961–1998 period. Presently, little is known about the chemistry or flux of DOM associated with these glacially derived waters, although the presence of compounds from both microbial and fossil fuel sources have been reported for glacial waters associated with the Juneau Icefield [Stubbins et al., 2012].
All catchments within the Yukon River basin are subject to changes in DOC export related to climate effects. Presently, drainages such as the Tanana are experiencing melting of Alpine glaciers and perennial snow fields [Arendt et al., 2002], whereas drainages dominated by permafrost and organic-rich soils are experiencing permafrost degradation, the formation of thermokarst features, and drying of areas where thermokarst lakes have drained [Yoshikawa and Hinzman, 2003; Riordan et al., 2006; Jorgenson et al., 2001, 2006]. Continued glacier melting and permafrost thaw within different catchments could profoundly influence DOM chemistry and DOC export within the Yukon River basin by changing regional hydrology, soil productivity, and microbial activity. While Alaska's glaciers continue to melt at a rapid rate, there are uncertainties about long-term permafrost thaw and its influence on carbon cycling and DOC export [Arendt et al., 2002; Froese et al., 2008; Grosse et al., 2011]. It is anticipated that, as permafrost thaws, the groundwater contribution to overall river flow throughout the Yukon River basin will increase substantially [Walvoord et al., 2012]. This change in hydrology is projected to result in decreased DOC concentrations and decreased SUVA254 values [O'Donnell et al., 2010; O'Donnell et al., 2012]. Our results suggest that these changes will also result in more Δ14C-depleted DOC in the river.
For large Arctic rivers, such as the Yukon, the ability to observe and quantify changes in DOC composition and export resulting from climate change is complicated by the diversity of subwatersheds, large disparity in discharge between the river main stem and impacted tributaries, and differences in the rates of processes driving carbon cycling within subwatersheds. As a result, it is difficult to link changes in soil dynamics to DOC export when sampling only at the mouths of large rivers. As our data indicate, however, drainages within the Yukon River basin exhibit different DOM characteristics due to different geologic, hydrologic, and permafrost influences. A fruitful strategy for assessing climate-based influences in the Yukon River basin, therefore, is the identification and monitoring of smaller sentinel tributaries especially sensitive to the melting of glaciers, changes in watershed hydrology resulting from permafrost thaw, and/or the mobilization of organic matter in permafrost soils currently below the active layer.
This study was supported by the United States Geological Survey National Stream Quality Accounting Network (http://water.usgs.gov/nasqan) and the USGS National Research Program (http://water.usgs.gov/nrp). R.G.M.S acknowledges support from NSF grants DEB-1145932, OPP-1107774, and ANT-1203885. We wish to thank Kenna Butler (USGS-Boulder) for analytical assistance. We also thank Jonathan O'Donnell of the U.S. National Park Service and two anonymous reviewers for their critical reviews of the manuscript. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.