Carbon export and cycling by the Yukon, Tanana, and Porcupine rivers, Alaska, 2001–2005



[1] Loads and yields of dissolved and particulate organic and inorganic carbon (DOC, POC, DIC, PIC) were measured and modeled at three locations on the Yukon River (YR) and on the Tanana and Porcupine rivers (TR, PR) in Alaska during 2001–2005. Total YR carbon export averaged 7.8 Tg C yr−1, 30% as OC and 70% as IC. Total C yields (0.39–1.03 mol C m−2 yr−1) were proportional to water yields (139–356 mm yr−1; r2 = 0.84) at all locations. Summer DOC had an aged component (fraction modern (FM) = 0.94–0.97), except in the permafrost wetland-dominated PR, where DOC was modern. POC had FM = 0.63–0.70. DOC had high concentration, high aromaticity, and high hydrophobic content in spring and low concentration, low aromaticity, and high hydrophilic content in winter. About half of annual DOC export occurred during spring. DIC concentration and isotopic composition were strongly affected by dissolution of suspended carbonates in glacial meltwater during summer.

1. Introduction

[2] Hydrology is changing across subarctic and arctic regions. Permafrost degradation, construction of hydroelectric dams, and changing climate have resulted in changed amount and/or timing of water discharge (Q) in recent decades [Peterson et al., 2002; Ye et al., 2003; McClelland et al., 2004, 2006; Dery and Wood, 2005]. Coincident changes in the ecology and biogeochemistry of northern watersheds [Serreze et al., 2000; Chapin et al., 2005; Hinzman et al., 2005] have altered the relationship between Q and carbon (C) transport in the Yukon [Striegl et al., 2005] and possibly in other northern river basins. However, there are few historical water chemistry data to confirm this. Until 2001, data on flow and C chemistry of the Yukon River (YR) and its tributaries were mostly of a short term and sporadic nature [Kempe, 1982; Telang et al., 1991; Brabets et al., 2000; U.S. Geological Survey National Water Information System, Water Data for the Nation,]. During 2001–2005, the National Stream Quality Accounting Network, the Alaska Water Science Center, and the National Research Program of the United States Geological Survey (USGS) combined efforts to conduct a systematic study of Q and water chemistry at three locations on the YR and two of its major tributaries, the Tanana River (TR) and the Porcupine River (PR) [Schuster, 2003;]. This paper focuses on dissolved and particulate C measurements made during the Yukon basin study. It documents Q and C concentration, export, and yield and C composition at the five locations. Additionally, the three YR sites allow for interpretation of the relative importance of within-river cycling of organic and inorganic C (OC, IC). Although such interpretations are limited by the large size of the basin and the long distances between sampling locations, they provide a platform for hypothesis formulation and for more detailed biogeochemical assessment of the largest unregulated river in the United States.

2. Study Area

[3] The 3340 km Yukon River drains 853,300 km2 of northwest Canada and Alaska, United States, discharging to Norton Sound on the Bering Sea (Figure 1). The basin is mostly wilderness, with ∼115,000 of its <130,000 inhabitants residing in or near Fairbanks, AK and Whitehorse, Yukon Territory (YT). Precipitation varies across the basin, ranging from ∼250 mm yr−1 in the upper Porcupine basin, to ∼280 mm yr−1 in other areas of interior AK and YT, ∼400 mm yr−1 near the Yukon Delta and the Bering Sea, and >1500 mm yr−1 in headwater areas of the St. Elias and Coast Mountains. Permafrost underlies much of the basin and thawing of permafrost is widespread, causing deepening of the soil active layer, slope and bank erosion, and thermokarst formation [Osterkamp and Romanovsky, 1999; Jorgenson et al., 2001]. Basin lowlands include black spruce forest, extensive areas of wetlands and lakes known as “flats”, shrublands, and tundra. Vegetated uplands include spruce, pine, aspen, and birch forest, shrublands and tundra. There are extensive ice and snow fields in headwater areas of Alaska Range and the Wrangell, St. Elias, and Coast Mountains [Brabets et al., 2000].

Figure 1.

Map of the Yukon River basin showing measurement station locations and watershed boundaries. YRE, Yukon River at Eagle; PR, Porcupine River; YRS, Yukon River near Stevens Village; TR, Tanana River at Nenana; YRP, Yukon River at Pilot Station.

