Northern high-latitude rivers transport large amounts of terrestrially derived dissolved organic matter (DOM) from boreal and arctic ecosystems to coastal areas and oceans. Current knowledge of the biodegradability of DOM in these rivers is limited, particularly for large rivers discharging to the Arctic Ocean. We conducted a seasonally comprehensive study of biodegradable dissolved organic carbon (BDOC) dynamics in the Yukon River and two of its tributaries in Alaska, USA. Distinct seasonal patterns of BDOC, consistent across a wide range of watershed size, indicate BDOC is transported year-round. Relative biodegradability (%BDOC) was greatest during winter, and decreased into spring and summer. Due to large seasonal differences in DOC concentration, the greatest concentrations of BDOC (mg C L−1) occurred during spring freshet, followed by winter and summer. While chemical composition of DOM was an important driver of BDOC, the overriding control of BDOC was mineral nutrient availability due to wide shifts in carbon (C) and nitrogen (N) stoichiometry across seasons. We calculated seasonal and annual loads of BDOC exported by the Yukon River by applying measured BDOC concentrations to daily water discharge values, and also by applying an empirical correlation between %BDOC and the ratio of DOC to dissolved inorganic N (DIN) to total DOC loads. The Yukon River exports ∼0.2 Tg C yr−1 as BDOC that is decomposable within 28 days. This corresponds to 12–18% of the total annual DOC export. Furthermore, we calculate that the six largest arctic rivers, including the Yukon River, collectively export ∼2.3 Tg C yr−1 as BDOC to the Arctic Ocean.
 The northern permafrost zone includes circumpolar arctic, subarctic, and boreal regions, and contains an estimated 1670 Petagrams (Pg) of organic carbon (OC) in perennially frozen (permafrost) and seasonally thawed soils [Schuur et al., 2008; Tarnocai, 2009]; more than twice the entire atmospheric C pool [Schuur et al., 2008]. Recent increases in surface air temperatures of 0.35°C per decade over terrestrial regions between 50°N and 70°N [Serreze and Francis, 2006] are broadly impacting physical and ecological systems of this region, and permafrost thaw may lead to wide-scale mobilization of C to the atmosphere, and to aquatic systems as gaseous, dissolved, and particulate C [McGuire et al., 2009; Grosse et al., 2011]. Mobilization of even a fraction of this huge OC pool may significantly impact C cycling in northern freshwater aquatic systems and the Arctic Ocean, which receives more than 10% of global riverine discharge and dissolved organic carbon (DOC) loads [Opsahl et al., 1999].
 Chemical composition is a central control on potential DOC biodegradability in terrestrial and aquatic systems [Kalbitz et al., 2003a; del Giorgio and Davis, 2003]. Current understanding of DOM chemical composition in high-latitude rivers suggests that biodegradable DOC (BDOC) is exported by rivers to coastal regions of the Arctic Ocean, although very little is known about the amount and variability of riverine BDOC exports. Two recent studies of arctic rivers indicate that spring freshet DOC is highly biodegradable, and less so in early summer [Holmes et al., 2008; Mann et al., 2012]. BDOC was correlated with DOM chemical composition [Mann et al., 2012], and increased with inorganic nutrient additions [Holmes et al., 2008]. These findings challenge previous interpretations of high-latitude riverine DOM as mostly refractory [Dittmar and Kattner, 2003; Meon and Amon, 2004]. While it is important to know how much DOM is exported to the Arctic Ocean and how that may be altered by changing climate, it is crucial to understand DOM reactivity in terms of the potential impact on estuarine and marine productivity, and as a positive feedback to climate in the case of rapid mineralization to CO2.
 We conducted a multiyear study aimed at characterizing the seasonal and spatial variability of DOC biodegradability in three broadly different rivers of the Yukon River basin in Alaska, including the Yukon River above head of tide before it discharges to the Bering Sea. We assess BDOC across the entire hydrograph, including winter, and determine relations with DOM chemical composition and inorganic nitrogen concentrations, as well as temperature sensitivity of BDOC. We present seasonal and annual loads of BDOC exported by the Yukon River, and estimate BDOC loads exported by the five other major arctic rivers discharging to the Arctic Ocean. This is the first study of BDOC exports from northern high-latitude rivers that includes complete seasonal sampling and year-round calculations of fluxes.
2.1. Site Description
 The Yukon River is the largest river in Alaska (3340 km long; ∼200 km3water discharge per year), and the fifth largest river draining to the Arctic Ocean. The 854,700-km2 basin (R. G. Striegl, et al., Carbon dioxide and methane emissions from the Yukon River system, submitted to Global Biogeochem. Cycles, 2012). covers diverse terrain in northwest Canada and Alaska, including high mountainous regions, forested uplands, and wide expanses of lowlands rich in wetlands and lakes. The basin is predominantly subarctic or boreal, and continuous and discontinuous permafrost is present throughout [Brabets et al., 2000; Striegl et al., 2007]. We chose to study three rivers that cover a gradient in watershed size, terrain, and water discharge: the Yukon River, the Porcupine River, and Beaver Creek (Figure 1 and Table 1). The Yukon River at Pilot Station (Yukon) includes water discharge from most of the basin above tidal influence. The Porcupine River (Porcupine) is a major tributary to the Yukon (km 1450 from the Yukon headwaters) that drains 108,520 km2of the northeastern portion of the Yukon River basin. This region is underlain by continuous permafrost and dominated by lowland wetlands. Beaver Creek is a mid-size tributary that flows north to the Yukon and has a ∼7000-km2watershed. Beaver Creek originates in the upland forest of the White Mountains, where permafrost is continuous, flows through the Yukon Flats, which has discontinuous permafrost and extensive lowland forest, wetlands, and lakes, and discharges to the Yukon (km 1678). Two sampling locations were established on Beaver Creek to capture the distinct terrain of the watershed: Beaver Creek above Victoria Creek (BC-Victoria) includes drainage from 3315 km2of mountainous terrain; Beaver Creek near Michel Lake (BC-Michel) includes drainage from the uplands and the lake-rich lowlands (drainage area = 6164 km2). The four sampling sites were co-located with USGS gaging stations, where discharge (Q) is continuously monitored (Table 1). Station descriptions and water quantity and quality data are accessible at http://waterdata.usgs.gov/NWIS and http://ak.water.usgs.gov/yukon.
