Chemical characteristics of fulvic acids from Arctic surface waters: Microbial contributions and photochemical transformations



[1] Dissolved organic matter (DOM) originating from the extensive Arctic tundra is an important source of organic material to the Arctic Ocean. Chemical characteristics of whole water dissolved organic matter (DOM) and the fulvic acid fraction of DOM were studied from nine surface waters in the Arctic region of Alaska to gain insight into the extent of microbial and photochemical transformation of this DOM. All the fulvic acids had a strong terrestrial/higher plant signature, with uniformly depleted δ13C values of −28‰, and low fluorescence indices around 1.3. Several of the measured chemical characteristics of the Arctic fulvic acids were related to water residence time, a measure of environmental exposure to sunlight and microbial activity. For example, fulvic acids from Arctic streams had higher aromatic contents, higher specific absorbance values, lower nitrogen content, lower amino acid–like fluorescence and were more depleted in δ15N relative to fulvic acids isolated from lake and coastal surface waters. The differences in the nitrogen signature between the lake and coastal fulvic acids compared to the stream fulvic acids indicated that microbial contributions to the fulvic acid pool increased with increasing water residence time. The photo-lability of the fulvic acids was positively correlated with water residence time, suggesting that the fulvic acids isolated from source waters with larger water residence times (i.e., lakes and coastal waters) have experienced greater photochemical degradation than the stream fulvic acids. In addition, many of the initial differences in fulvic acid chemical characteristics across the gradient of water residence times were consistent with changes observed in fulvic acid photolysis experiments. Taken together, results from this study suggest that photochemical processes predominantly control the chemical character of fulvic acids in Arctic surface waters. Our findings show that hydrologic transport in addition to biogeochemical alteration of the organic matter must be considered in order to predict the ultimate fate of Arctic DOM.

1. Introduction

[2] The Arctic Ocean receives about 10% of the global riverine discharge annually. This river discharge delivers approximately 25 Tg of dissolved organic carbon (DOC) per year to the Arctic Ocean [Aagaard and Carmack, 1989; Opsahl et al., 1999]. Thus, on a volume basis, the Arctic Ocean receives the largest load of terrestrially derived organic matter of any of the world's oceans. The fate of this organic matter is an important piece of the global carbon cycle. However, the ultimate fate of terrestrial dissolved organic matter (DOM) in the ocean (e.g., carbon dioxide versus bacterial biomass) has been shown to depend on the extent of chemical ‘weathering’ of DOM as it is transported from rivers to oceans [Smith and Benner, 2005].

[3] The relative importance of photochemical and microbial “weathering” processes is not currently resolved for the Arctic system [Amon and Meon, 2004; Benner et al., 2005]. It is possible that photochemical processes may play a large role in shaping Arctic DOM character because snowmelt and delivery of fresh, terrestrial DOM into surface waters occurs at the onset of summer, a 3-month period with constant sunlight. Relative to microbial processes, photochemical processes are not expected to be as strongly temperature-dependent over the range that occurs in surface waters. Cool temperatures and low primary productivity combined with 24 hours of sunlight during the summer season may increase the importance of photochemical alteration of DOM in Arctic surface waters compared to temperate regions. Both photochemical and microbial weathering processes would be expected to impart distinct chemical signatures. For example, photochemical degradation of DOM has been shown to cause a decrease in aromaticity [Brooks et al., 2007].

[4] Another consequence of low rates of primary production in Arctic surface waters is that the terrestrial-aquatic linkage driven by the flux of allochthonous DOM may play a more important role in downstream aquatic food webs in the Arctic compared to temperate zones [O'Brien et al., 1997]. Secondary bacterial production has been estimated to be as high as 66% of primary production in Arctic lakes; reflecting the use by microorganisms of terrestrially derived organic carbon flushed into the lakes during snowmelt [Crump et al., 2003]. The effects of chemically distinct pools of DOM on bacterial community shifts and carbon fluxes in Arctic lakes have been well studied [Crump et al., 2003; Judd and Kling, 2002; Michaelson et al., 1998], but little is known about DOM chemical composition. In the latter studies, differences in DOM chemical character have been inferred by observed seasonal shifts in lake bacterial production and community [Crump et al., 2003; Judd and Kling, 2002].

[5] Differences in DOM character may be evident in the fulvic acid fraction of the DOM, which represents a major DOM fraction and is the dominant light-absorbing fraction in natural waters [McKnight and Aiken, 1998]. For DOM in general, fulvic acids are derived from two classes of precursor organic material, i.e., decomposed plant material and soils of terrestrial (allochthonous) origin and organic material produced by algae and bacteria (autochthonous). Chemical properties of fulvic acids isolated from diverse environments fall on a continuum between the International Humic Substance Society (IHSS) reference fulvic acid from the Suwannee River in Georgia, representing a terrestrial end-member, and fulvic acids isolated from Lake Fryxell and Pony Lake in Antarctica, representing microbial end-members [McKnight et al., 2001; Schwede-Thomas et al., 2005]. For example, the Suwannee River fulvic acid has high aromatic carbon content and is low in nitrogen, reflective of its higher plant precursor material such as lignin. In contrast, Lake Fryxell fulvic acid contains roughly six times the nitrogen of Suwannee River fulvic acid and is significantly less aromatic; consistent with its derivation from microbial organic matter.