[4] Tributaries to the YR can be categorized into three general types with respect to C. Meltwater-dominated rivers, including the TR, have high sediment load, potentially high particulate and dissolved IC (PIC, DIC), and low dissolved OC (DOC). They originate from perennial ice and snow fields and alpine glaciers and may carry large amounts of unweathered glacial flour, including carbonates. Black-water-dominated rivers, including the PR, are highly colored with DOC and originate from permafrost wetlands and other peat-dominated lands. They typically have high DOC, low DIC and low sediment concentration. Clear water rivers typically originate from zones of groundwater discharge, have low DOC and sediment concentration and can have high DIC concentration. Although clear water streams exhibit a unique C chemistry, they tend to be smaller than, and drain to, the other tributary types.

[5] The Tanana and Porcupine watersheds are thought to be particularly sensitive to climate warming, but for different reasons. The Tanana and similar watersheds originating in mountainous terrain in the south central and southeastern portions of the Yukon basin are experiencing extensive melting of alpine glaciers and perennial snow fields. The Porcupine and many other low-lying watersheds in the Yukon basin are experiencing progressive permafrost degradation, thermokarst formation, and drying in locations where surface waters perched on thawing permafrost have drained [Jorgenson et al., 2001; Yoshikawa and Hinzman, 2003; Riordan et al., 2006].

[6] The YR integrates drainage from throughout the basin, receiving a mix of water from all tributary types. It flows clear (Secchi transparency > 1 m) from its headwaters in British Columbia to its confluence with the White River (WR; km 815), after which it is sediment laden, having a transparency of ∼0.02 m. Mean cross-sectional velocities can range from <0.12 m s−1 under ice to >3.0 m s−1 during high flow. Ice cover is usually from November into May, averaging ∼200 days yr−1. Late summer water temperature can reach 20°C. The river is a major waterway for salmon migration from the Bering Sea, but provides limited habitat for benthic or primary production in AK portions of the river.

3. Methods

3.1. Data Collection

[7] USGS established five fixed stations in the Yukon basin, AK to continuously measure Q and to systematically measure water and sediment chemistry during 2001–2005 (Figure 1 and Table 1). Yukon River at Eagle (YRE; km 1043) receives a mix of waters from snow and ice fields, headwater lakes, and boreal forest. Porcupine River near Fort Yukon (PR), located 201 km upstream of the confluence with YR (km 1450), is a high DOC river receiving black water drainage. Yukon River near Stevens Village (YRS; km 1744) includes drainage from PR and from the 35,000 km2 Yukon Flats. The Tanana River near Nenana (TR), 286 km upstream of the confluence with YR (km 1973), receives sediment-laden meltwater from the Alaska Range and drainage from extensive boreal forest and the Fairbanks area. Yukon River at Pilot Station (YRP; km 3137) receives drainage from the entire Yukon basin above the Yukon Delta and is the furthest downstream location where the river is contained in a single channel and is unaffected by tidal fluctuation. Station descriptions and water quantity and quality data for the five fixed stations are accessible at and

Table 1. Station Name, Location, and Water Discharge at the Five Measurement Stations for 2001–2005 and Mean Annual, Peak, and Low Q for the Period of River Flow Record
Station NameUSGS Station NumberLatitude, °N, Longitude, °W (NAD 83)Drainage Area, km2Elevation, mAnnual Q, km3 yr−1Period of RecordMean Annual Q, km3 yr−1Peak Q, m3 s−1Low Q, m3 s−1
Yukon River at Eagle1535600064°47′22″, 141°47′22″294,00025991.174.565.471.079.11951–200575.115,400 (12 Jun 1964)204 (Feb 1951)
Porcupine River near Fort Yukon1538900066°59′26″, 143°08′16″76,40015810.812.111.19.309.521965–1978, 2001–200512.48,470 (24 May 1973)0 (Mar 1966, Mar 1967)
Yukon River near Stevens Village1545350065°52′32″, 149°43′04″508,4007312310898.996.61081977–200510626,500 (15 Jun 1964)396 (Apr 1997)
Tanana River at Nenana1551550064°33′55″, 149°05′30″66,30010323.323.822.424.124.71963–200521.85,270 (18 Aug 1967)113 (Mar 1974)
Yukon River at Pilot Station1556544761°56′04″, 162°52′50″831,40062231882251902271976–1995, 2001–200520535,100 (16 May 2005)991 (Feb–Mar 1984)