Table 1. Site Descriptions
Longitude (NAD 27)
Drainage Area (km2)
Distance From Headwaters to Sample Site (km)
USGS Station Number
Beaver Creek above Victoria
Beaver Creek near Michel Lake
Porcupine River near Ft. Yukon
Yukon River at Pilot Station
2.2. Sample Collection
 Our objective was to capture detailed seasonal and spatial variability in discharge and water chemistry by sampling across the three distinct seasons at all sites (Figure 2). We define the three seasons as: spring = May 1–June 30; summer = July 1–October 31; and winter = November 1–April 30, consistent with related published studies [Striegl et al., 2005, 2007; Holmes et al., 2011]. Furthermore, we consider annual discharge and water chemistry during October 1–September 30, defined as a “Water Year” (WY). In 2009, ten samples were collected at Yukon, with frequent sampling during the spring freshet (5 spring, 3 summer, 2 winter). In 2010, six samples were collected at Yukon (2 spring, 3 summer, 1 winter), five samples at Porcupine (2 spring, 2 summer, 1 winter), and six samples at both BC-Victoria and BC-Michel (2 spring, 2 summer, 2 winter, for each site). Samples were collected using the USGS equal discharge increment (EDI) sampling protocol (http://pubs.usgs.gov/twri/). Large volume samples (10–15 L) were filtered in the field with pre-rinsed Gelman AquaPrep 600 capsule filters (0.45μm). Samples were composited and processed according to established USGS protocols (http://water.usgs.gov/owq/FieldManual/). Unfiltered water was collected for BDOC incubation inoculation at the same time (see section 2.4, BDOC incubations). Water discharge, temperature, and pH were measured at the time of sample collection. Samples were kept cold and in the dark, and shipped on ice to the USGS laboratory in Boulder (Colorado, United States) within one week of collection.
2.3. Chemical Analyses
 DOC concentrations were analyzed using the platinum catalyzed persulfate wet oxidation method of the OI Analytical Model 700TM TOC Analyzer™ (detection limit 0.2 mg C L−1) [Aiken, 1992]. Nitrate and ammonium concentrations were measured using Dionex Model DX-300 and DX-500 Ion Chromatographs with conductivity detectors (detection limits = 0.003 and 0.013 mg N L−1, respectively). Hereafter we refer to the sum of nitrate and ammonium concentrations as dissolved inorganic nitrogen (DIN). Nitrate and ammonium are the major DIN species in the Yukon River [Dornblaser and Striegl, 2007] and in major arctic rivers [Holmes et al., 2011].
 DOM chemical composition was assessed using optical measurements (UV-visible absorbance, and excitation-emission fluorescence scans), and chromatographic separation of DOC fractions by XAD resins. UV-visible absorbance scans were collected on room temperature samples using an Agilent Model 8453™ photo-diode array spectrophotometer betweenλ= 200 and 800 nm and a 10-mm quartz cuvette, and were referenced to a blank spectrum of nanopure water. Specific ultraviolet absorbance (SUVA254), an indicator of relative aromatic C content, was determined by dividing the UV absorbance (λ = 254 nm) by DOC concentration [Weishaar et al., 2003]. Excitation-emission fluorescence matrices (EEMs) were collected over an excitation (ex) range of 240–450 nm every 5 nm, and an emission (em) range of 300–600 nm every 2 nm using a Jobin-Yvon Horiba Fluoromax-3™ fluorometer. Optically dense samples were diluted to an absorbance of less than 0.2 atλ= 254 nm prior to analysis to minimize inner-filter effects [Ohno, 2002]. EEMs were corrected using the methods of Murphy et al. . The Fluorescence Index (FI) was calculated at ex λ = 370 nm as the ratio of intensities at em λ = 470 nm to em λ = 520 nm [Cory et al., 2010]. FI indicates the relative contribution of low molecular weight non-aromatic DOM versus high molecular weight aromatic DOM, and has been used as an indicator of the relative contribution of microbially derived (FI = 1.7–2.0) and terrestrially derived (FI < 1.4) aquatic fulvic acids [McKnight et al., 2001; Cory et al., 2010]. Biological processing and photochemical degradation of DOM may also modify FI [Wickland et al., 2007; Mann et al., 2012]. Corrected EEMs were fit to the previously validated 13 component parallel factor analysis (PARAFAC) model presented by Cory and McKnight to quantify the relative abundance of fluorescing components. BDOC has been related to the relative contribution of protein-like fluorescence in several studies [Fellman et al., 2010], so we present and discuss only the relative abundance of the two protein-like fluorophores: tyrosine and tryptophan. DOC was chromatographically fractionated using Amberlite XAD-8 and XAD-4 resins [Aiken, 1992]. The hydrophobic organic acid (HPOA) fraction is generally composed of 90–95% fulvic acid and 5–10% humic acid. The hydrophilic acid (HPI) fraction contains lower molecular weight organic acids, and is generally more biodegradable than HPOA [Qualls, 2005; Wickland et al., 2007].