[6] Controls on fulvic acid concentration and chemical properties include the relative contribution of different organic precursor materials, photochemical and microbial alteration, and sorption onto mineral surfaces. In a study of Arctic soil and surface waters during spring thaw, Michaelson et al. [1998] found that the fulvic acid component comprised a greater fraction of the DOM in the soil and stream waters than in the lake water compared to the amount of low molecular weight organic acids and neutrals which comprise the nonfulvic fraction of DOM. These results are consistent with the general trend that higher fulvic acid content is characteristic of DOM derived from terrestrial (soil and higher plant) organic matter while lower fulvic acid content is characteristic of autochthonous DOM [McKnight and Aiken, 1998]. The results of Michaelson et al. [1998] also suggest that changes in the character of aquatic fulvic acids in the Arctic may vary with water residence time, as terrestrially derived organic matter from the tundra is increasingly exposed to photochemical and microbial processing in streams, lakes and coastal waters.

[7] Photochemical and microbial processes oxidize terrestrially derived fulvic acid to produce carbon dioxide [Moran et al., 2000; Xie et al., 2004]. The net result is that reduced carbon of terrestrial origin is either oxidized fully to carbon dioxide or partially to unknown byproducts. The “leftover” fulvic acid pool, which contains the byproducts of photochemical and microbial oxidation as well as byproducts of microbial metabolism, provides clues about the extent of these processes in the environment. Because aromatic carbon containing molecules account for the light-absorbing properties of DOM [Weishaar et al., 2003], the breakdown of aromatic carbon moieties in the fulvic acid fraction is expected to result in loss of DOM absorbance and fluorescence, which has been extensively documented for DOM photodegradation [Ma and Green, 2004; Molot et al., 2005; Moran et al., 2000]. Whereas an increasing autochthonous (microbial) contribution to a terrestrially derived fulvic acid would likely increase nitrogen content because of the higher nitrogen content of microbially derived organic matter compared to the nitrogen-poor lignaceous precursor material of terrestrial DOM [McKnight and Aiken, 1998; Brown et al., 2004; Fulton et al., 2004]. Further, specific shifts in fulvic acid fluorescence have been associated with microbial contributions to the fulvic acid pool [McKnight et al., 2001].

[8] Thus the goal of this study was to determine the dominant chemical changes in Arctic whole water DOM and dissolved fulvic acids occurring in freshwaters and coastal zones and to evaluate the influences of photochemical degradation and microbial contributions on these chemical changes. Our approach was to examine the chemical characteristics of DOM in Arctic streams, lakes and coastal waters that would be expected to change upon microbial production or photochemical degradation. We isolated fulvic acids from Arctic streams and lakes in the vicinity of the Toolik Lake Long Term Ecological Network Station (Arctic LTER) and from coastal surface waters in Prudhoe Bay. Although these samples do not correspond necessarily to samples along a single hydrologic flowpath, they represent samples with varying degrees of environmental exposure reflected in the differing water residence times previously determined through studies of the Arctic LTER. Photodegradation studies of the whole water DOM and isolated fulvic acids were conducted to gain insight on photochemical shifts in chemical characteristics.

2. Site Description

[9] Sample collection and processing occurred at the Arctic LTER site, which is based at University of Alaska's Toolik Field Station located in the northern foothills of the Brooks Range Alaska (68°38′N, 149°43′W, elevation 760 m). A complete site description can be found at and is given by Kling et al. [2000]. The area around Toolik Field Station is characterized by continuous permafrost, 24 hours of sunlight during the summer growing season (May–August), and periods of limited daylight to complete darkness during the winter (September–April). Ice-out and snowmelt generally occur in May or June. Tussock tundra is the dominant vegetation, and areas of wet sedge tundra, drier heath tundra on ridge tops and other well-drained sites with river-bottom willow communities are found in the watershed (

[10] Seven preparative-scale samples were collected from lakes and streams within 20 km of the Toolik Field Station during the summers of 2002 and 2003 (Table 1). The average Secchi depth of lakes around Toolik Field Station is ∼4 m [Kling et al., 2000]. The lakes samples included Toolik Lake (TL), Island Lake (IL), and Campsite Lake (CL). Toolik Lake is the reference lake for the Arctic LTER and is typical of the lakes in the region: it is a deep (25 m maximum), oligotrophic kettle lake, and is ice free from about mid-June through September. Despite its northern location, Toolik Lake is dimictic. Other features of Toolik Lake include its several inlet streams and one outlet stream, and a water residence time estimated to be approximately 1 a (365 days [O'Brien et al., 1997]). Campsite Lake and Island Lake are both smaller and shallower than Toolik Lake and lack the coloration of Toolik Lake water. Following the approach of Kling et al. [2000], we estimated minimum water residence times for Campsite and Island Lake by comparing the ratio of the lake volume to the drainage area to that of Toolik Lake. This calculation yielded water residence times for Campsite Lake and Island Lake of 2404 and 885 days, respectively (Table 1). Toolik and Island Lakes were ice-free when sampled in mid-June 2002, several weeks after snowmelt and ice-out. Campsite Lake was sampled in June 2003, also several weeks after snowmelt, but was not completely ice-free.