[8] The sampling objective was to characterize all flow conditions at all sites over the 5-year sampling period using USGS equal discharge increment (EDI) sampling protocol ( Samples were collected at approximately 3 week intervals from ice melt through September. One sample was collected each year in March to characterize winter base flow. Water discharge measurements were made before each EDI sample was collected. Additional samples to characterize DOC chemistry and specific ultraviolet absorbance (SUVA), dissolved carbon dioxide [CO2], [DIC], and δ13C and fraction modern (FM) of 14C in carbon species were collected from just beneath the water surface at the center of flow [Schuster, 2003; Striegl et al., 2005]. During the summers of 2003 (YRS) and 2004 (YRE, YRP, and YR at Carmacks, YT), more frequent samples were collected to characterize temporal variability of DIC, and δ13C-DIC. Carmacks is upstream of the White River, the major source of PIC to the YR.

[9] DOC was chromatographically fractionated using XAD resins into hydrophobic organic acid (HPOA), transphilic acid (TPIA) and hydrophilic fractions (HPI) [Aiken et al., 1992]. The carbonate content of suspended sediment was determined by X-ray diffraction of aliquots of suspended sediment [Eberl, 2004]. PIC was calculated as the product of the EDI sediment concentration and measured carbonate content of the suspended sediment. Aliquots of DOC were incubated for four weeks in the laboratory at 20°C to estimate potential within-river DOC mineralization [Kalbitz et al., 2003].

3.2. Loads and Yields

[10] River C loads (mass C t−1) were calculated from continuous Q data and water and suspended sediment chemistry measurements 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 used 53 measurements at YRE, 34 at PR, 41 at YRS, 38 at TR, and 43 at YRP, collected during October 2000 to September 2005. LOADEST centers Q 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 calculated 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 confirm normal distribution of the data.

[11] Yields were calculated by dividing total Q (m3 t−1) or C load for a flow period by watershed area (m2). Water yield is presented as mm water t−1 and carbon yield is presented as mmol C m−2 t−1.

4. Results

4.1. Water Discharge (Q)

[12] Water discharge is highly seasonal in the Yukon basin, with peak Q on the YR and PR occurring mid to late June, tapering through fall and reaching low flow in late winter (Figure 2). Smaller streams and rivers within subbasins may freeze solid in winter, depending on local groundwater conditions. For YR and PR, 34–51% of annual Q occurred during spring. The TR and streams dominated by glacial meltwater and overflow of alpine lakes typically peak during summer, usually in mid to late July (Figure 2) [Brabets et al., 2000]. Average annual Q at YRP during 2001–2005 was 211 km3, within 3% of the mean for 25 years of record. Average annual Q at PR was ∼15% less than the 19 year average, and Q at TR was ∼8.5% greater than the 43-year average (Table 1). There is some indication that annual Q may be trending downward at PR (Sen-Theil-Kendall trend line; P = 0.19) and slightly upward at TR (P = 0.19). Long-term data were insufficient to determine trends in annual Q at YRS or YRP.

Figure 2.

Water discharge hydrographs for the five measurement stations, water years 2001–2005.

4.2. Carbon

4.2.1. Total Carbon Export

[13] An average of 7.8 Tg C yr−1 discharged past YRP during 2001–2005; 30% as organic C and 70% as inorganic C (Table 2). Following Striegl et al. [2005], annual flow was divided into three seasonal flow periods based on hydrology and on the chemical character and concentration of C (Table 3). Total C loads generally follow Q, with peak daily C export occurring during spring, except at TR which peaks during summer (Figure 2 and Table 3).

Table 2. Average Annual DOC, POC, DIC, PIC, and Total C Loads for the Five Measurement Stations for Water Years 2001–2005, With Estimation Error for Each Calculated Loada
Station NameDOC, 106 kg yr−1DOC Error, %POC, 106 kg yr−1POC Error, %DIC, 106 kg yr−1DIC Error, %PIC, 106 kg yr−1PIC Error, %Total C, 106 kg yr−1
  • a

    Water year is 1 October to 30 September.