2.4. Biodegradable DOC (BDOC) Incubations
 BDOC incubations were conducted at 15°C, and additionally at 5°C for Yukon samples to determine the influence of temperature on BDOC. These standard temperatures allow for determination of seasonal changes in BDOC independent of temperature, and are within the annual range of riverine temperatures. For all incubations, 1L of 0.45-μm filtered water was inoculated with 10 mL of 1.6-μm filtered water collected from the same location (inoculant: sample ratio = 1:100) and gently mixed. We chose to inoculate incubations using site and date-specific water, rather than a standard inoculum because our primary aim is to assess BDOC under field conditions. Any differences we measure in BDOC among sites or seasons may therefore be due in part to differences in microbial community. Sets of 100-mL amber glass serum bottles were filled with 50 mL of inoculated sample per bottle (5 bottles per set), and capped with Teflon-lined stoppers and crimp seals. Two bottles per set were immediately acidified to pH < 3 with 1 mL of concentrated H3PO4. Bottles were placed in a 5°C or 15°C incubator; after a minimum of two hours at temperature, headspace carbon dioxide (CO2) concentrations were analyzed using a LI-COR 6252 infrared gas analyzer having N2carrier gas (0.5-mL injections, four reps per bottle). The acidified bottles signify the initial dissolved inorganic carbon (DIC) concentration; after analysis they were discarded. The remaining unacidified bottles were returned to the appropriate incubator and stored upside down to prevent headspace leakage through the septa. Headspace CO2 concentration analyses were repeated on days 7, 14, 21, and 28. After each analysis, 2 mL of CO2-free air was added back to each bottle to maintain constant pressure and to replenish oxygen. The dilution of headspace CO2 from sample removal and addition of zero air was accounted for in calculations. On day 28, bottles were acidified to pH < 3 after CO2 analysis, and then analyzed for DIC concentration. We calculated dissolved CO2 concentrations (headspace + aqueous) using known CO2 equilibrium constants [Plummer and Busenberg, 1982] adjusted for ambient temperature and pressure [Striegl et al., 2001]. Total CO2production was quantified as the difference between initial and final DIC concentrations, which accounts for any pH changes during the incubation. DIC production was assumed to originate completely from DOC respiration. DOC biodegradation over the 28-day incubation period is expressed as % BDOC (mg C-DIC produced/initial mg C-DOC; n = 3 per sample). The temperature sensitivity of %BDOC, expressed as Q10, was calculated for Yukon samples using the following equation:
 Annual and seasonal DOC and DIN loads (mass C or N t−1) at Yukon were calculated from continuous discharge and water chemistry data for WY 2009–2011 using the FORTRAN Load Estimator (LOADEST) program [Runkel et al., 2004]. The model requires at least 12 direct measurements of chemistry over a wide range of flow conditions to calculate loads by applying the method of adjusted maximum likelihood estimation (AMLE). We calculated DOC loads for WY2008–2009 using 46 DOC measurements collected January 2008 - September 2009; DOC loads for WY2010–2011 were calculated using 38 DOC measurements collected October 2009 - June 2011. We calculated DIN loads (WY2009–2011) using 19 measurements of nitrate and ammonium concentrations collected January 2009 - September 2011. There are not enough data available to accurately constrain loads for BC-Victoria, BC-Michel, and Porcupine for this time period. LOADEST centers discharge and chemical concentration data to eliminate colinearity and automatically selects one of nine predefined regression models to fit the data. Model output lists the coefficient of variation of estimated loads for the modeled flow period, the r2 of the AMLE, residuals data, and the serial correlation of residuals to quantify estimation error and to evaluate model output.
2.6. BDOC Loads Calculations
2.6.1. Yukon River
 We estimated annual and seasonal BDOC loads (kg C d−1) at Yukon based on BDOC incubation results, and DOC and DIN loads. We used two different approaches; in Method 1, we multiplied %BDOC (15°C) by initial DOC concentration to calculate corresponding BDOC concentrations (mg BDOC L−1). BDOC concentrations (n = 16) and daily discharge were input to the LOADEST model to estimate daily BDOC load for WY2009–10. In Method 2, we applied a linear equation relating log-normalized %BDOC (15°C) to the log-normalized molar ratio of [DOC]:[DIN] for Yukon (see Resultssections 3.6 and 3.8.1, equation (2) for details) to the daily molar ratio of [DOC]:[DIN] from the DOC and DIN loads. Specifically, we first determined daily [DOC]:[DIN] from the daily DOC and DIN loads; we then calculated daily ln([DOC]:[DIN]), and applied equation (2) to calculate daily ln%BDOC. Finally, we applied daily %BDOC (inverse ln%BDOC) to the daily DOC loads to estimate daily BDOC load for WY2009–11. For both Methods 1 and 2, the daily BDOC loads were summed for seasonal and annual BDOC loads. Upper and lower estimates of the loads were calculated for Method 1 by multiplying seasonal or annual BDOC loads by the coefficient of variation of the LOADEST run (CV = 8.2%). Upper and lower load estimates for Method 2 were calculated by applying the standard error of the estimate for equation (2) (±0.28856) to the daily loads. BDOC load is considered as the portion of the total DOC load which is decomposable over 28 days at 15°C under native [DOC]:[DIN] condition.
 We applied the two BDOC load estimation methods assuming that while Method 1 may be more accurate, the utilization of a relation between %BDOC and water chemistry parameters that are often measured in surface waters (Method 2) has the potential to be more broadly applied to other systems in the absence of direct BDOC incubation data. We reason that if BDOC loads estimated by Methods 1 and 2 are comparable, then Method 2 is a valid approach.