Yukon River at Eagle4087.431422.716392.345712.92818
Porcupine River near Fort Yukon1248.42710.21874.51813.0356
Yukon River near Stevens Village7816.038210.222963.74339.93892
Tanana River at Nenana8010.9888.75783.87614.3822
Yukon River at Pilot Station15856.07517.249633.150610.97805
Table 3. Seasonal and Annual DOC, POC, DIC, and PIC Loads and 5-year Means for the Five Measurement Stations, 2001–2005a
Station20012002200320042005b5-year Means
  • a

    Spring is 1 May to 30 June; summer–autumn is 1 July to 31 October; winter is 1 November to 30 April. All loads are in 106 kg C.

  • b

    Summer–autumn 2005 uses estimated flow for 1–31 October 2005.

Yukon River at Eagle
Porcupine River near Fort Yukon
Yukon River near Stevens Village
Tanana River at Nenana
Yukon River at Pilot Station

4.2.2. Organic Carbon

[14] Organic carbon comprised 26–42% of total C flux, with highest DOC loads occurring during spring at all stations except TR (Table 3). SUVA, a measure of DOC aromatic carbon content [Weishaar et al., 2003], and DOC concentration were greatest during spring and least during low flow in winter (Table 4). Within the YR, average DOC concentration and SUVA increased from YRE to YRP.

Table 4. Mean Measured DOC, SUVA, DIC, and pCO2, ± Percent Variance for the Five Measurement Stations, 2001–2005a
 Yukon River at EaglePorcupine River Near Fort YukonYukon River Near Stevens VillageTanana River at NenanaYukon River at Pilot Station
  • a

    Errors are in percent variance. The values given in parentheses are the number of samples.

DOC, μmol L−1
   Spring640 ± 10.5 (12)1000 ± 10.7 (12)890 ± 14.0 (11)420 ± 14.7 (13)900 ± 7.0 (11)
   Summer–autumn340 ± 17.3 (16)670 ± 11.0 (17)360 ± 11.9 (17)220 ± 16.3 (18)430 ± 10.0 (19)
   Winter140 ± 3.5 (4)160 ± 3.2 (4)180 ± 6.7 (5)120 ± 2.9 (5)220 ± 2.7 (5)
   Annual460 ± 12.7 (32)760 ± 8.8 (33)510 ± 11.5 (33)280 ± 11.8 (36)560 ± 9.0 (35)
SUVA, L mg C−1 m−1
   Spring3.3 ± 4.2 (12)3.5 ± 1.4 (12)3.4 ± 1.5 (11)3.0 ± 4.7 (13)3.5 ± 2.3 (11)
   Summer–autumn2.8 ± 10.4 (16)3.1 ± 4.8 (17)2.9 ± 4.8 (17)2.6 ± 8.5 (18)3.1 ± 4.2 (19)
   Winter1.9 ± 4.2 (4)2.1 ± 8.1 (4)2.1 ± 6.2 (5)2.0 ± 11.0 (5)2.3 ± 3.5 (5)
   Annual2.9 ± 3.4 (32)3.1 ± 3.2 (33)3.0 ± 3.3 (33)2.7 ± 3.7 (36)3.1 ± 3.2 (35)
DIC, μmol L−1
   Spring1530 ± 3.5 (10)1070 ± 70 (12)1530 ± 6.9 (10)1770 ± 6.3 (13)1480 ± 4.2 (10)
   Summer–autumn1860 ± 3.2 (16)1730 ± 5.9 (17)1900 ± 3.8 (17)1840 ± 4.1 (17)1890 ± 3.0 (19)
   Winter2580 ± 2.1 (7)4270 ± 1.6 (8)2930 ± 2.5 (6)3410 ± 1.1 (6)4100 ± 1.9 (8)
   Annual1910 ± 3.9 (33)2070 ± 10.1 (37)2020 ± 5.1 (33)2080 ± 5.5 (36)2250 ± 7.4 (37)
pCO2, μatm
   Spring1280 ± 20.2 (10)1610 ± 12.1 (12)957 ± 15.0 (10)1390 ± 11.4 (13)1530 ± 9.5 (9)
   Summer–autumn908 ± 11.3 (16)1240 ± 7.8 (17)989 ± 13.3 (17)1580 ± 23.2 (18)1650 ± 24.0 (19)
   Winter1820 ± 4.5 (7)3720 ± 6.0 (8)2340 ± 3.8 (6)3480 ± 6.7 (5)8280 ± 3.6 (8)
   Annual1220 ± 9.1 (33)1890 ± 9.9 (37)1270 ± 10.1 (33)1830 ± 12.4 (36)3090 ± 16.5 (36)

[15] HPOA, TPIA and HPI fractions comprised >79% of the DOC in all samples, with HPOA being the most abundant fraction (Table 5). HPOA was the most aromatic fraction, followed by TPIA and HPI. Like DOC, HPOA concentration and aromaticity were greatest during spring and increased from YRE to YRP. HPI comprised a larger percentage of DOC during winter than during other flow periods. Incubation results indicated that DOC half life (20°C) was 12 ± 6 d for winter samples and 48 ± 20 d for all other samples.

Table 5. Five-Year Average DOC Fractions, as Percent HPOA, TPIA, and HPI of Total DOC Concentration
SeasonDOC FractionYukon River at EaglePorcupine River Near Fort YukonYukon River Near Stevens VillageTanana River at NenanaYukon River at Pilot Station
Springtotal recovery86.486.679.489.080.8
Summer–autumntotal recovery96.184.391.110584.5
Wintertotal recovery10188.410310494.2

[16] The δ13C-DOC values were −27.3, −26.6, −24.1, and −26.4‰ for PR, TR, YRE, and YRP, respectively. The respective FM of 14C-DOC was 1.07, 0.97, 0.94, and 0.96, corresponding to radiocarbon ages of modern, 244, 497, and 327 years before present (B.P.). This indicates the presence of some aged DOC in all the rivers, except the Porcupine.

[17] POC loads were variable and mostly associated with high flows (Table 3). POC was mostly composed of terrestrial plant debris, having δ13C-POC values of −29.9, −25.7, and −26.4 ‰ at YR above WR, YR above PR, and YRP, respectively. FM 14C-POC was 0.70, 0.63, and 0.67 at those locations, with corresponding radiocarbon ages of 2840, 3710, and 3250 years B.P.

4.2.3. Inorganic Carbon

[18] DIC comprised 52–70% of total C flux at all stations (Table 3). With pH ranging from 6.2 to 8.4 for all stations and times, most DIC was [HCO3]. DIC concentrations were greatest during base flow in winter and were diluted by spring runoff (Table 4). Partial pressure of CO2 (pCO2) was generally greater than equilibrium with the atmosphere throughout the year, indicating CO2 flux to the atmosphere during the ice free period. The exception was during the summer of 2001, when subatmospheric pCO2 was measured in the YR [Schuster, 2003]. High [CO2] in ground water, within-river respiration, and lack of atmospheric exchange during winter result in build up of pCO2 under ice (Table 4).

[19] Yukon River DIC concentrations were influenced by inflow of suspended particulate carbonates from the White River [Eberl, 2004]. Peak PIC loads coincided with the glacial meltwater peak, with ∼75% of annual PIC load passing YRE during summer-autumn (Table 3). WR had δ13C-PIC = −0.1‰ and FM 14C-PIC = 0. During 2001, passing of the PIC peak at YRS corresponded with enrichment of δ13C-DIC from −12.7‰ in early June to −5.2‰ in mid-August, depletion of 14C-DIC from FM 0.69 to 0.57, and depletion of [CO2] to below atmospheric equilibrium. Input and dissolution of WR carbonates also had a dramatic effect on δ13C-DIC versus [DIC] along the YR (Figure 3).

Figure 3.

Plot of δ13C-DIC versus DIC in the Yukon River. Unit for δ13C is per mil, relative to Peedee belemnite. YRS is from June–August 2001; all other stations are from June–August 2004. Yukon R. at Carmacks is upstream of the confluence of the Yukon and White rivers.

5. Discussion

5.1. Carbon Cycling

[20] The chemistry and amount of C exported by rivers results from interactions among geologic and atmospheric inorganic C, terrestrial and aquatic organic C, climate, and hydrology. With vast stores of aged OC in permafrost and soils, it is important to understand current sources, ages, and proportions of C species so that future climate change effects can be assessed [Raymond and Bauer, 2001a].