2.6.2. Major Arctic Rivers
 The Pan-Arctic River Transport of Nutrients, Organic Matter, and Suspended Sediments (PARTNERS) project measured discharge and water chemistry of the six largest arctic rivers in Russia, Canada, and Alaska, during all seasons of 2003–06 [Holmes et al., 2011]. Seasonal and annual loads of DOC and DIN (nitrate + ammonium) for the Kolyma, Ob, Yenisey, Lena, Mackenzie, and Yukon Rivers during 1999–2008 were calculated using LOADEST [Holmes et al., 2011]. We used these published values to estimate average annual and seasonal BDOC loads by Method 2 (described above). We applied the relation between log-normalized %BDOC and [DOC]:[DIN] for the Yukon River (see Resultssection 3.6 and 3.8.1, equation (2)) to the average annual and seasonal [DOC]:[DIN] molar loads for each river (1999–2008), to calculate average annual and seasonal %BDOC values. We then estimated BDOC loads from %BDOC and total DOC loads for each river; upper and lower ranges of BDOC loads were calculated using the standard error for equation (2). While these values are not determined from actual incubations, they are useful for considering the possible magnitude of BDOC in these rivers where currently very little incubation data exist.
 All statistical analyses were conducted using Statistica 10 (StatSoft, Inc.). We used the Simple Regression function to test for significant correlations between %BDOC and various parameters, and between Q10 and various parameters.
3.1. Water Discharge
 Annual discharge (Q) spanned a wide range: BC-Victoria = 0.457 km3, BC-Michel = 0.626 km3, Porcupine = 18.5 km3, and Yukon = 207 and 184 km3(WY2009 and WY2010, respectively). Seasonal patterns of discharge were typical for high-latitude rivers, with some notable exceptions (Figures 2a–2d). Spring and summer each represented 45% of annual discharge at both BC-Victoria and BC-Michel, with winter comprising the remaining 10%. Porcupine typically follows this pattern [Striegl et al., 2007], but WY2010 was strikingly different (Figure 2c). Spring discharge (23% of annual Q) was much lower than average [Striegl et al., 2007], and there was an extended period of high discharge during mid-July through late August (summer Q = 72% of annual Q). This is likely a response to greater than average precipitation during July and August in the upper Porcupine watershed (Environment Canada Old Crow RCS Weather Station,http://www.climate.weatheroffice.gc.ca/). Discharge at Yukon during WY2009 was similar in seasonal pattern and magnitude to the long-term record [Striegl et al., 2007], whereas WY2010 had smaller spring discharge and greater summer discharge. The seasonal distribution of annual discharge was spring = 39%, 26%, summer = 45%, 58%, and winter = 16%, 16% (WY2009 and WY2010, respectively).
3.2. Dissolved Organic Carbon Concentrations and Chemical Composition
 DOC concentrations were strongly seasonal at all sites, and generally greater at Yukon than the other sites (Table 2). Specific UV absorbance (SUVA254) and Fluorescence Index (FI) values were similar in magnitude and seasonal pattern across all sites (Table 2), suggesting common DOM chemical composition and source signature with hydrologic regime, and minimal biological processing between sites. During winter, SUVA254 values were low (mean ± stdev for all sites = 2.2 ± 0.3 L mg C−1 m−1) and FI values were high (1.52 ± 0.03), indicating low DOM aromaticity and a “microbial” source signature or, alternatively, extensive degradation. Spring freshet SUVA254 values indicate high aromaticity (3.4 ± 0.4 L mg C−1 m−1), and FI values indicate DOM dominated by terrestrial plant sources (1.39 ± 0.02). Summer SUVA254 and FI values are similar to spring (SUVA254 = 3.0 ± 0.3 L mg C−1 m−1, FI = 1.41 ± 0.02). Seasonal trends in SUVA254 and FI are consistent with other studies of the Yukon River Basin [Spencer et al., 2008, 2009; O'Donnell et al., 2010], and with the Kolyma River in Siberia [Neff et al., 2006; Mann et al., 2012]. PARAFAC analyses of fluorescence EEMs indicate that the relative abundance of tryptophan-like fluorescence is greatest in winter and lowest during spring and summer (Table 2). Tyrosine-like fluorescence is relatively stable across all seasons (Table 2). Mann et al. similarly observed high contributions from protein-like fluorescence in the Kolyma River just before spring ice-out.
Table 2. Discharge, Water Chemistry, and Biodegradable DOCa
Mean Daily Discharge (m3 s−1)
DOC (mg C L 1)
Nitrate (mg N L−1)
Ammonium (mg N L−1)
DOC: DIN (molar)
% BDOC 5°C
% BDOC 15°C
SUVA254= specific UV absorbance at 254nm; FI = Fluorescence Index; % Tryp = %Tryptophan-like fluorescence; %Tyros = %Tyrosine-like fluorescence;%HPOA = % hydrophobic organic acids; %HPI = % hydrophilic acids; DIN is the sum of nitrate and ammonium concentrations. BDOC is presented as Mean (Stdev). ND = No Data.
 Relative amounts of HPOA and HPI fractions were broadly similar across all sites (Table 2). HPOA was dominant in all samples, but was less important in winter (mean ± stdev = 45% ± 3%) than in spring (53% ± 4%) and summer (51 ± 4%). HPI also varied seasonally (winter = 21% ± 3%, spring = 17 ± 1%, summer = 18 ± 3%). Earlier studies in the Yukon River basin [Striegl et al., 2007; O'Donnell et al., 2010] have measured similar seasonal patterns.