[21] The chemical character and 13C composition of DOC suggest terrestrial origin with little down river biogeochemical alteration. Half-lives of DOC determined by incubations also suggest that most DOC is exported from the basin before it is mineralized. Increased HPI fraction and low DOC concentration in winter are consistent with groundwater input, as hydrophobic DOC is preferentially sorbed on mineral soil [Kaiser and Zech, 1997] and the increased flow path allows for DOC consumption in soil and groundwater [Striegl et al., 2005]. The HPI fraction is relatively labile [Qualls and Haines, 1992]; resulting in shorter DOC half lives during winter.

[22] During August–September 2004, YRP DOC averaged FM 14C-DOC = 0.96, indicating a radiocarbon age of 330 yrs B.P. Previous studies have recorded modern DOC in arctic rivers [Benner et al., 2004]. Although YR samples are not dominated by old DOC, there is an aged component consistent with mixing of a large pool of modern DOC with a small pool of old DOC of unknown origin [Raymond and Bauer, 2001b]. It is possible that this aged DOC component exists in other arctic rivers. In addition to the >3000 years old POC measured by this study, others have also measured aged POC in the Yukon [Guo and MacDonald, 2006] and other high latitude rivers [Goni et al., 2005]. Annual floods erode deeply into YR river banks, dropping trees and other vegetation, roots, and old soil organic matter into the river. The 14C-depleted POC we measured is consistent with significant contribution from these old carbon pools.

[23] PIC load decreased from YRE to YRS, probably due to particle settling and PIC dissolution [Eberl, 2004]. Weathering consumes ∼30% of PIC during transport, has substantial effect on the isotopic composition of DIC (Figure 3), and occasionally consumes [CO2] to below atmospheric equilibrium, but has small effect on the overall DIC mass balance. Given PIC export of 500 × 106 kg C yr−1 (Table 3) and assuming this represents 2/3 of PIC input, weathering only contributes ∼5% of total DIC export at YRP. Mass balance calculations of silicate and nonsilicate weathering at YRP suggest that ∼54% of C-DIC originates from the atmosphere or OC respiration [Blum et al., 1998; Anderson et al., 2000]. Decreased δ13C-DIC with increased [DIC] in YR at Carmacks suggests a respiration source (Figure 3). However, PIC dissolution masks the respiration signal downstream of the White River, and is only marginally perceptible at YRP.

5.2. Watershed Carbon Yield

[24] Carbon yield (mmol C m−2 watershed area t−1) with respect to water yield (mm water t−1) provides valuable comparative information on C source and export from watersheds within a basin. Five-year averages of seasonal C species yields versus water yields at the five stations are listed in Table 6 and shown in Figure 4. Total C yield at YRP averaged 0.73 mol m−2 yr−1 and was highest at TR (1.03 mol m−2 yr−1) where DIC dominated C flux. DIC yield increased linearly with water yield among YR subbasins for all seasons and all stations (r2 = 0.95, n = 75, p < 0.001). Although the slope of DIC versus water yield (Figure 4a) was steepest during winter due to high base flow DIC concentrations, it was generally similar for all seasons throughout the basin.

Figure 4.

Mean seasonal (a) DIC, (b) DOC, (c) PIC, and (d) POC yield versus water yield for the five measurement stations, 2001–2005. Blue triangles, winter; green diamonds, spring; red squares, summer–autumn.

Table 6. Seasonal and Annual DIC, DOC, PIC, POC, and Water Yields for the Five Measurement Stations, 2001–2005
StationDIC, mmol/m2DOC, mmol/m2PIC, mmol/m2POC, mmol/m2Water, mm
Yukon River at Eagle
Porcupine River near Fort Yukon
Yukon River near Stevens Village
Tanana River at Nenana
Yukon River at Pilot Station

[25] DOC chemistry and export group into three distinct seasons at YRP [Striegl et al., 2005] and throughout the basin (Tables 5 and 6 and Figure 4b). Winter DOC concentrations were fairly constant at all stations, averaging 161 ± 35 μmol C L−1 (n = 21). This resulted in a linear relation between DOC and water yields during winter, with DOC yield proportional to areal groundwater discharge. During spring, water yield was fairly constant across the basin, while DOC yield was variable, with highest DOC yield at PR and lowest yield at TR. Much of the land surface remains frozen in early spring and new DOC runs off with melt water. Annual DOC yield was greatest from wetland influenced portions of the basin (PR and YRP) and least from subbasins having a higher proportion of bare rock, ice and snow, and forested uplands (TR and YRE) [Brabets, 2000]. Highest annual DOC yield (0.16 mol C m−2 yr−1) was at YRP (Table 6). This suggests that our fixed station measurements missed at least one major DOC source, most likely the remote Koyukuk River (Figure 1). Although the Koyukuk basin has flats and wetlands similar to the PR basin, it is wetter and likely yields more OC.

[26] In contrast to the spring pattern, DOC yields during summer-autumn were fairly constant across the basin, while water yields varied. At first appearance, this suggests limited DOC sources available for hydrologic export. However, closer inspection suggests that DOC export was water limited, not C limited (Figure 5). Coincident occurrence of low runoff in the PR watershed, where DOC source is relatively high, and high runoff in the TR watershed, where DOC source is relatively low, resulted in similar yields at all stations. The slope of DOC yield versus water yield at PR was about twice as steep as at TR and slopes for the YR stations were intermediate.

Figure 5.

Summer–autumn DOC yields for the five measurement stations, 2001–2005, and for YRP, 1978–1980.

[27] The only historical DOC export data available for the Yukon River were collected at YRP during 1978–1980 [Striegl et al., 2005] and those yields fall more in line with 2001–2005 PR yields than with 2001–2005 YRP yields (Figure 5). There is weak suggestion of an increased trend in summer–autumn Q at TR (Sen-Theil-Kendall trend line, P = 0.16), which could contribute to shifting the YRP DOC versus water yield relation to the right, but no trends in summer–autumn Q were observed at other stations. However, winter Q has increased significantly at PR (P < 0.001), TR (P = 0.004) and YRE (P < 0.001). This increase in base flow supports the argument that change in water flow path, and not change in water quantity, has produced a downward shift in the relation between summer–autumn DOC export and water export from the Yukon basin in recent decades [Striegl et al., 2005].

[28] PIC yield (Figure 4c) was directly related to the export of suspended sediment having high carbonate content from the White River basin [Eberl, 2004]. Therefore fixed station yields were greatest at YRE and decreased downstream to YRS and YRP because of sediment dilution and PIC dissolution. Yields were intermediate at TR, due to high suspended sediment concentration and low carbonate content, and were lowest at PR, which had low suspended sediment concentration and presumably low carbonate content.

[29] POC yields (Figure 4d) reflected their primary source from bank erosion and generally increased with water yield. Yields were greatest in the upper Yukon basin during spring, reflecting the tremendous bank scour by ice that occurs during spring melt. POC yields were smallest at PR and were negligible during winter at all stations.

5.3. Carbon and Climate Change

[30] Net primary production (NPP) in the Yukon River basin is currently about 110 Tg C yr−1 (10.8 mol C m−2 yr−1 (D. Kicklighter, unpublished data, 2006)). About 6% of this is exported from the basin annually as DOC and DIC. NPP is expected to change in response to climate warming and interactions among ecology, hydrology, and C export can also be expected to change [Serreze et al., 2000; Hinzman et al., 2005; Arctic Climate Impact Assessment,]. Thawing of permafrost will initially result in higher DOC concentrations in melt water [Michaelson et al., 1998; Kawahigashi et al., 2004] and possibly in increased DOC yield [Frey and Smith, 2005]. However, as flow paths deepen and more DOC is consumed in soil and groundwater, DOC yield may decrease coincident with increased respiration and DIC yield [Striegl et al., 2005]. Carbon yield is proportional to water yield in all settings, therefore understanding of change in C export in response to climate change will be directly linked to understanding of change in hydrology from the small watershed to the basin scale.


[31] Data collection and laboratory analyses were a team effort by many scientists and volunteers. We thank all of you for your contributions. The 13C-DIC analyses were conducted by the Stable Isotope Laboratory, Department of Oceanography, Florida State University. The 14C samples were processed at Yale University and analyzed at NOSAMS or University of Arizona. N.-H. Oh contributed the ion balance calculation, and M. Walvoord conducted the hydrologic trends calculations. D. Clow, T. Brabets, R. Smith, and J. Herman reviewed the draft manuscript and contributed to its improvement.