3.3. Dissolved Inorganic Nitrogen Concentrations and [DOC]:[DIN]
 Nitrate concentrations spanned a large range, with greatest concentrations during winter and lowest concentrations in summer (Table 2). Similar seasonal patterns have been observed in the Yukon River basin [Dornblaser and Striegl, 2007], and other major arctic rivers [Holmes et al., 2011; Mann et al., 2012]. Ammonium concentrations were low and stable at BC-Victoria, BC-Michel, and Porcupine, but varied by almost one order of magnitude at Yukon (Table 2). Holmes et al.  report that ammonium concentrations in major arctic rivers are often below detection limit.
 The molar ratio of [DOC]:[DIN] (Table 2) showed opposing seasonal variation in organic C and inorganic N at all sites. In winter, [DOC]:[DIN] < 20, except at BC-Michel. During spring freshet ratios rapidly increase at all sites to >90. Summer [DOC]:[DIN] ratios are variable across sites, falling below 100 at Yukon and BC-Victoria, but remaining >100 at BC-Michel and Porcupine.
3.4. Biodegradable DOC Incubations
3.4.1. Seasonal and Spatial Patterns
 BDOC showed striking seasonal patterns consistent across sites (Table 2 and Figure 3). The greatest %BDOC for each site was measured in winter (35–53% BDOC). The 9% BDOC for BC-Michel (2/24/10) was the lowest winter value we measured. Spring freshet had the second highest %BDOC (17–38% BDOC), except at Porcupine. Frequent sampling during WY2009 spring freshet at Yukon illustrates distinct changes in %BDOC (Figures 2 and 3 and Table 2). The greatest %BDOC of spring freshet, and of the entire open-water season for Yukon, occurred on the rising limb of the hydrograph (Figures 2 and 3). The % BDOC decreased as spring freshet proceeded, with the lowest measured values at peak discharge and the falling limb of the hydrograph. Summer %BDOC was variable among sites: BC-Victoria and BC-Michel had the lowest values (3–8%BDOC), and Porcupine was low until the July high discharge event, when %BDOC was four times that of spring (Table 2). Summer %BDOC at Yukon was generally higher than spring peak discharge (10–36%BDOC) (Table 2 and Figure 3).
3.4.2. Response to Temperature
 The temperature sensitivity of %BDOC at Yukon, expressed as Q10, ranged from almost no effect (1.1) to 3.2 at peak spring discharge (Table 2 and Figure 3). The mean Q10 of 2.0 ± 0.6 that we observed is a widely accepted value for organic matter decomposition in terrestrial ecosystems [Davidson and Janssens, 2006]. There is a positive relation between Q10 and %HPOA (R2 = 0.33, F = 5.5, p < 0.05), which is consistent with the observation that temperature sensitivity of decomposition increases with greater molecular complexity of the substrate [Davidson and Janssens, 2006].
3.5. Relations Between BDOC and DOM Chemical Composition
 BDOC, on a percentage basis, was significantly related to SUVA254 and Fluorescence Index (FI) (Table 3). The %BDOC (15°C) decreased as SUVA254 increased for both Yukon and all sites combined. The %BDOC is significantly related to FI for both Yukon and all sites combined (Table 3), where DOM that has a terrestrial plant source signature (low FI value) has lower %BDOC than DOM having a more microbial source signature (high FI value). A high FI can also indicate extensive degradation of DOM, and thus should be interpreted with care. The %BDOC was significantly positively related to % tryptophan-like fluorescence for Yukon and for all sites combined (Table 3). We found significant relations between %BDOC and total % protein-like fluorescence that were weaker than with % tryptophan-like fluorescence alone (Table 3), and no significant relations with % tyrosine-like fluorescence alone.
Table 3. Linear Regressions Between %Biodegradable DOC (15°C) and Chemical Parametersa
SUVA254= specific UV absorbance at 254 nm; FI = Fluorescence Index; %Tryp = %Tryptophan-like fluorescence; %Tyros = %Tyrosine-like fluorescence;%HPOA = % hydrophobic organic acids; %HPI = % hydrophilic acids; DIN is the sum of nitrate and ammonium concentrations; [DOC]:[DIN] is the molar ratio; %BDOC is at 15°C.
%BDOC = (−24.6 * SUVA254) + 98.5
%BDOC = (−12.7 * SUVA254) + 57.3
%BDOC = (266 * FI) − 355
%BDOC = (171 * FI) − 224
%BDOC = (2029 * %Tryp) + 3.32
%BDOC = (1281 * %Tryp) + 7.22
% Tryp + % Tyros
%BDOC = (1307 * (%Tryp + %Tyros)) − 33.4
%BDOC = (359 * (%Tryp + %Tyros)) + 3.01
%BDOC = (−234 * %HPOA) + 144
%BDOC = (−151 * %HPOA) + 96.3
%BDOC = (223 * %HPI) − 21.2
%BDOC = (217 * DIN) − 9.48
%BDOC = (140 * DIN) + 1.35
ln%BDOC = (−0.694 * ln(DOC:DIN)) + 5.71
ln%BDOC = (−0.589 * ln(DOC:DIN)) + 5.17
 There are weak significant relations between %BDOC and both %HPOA and %HPI (Table 3). The %BDOC increases with decreasing %HPOA and with increasing %HPI. This is consistent with the observation that hydrophobic acids are less accessible to microbial degradation than hydrophilic compounds [Kalbitz et al., 2003a].
3.6. Relations Between BDOC and Dissolved Inorganic N
 The %BDOC (15°C) is highly significantly related to DIN concentration for both Yukon, and for all sites combined (Table 3). We find an even stronger linear relation between %BDOC and the molar ratio of [DOC]:[DIN] for both Yukon and all sites combined after data are lognormalized, where %BDOC is greatest at low molar ratios (Table 3 and Figures 4a and 4b). Winter samples have the lowest [DOC]:[DIN] ratios and greatest %BDOC, and spring and summer samples generally have higher ratios and lower %BDOC (Table 2 and Figures 4a and 4b). During spring freshet WY2009 at Yukon, [DOC]:[DIN] shifts rapidly from 34 on the rising limb of the hydrograph (when BDOC concentration is greatest), to >100 within days, with a corresponding threefold decrease in %BDOC. As spring freshet progresses [DOC]:[DIN] ratios increase to >200, and %BDOC decreases to 7% at peak discharge. Summer [DOC]:[DIN] ratios fall below 100, and %BDOC increases moderately. The ratio of [DOC]:[DIN] appears to explain seemingly anomalous %BDOC results that do not fit the general seasonal patterns we observed. Specifically, the low %BDOC measured at BC-Michel on 2/24/10 and at Porcupine on 5/24/10 may be attributed to moderate to high [DOC]:[DIN] values of 42 and 589, respectively.
3.7. DOC and DIN Loads, Yukon River
 Annual and seasonal loads of DOC (kg C time−1) and DIN (kg N time−1) at Yukon during WY2009–2011 are presented in Table 4. Annual DOC loads varied during the three years; WY2009 was similar to the 2001–05 mean of 1585 × 106 kg C yr−1 [Striegl et al., 2007], while WY2010 was ∼30% smaller, and WY2011 was ∼25% greater, than WY2009. These differences can be attributed to a small spring DOC load in WY2010, and a large summer DOC load in WY2011. Annual DIN loads were similar for the three years, and are comparable to loads presented by Dornblaser and Striegl , and Guo et al. . Seasonal DIN loads were similar to 2001–05 seasonal means of 6.54, 9.05, and 6.00 × 106 kg N for spring, summer, and winter respectively [Dornblaser and Striegl, 2007]. In contrast to DOC loads, a disproportionately large amount of DIN is transported in the winter (28–31% of annual) compared to spring (25–33%) and summer (36–44%) relative to seasonal discharge.
Table 4. Annual and Seasonal DOC, DIN, and Biodegradable DOC (BDOC) Loads, Yukon River at Pilot 2009–2011a
See Section 2.6 for description of Method 1 and Method 2 for estimation of BDOC loads (Total and Range). Briefly, Method 1 uses LOADEST and BDOC concentrations to calculate BDOC Loads; Method 2 uses Equation 2 (see Section 3.8.1) and DOC and DIN Loads to calculate BDOC Loads. BDOC Loads are presented as Mean (Range). NA = Not Available.
Summer 2010 BDOC load Method 1 is July 1- Sept 30 due to calculation constraints.
 We estimated BDOC using two methods (see Methods section 2.6); in Method 2, we applied the linear equation relating log-normalized %BDOC (15°C) to the log-normalized molar ratio of [DOC]:[DIN] for Yukon (equation (2), Table 3, and Figure 4a) to the daily molar ratio of [DOC]:[DIN] from the DOC and DIN loads:
The annual BDOC loads estimated by Method 1 and Method 2 for WY2009–2010 were within 5% of each other, and seasonal loads were within 7–14% of each other (Table 4), indicating that estimation of BDOC loads by [DOC]:[DIN] for the Yukon is a valid approach. Therefore, we limit our discussion to BDOC loads estimated by Method 2.
 The annual BDOC loads averaged 15%, 18%, and 12% of total annual DOC load for the three years. There was a strong seasonality where the relative proportion of the total DOC load calculated as BDOC increased from spring (10–15%) to summer (11–17%) to winter (41–51%). The absolute BDOC load (kg C) was smallest in winter, and was greatest in either spring or summer depending on the year (2009: spring > summer; 2010–11: spring < summer). The seasonal distribution of annual BDOC load is 36% ± 7 in spring, 44% ± 7 in summer, and 20% ± 3 in winter during 2009–11. Although winter has the smallest BDOC load, it is a significant portion of the annual BDOC load.
3.8.2. Major Arctic Rivers
 The DOC and DIN loads, and the [DOC]:[DIN] molar ratios for the six major arctic rivers studies by the PARTNERS project are highly seasonal (Table 5) [Holmes et al., 2011]. Considering the similar seasonal patterns in [DOC]:[DIN] ratios across major arctic rivers, the importance of DIN to heterotrophic growth and DOC metabolism in arctic systems [Kirchman and Wheeler, 1998; Thingstad et al., 2008], and the significant control of C:N stoichiometry on microbial processes coupling C and N cycles [Thingstad et al., 2008; Taylor and Townsend, 2010], we suggest that [DOC]:[DIN] is an important and perhaps overriding control of BDOC in all these rivers. Furthermore, these rivers exhibit spatial and seasonal synchrony in bacterioplankton community assemblages, indicating similar relations between hydrology, biogeochemistry, and microbial communities [Crump et al., 2009]. Therefore, we utilize the relation between lognormalized %BDOC and [DOC]:[DIN] for the Yukon River (equation (2)) and apply it to the average DOC and DIN loads for the other major arctic rivers to calculate BDOC loads (see Methods section 2.6).
 We estimated mean annual %BDOC ranged from 8% (Lena River) to 19% (Ob River); the Yukon River was calculated at 16%, which is similar to our 2009–11 estimates (Tables 4 and 5 and Figure 5). The annual BDOC loads were smallest for the Kolyma River, and increased from the Yukon and Mackenzie Rivers < Lena River < Yenisey River < Ob River. The Lena and Yenisey Rivers have greater annual discharge than the Ob River, but the [DOC]:[DIN] ratio is two to four times lower for the Ob River than the Yenisey and Lena Rivers. The seasonal %BDOC trend within each river was winter > spring, summer (Table 5 and Figure 5), similar to our calculations for the Yukon River. When we sum the annual BDOC loads for all rivers, we find that the collective total average annual export of BDOC to the Arctic Ocean (defined here as decomposable over 28 days at 15°C under native [DOC]:[DIN] condition) is 2.3 Tg C yr−1.
Table 5. Calculated Biodegradable DOC Loads for Major Rivers Flowing to the Arctic Ocean, 1999–2008a
 We acknowledge that while these values are not determined from actual incubations, they are useful for considering the possible magnitude of BDOC in these remote rivers where currently very little incubation data exist. Some data are available on the biodegradability of DOC for the Kolyma and Yenisey Rivers (Figure 5). The results from Mann et al.  for the Kolyma River in spring are within the range we estimated, but their winter results are lower than we estimated (see Discussion section 4.1 for further discussion). Kawahigashi et al. measured BDOC from six tributaries of the Yenisey River spanning a latitudinal gradient in mid-August 2002, and their mean %BDOC of 8.9% ± 4.8 compares well with our estimated 7% BDOC (range = 5–9%) for the Yenisey River in summer.
 Northern high-latitude rivers deliver large amounts of terrestrially derived DOC to the coastal and oceanic environments of the Arctic Ocean. The relative biodegradability of riverine DOC is strongly seasonal, and inorganic N availability is a primary control of DOC biodegradability in the Yukon River system. DOC transported during winter and early spring is highly biodegradable as a result of low nutrient limitation and labile DOM chemical composition. Greatest BDOC exports from the Yukon River occur during spring and summer, although winter accounts for approximately 20% of the annual BDOC load. Based on similar seasonal dynamics of DOC and DIN in other major arctic rivers, we assert that biodegradable DOC is delivered to the Arctic Ocean year-round.
4.1. BDOC Seasonal and Spatial Variability
 Despite the huge range in drainage areas, we found remarkably similar BDOC amounts and seasonal trends across sites. There was no significant difference in %BDOC or BDOC concentration among sites, suggesting hydrology is an overall driver of delivering terrestrially derived BDOC to aquatic systems across a range of stream sizes in the Yukon River Basin. Therefore we expect a consistent response of BDOC to permafrost thaw across the Yukon River Basin, which covers a relatively small latitudinal range. Arctic and boreal rivers draining watersheds that cover large gradients in permafrost extent have latitudinal differences in %BDOC [Kawahigashi et al., 2004], and thus response to permafrost thaw may be more complex. Consistent %BDOC and DOM chemical composition among our sampling sites suggests that minimal processing of DOC occurs within the Yukon River main stem during transport. Factors such as low water temperatures in winter, low nutrient concentrations in spring and summer, high sediment loads in the Yukon [Dornblaser and Striegl, 2009] that inhibit UV penetration into the water column, and the lack of large reservoirs that could increase residence times, likely contribute to this. An outcome is that DOM quality is relatively well preserved, even at the mouth of the Yukon, allowing for a greater characterization of the source material and for connecting watershed processes with DOM composition.
 The high summer %BDOC at Porcupine suggests that BDOC is present in soils and/or litter layers even in summer, and that hydrologic transport to aquatic systems, rather than BDOC supply, limits riverine BDOC in late summer. This may be specific to watersheds dominated by wetlands, such as the Porcupine River basin. Fellman et al. [2009b]measured increased %BDOC during high-flow storm events in streams draining wetland-dominated catchments, but decreased %BDOC in an upland catchment stream. The late season increase of %BDOC in the Yukon may be due to contributions of high BDOC glacial meltwater sources [Hood et al., 2009], such as the Tanana River basin [Striegl et al., 2007]. Increasing contributions of groundwater and permafrost thaw water in late summer may also have high BDOC [Balcarczyk et al., 2009; Wickland et al., 2010].
 Our results are comparable to the few other studies of %BDOC in high-latitude streams and rivers.Holmes et al.  measured 14–33% BDOC during spring freshet on three arctic rivers in Alaska, and 0–9% BDOC in July at low discharge. Mann et al. measured a maximum of 20% BDOC during spring break-up on the Kolyma River, followed by 1–12% BDOC in early June. BDOC was 12–30% in southeast Alaska streams, and greatest in spring and late summer [Fellman et al., 2009a]. Small streams near Fairbanks, AK had 5–35% BDOC in June [Balcarczyk et al., 2009]. Kawahigashi et al. measured 4–28% BDOC in mid-August from six small tributaries to the Yenisey River spanning a latitudinal gradient.
 This is the first published study of winter BDOC from northern high-latitude rivers, to our knowledge. Related studies have determined that many components of winter flow, such as groundwater, permafrost thaw water, and glacial runoff, contain DOC that is highly biodegradable [Balcarczyk et al., 2009; Hood et al., 2009; Wickland et al., 2010].
 The closest comparison is Mann et al. who measured 0.1–9% BDOC in the Kolyma immediately before spring ice-out, which is substantially lower than our results. This may be attributed to differences in riverine DOM and/or other environmental conditions, or to BDOC assessment methodologies.Mann et al.  determined BDOC as the change in DOC concentration (DOC loss method), and we determined BDOC as CO2 production. While both methods are commonly used, results may not always be directly comparable as they consider the multiple processes involved in biodegradation to different degrees [Marschner and Kalbitz, 2003; McDowell et al., 2006]. The utilization of DOC by microorganisms includes synthesis of biomass, and complete mineralization to CO2 to obtain energy and inorganic nutrients [Marschner and Kalbitz, 2003]. The DOC loss method does not differentiate between microbial incorporation and mineralization, and net transfer of DOC into biomass is accounted for only if incubation samples are filtered prior to analysis [Marschner and Kalbitz, 2003; McDowell et al., 2006]. Microbes also produce DOC during DOM degradation [Kalbitz et al., 2003b], possibly reducing the net change in DOC concentration.
4.2. Controls of DOC Biodegradability: DOM Composition and Inorganic N
 Variability of %BDOC in the Yukon River and tributaries was explained by shifts in both DOM composition and [DOC]:[DIN] ratios. The %BDOC was significantly related to DOM optical properties, where %BDOC increased with lower SUVA254, higher FI, and greater contribution from protein-like fluorophores. Other studies have observed %BDOC increases with decreasing SUVA for aquatic and terrestrial DOM [Kalbitz et al., 2003a; Fellman et al., 2009a; McDowell et al., 2006], and with increasing % protein-like fluorescence (the sum of the % tryptophan-like and % tyrosine-like fluorescence) [Fellman et al., 2009a; Balcarczyk et al., 2009; Hood et al., 2009]. We found that %BDOC was most related to the tryptophan-like peak, which is thought to indicate amino acids and less-degraded peptide material [Fellman et al., 2010].
 The highly significant relation between BDOC and [DOC]:[DIN] suggests that while DOM chemical composition is an important driver of BDOC, the relative availability of inorganic nutrients is an overriding control of seasonal and spatial variability of riverine BDOC in the Yukon River Basin. Nutrient availability appears to be particularly limiting to DOC metabolism in spring and summer. It is well-established that additions of inorganic nutrients, including nitrate and ammonium, commonly result in higher %BDOC during laboratory incubations [Marschner and Kalbitz, 2003; McDowell et al., 2006; Holmes et al., 2008], so it is not surprising that riverine water enriched with DIN has greater %BDOC. This is consistent with the controls of stoichiometric shifts in [DOC]:[NO3−] molar ratio on microbial processes that couple C and N cycles, where C is limiting at low ratios and N becomes limiting at higher ratios [Taylor and Townsend, 2010]. Inorganic nutrient availability as a limiting control of both heterotrophic bacterial production and DOC metabolism has been demonstrated in marine, estuarine, and riverine systems [Kirchman and Wheeler, 1998; Caraco et al., 1998; Skoog et al., 1999; Middelburg and Nieuwenhuize, 2000; Sobczak et al., 2003; Thingstad et al., 2008]. Heterotrophic bacteria assimilate nitrate and ammonium because organic substrates often contain insufficient N to support bacterial growth completely [Kirchman and Wheeler, 1998; Middelburg and Nieuwenhuize, 2000]. The mechanism by which inorganic nutrient availability limits biodegradation of DOC is presumably through limitation on microbial growth. Thingstad et al. demonstrated the stoichiometric coupling between DOC and growth-limiting mineral nutrients in an Arctic pelagic ecosystem. Using controlled mesocosms they determined that heterotrophic uptake of DOC was inhibited when inorganic N and phosphorus limited bacterial growth rate. Arctic Rivers generally have among the greatest concentrations of DOM and the lowest concentrations of inorganic nutrients reported for rivers worldwide [Dittmar and Kattner, 2003]. We therefore expect that nutrient availability is an important control of BDOC in other northern high-latitude rivers.
 The fate of terrigenous DOM delivered by the six major arctic rivers to the Arctic Ocean has been increasingly studied during the past decade because of its importance to primary productivity and energy budgets, and because of recent shifts in arctic climate and hydrology [Hansell et al., 2004; Granskog et al., 2007; Thingstad et al., 2008; Stedmon et al., 2011]. Coastal and oceanic studies suggest that <50% of riverine DOM exported to the Arctic Ocean reaches the North Atlantic Ocean [Amon et al., 2003; Hansell et al., 2004], and that a significant portion does not survive transport across the shelf into the Arctic Ocean due primarily to microbial degradation [Hansell et al., 2004; Manizza et al., 2009; Stedmon et al., 2011]. While we estimate that <20% of DOC exported from major arctic rivers is biodegradable over 28 days, our finding of nutrient limitation of BDOC suggests rapid DOC mineralization in nutrient-rich coastal and oceanic waters. Our seasonal measurements of BDOC and [DOC]:[DIN] ratios suggest DOC exported in winter is quickly removed in near-coastal regions due to labile DOM chemical composition and high nutrient concentrations, while DOC exported during spring and summer may persist longer depending on nutrient availability in the marine environment.
 Our study of BDOC in the Yukon River and two of its tributaries provides new insight into the relative importance of DOM chemical composition and nutrient availability across seasons, and presents the first estimates of seasonal and annual BDOC fluxes from the six largest arctic rivers. Consistent seasonal patterns in BDOC across a wide range of watershed size indicate DOC transported during winter is highly biodegradable. This contradicts suggestions that winter riverine DOC is relatively recalcitrant based on age and/or certain measures of chemical composition [Stedmon et al., 2011; Mann et al., 2012]. Understanding winter riverine DOC biodegradability is particularly important as recent increases in winter base flow have been documented across arctic and boreal regions [Walvoord and Striegl, 2007; Smith et al., 2007]. We conclude that increased winter base flows will result in greater exports of biodegradable DOC to the Arctic Ocean, although low DOC concentrations and Q during winter result in relatively small BDOC loads compared to spring and summer.
 We thank K. Kelsey, J. Koch, D. Repert, and P. Schuster for assistance with laboratory analyses. We also thank E. Stets and two anonymous reviewers for their insightful comments on an earlier version of the manuscript. The National Research Program and the Climate Effects Network of the Water and the Climate and Land Use Change Mission Areas of USGS supported this research. Any use of trade or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.