Table 1. Site and Sample Descriptions
SiteAbbreviationTypeDescriptionEstimated Water Residence TimeSample Date, m/dd/yySample Volume, L
Oks SeepOksstreamdark water, shallow seep2 daysa6/30/0260
Imnaviat RiverIMstreamdark water first-order stream5 daysa6/23/02295
TWL CreekTWLstreamsmall first-order inlet stream to Toolik Lake, surrounded and by tussock tundra, moist acidic soils2 daysa6/30/02298
Toolik Inlet StreamTIstreamsecond-order inlet stream to Toolik Lake, water flows through series of lakes.∼ 0 daysa5/30/03360
Toolik LakeTLlake150 Ha, deep (25 m), oligotrophic kettle lake, dimictic1 ab6/19/02604
Island LakeILlakeoligotrophic kettle lake, shallower and clearer than Toolik Lake2.4 aa6/24/02596
Campsite LakeCLlakeoligotrophic kettle lake, smaller, shallower and clearer than Toolik Lake6.6 aa6/23/03158
Prudhoe BayPBcoastal oceancoastal ocean clear water, mostly ice-free at collection11–15 ac7/03/02139
Oliktok Pt.OPcoastal oceancoastal ocean site, ice covered at collection11–15 ac6/21/03694

[11] The stream samples included Toolik Inlet Stream (TI), Imnavait River (IM), Tussock Watershed Lower Creek (TWL), and a seep near Oksrukuyik Creek (Oks). These streams are typical of Arctic tundra streams in that they contain clear, darkly colored water with low nutrient concentrations [Kling et al., 2000]. TI Stream and TWL Creek drain into Toolik Lake. Because TI Stream was sampled at the onset of snowmelt and ice-out on 30 May 2003, we estimated that the water residence time for the TI sample to be less than a day (Table 1), even though under late summer conditions TI Stream primarily carries drainage from upstream lakes. Other streams were sampled several weeks after snowmelt in June 2002. TWL Creek is a small, narrow first order stream draining a 1.5 Ha area and is shaded by tussock tundra [Crump et al., 2003]. The Imnavait River is a larger first-order stream that is generally unshaded by vegetation. The Oks seep is an area of pooled water near the Oksrukuyik Creek, 20 km from the station. The water residence times for Oks Seep, IM River and TWL Creek were estimated to be on the order of two to five days following the approach of Kling et al. [2000, Table 1].

[12] Two preparative-scale marine samples were collected along the coast of the Arctic Ocean (latitude 70°01′N) near Prudhoe Bay (PB) and Oliktok Point (OP), 210 km north of Toolik Field Station. The PB sample was collected 3 July 2002 at an ice-free site with a 1 m depth. The OP sample was collected from a depth of 10 m on 21 June 2003 at a ice-covered site. Both coastal locations are within 50 km of the mouths the Kuparuk and Sagavanirktok Rivers, and within 100 km of the Colville River. During spring runoff, these three Arctic rivers bring more than one third of the annual load of dissolved organic carbon (DOC) to the coastal Beaufort Sea [Rember and Trefry, 2004]. It has been estimated that the average water residence time of river water in the surface Arctic Ocean is about 11–15 a [Bauch et al., 1995].

3. Methods

3.1. Whole Water Sample Collection and Fulvic Acid Isolation

[13] For this study, whole water DOM refers to the bulk DOM in a filtered water sample and the fulvic acid fraction of the DOM refers to hydrophobic acid fraction of DOM isolated by XAD-8 chromatography. Isolation of fulvic acid enables further characterization of an abundant DOM fraction using instrumental methods not applicable to whole water DOM, owing to dilute concentrations, interfering species, and/or requirements for a solid (powder) form of the sample. Samples were collected in 20-L plastic cubitainers and processed at the Toolik Lake Field Station. Initial volumes ranged from 60 to 694 L (Table 1). Samples were filtered through 0.25-μm and 0.45-μm glass fiber filters housed in a series of stainless steel filter towers (Balston Filters, Parker-Hannefin) and acidified to pH 2 with concentrated HCl (trace metal grade). Aliquots of the filtered, unacidified water were saved for DOC, fluorescence and UV-vis absorbance analyses. Fulvic acids were isolated using XAD-8 chromatography [Thurman and Malcolm, 1981]. Following reconcentration and desalting, fulvic acid solutions were shipped to Boulder, CO and lyophilized immediately.

3.2. DOM and Fulvic Acid Characterization

[14] The chemical characteristics of the whole water DOM and fulvic acids were examined to gain insight into the magnitude and nature of photochemical and microbial alterations of Arctic DOM originally flushed into surface waters from the tundra during snowmelt. The bulk characteristics examined (e.g., specific absorbance, 13C-NMR and δ13C) provide information on the average properties of organic molecules in the whole water or fulvic acid fraction. For the fulvic acid fraction, bulk characteristics generally pertain to the chemical form of the organic carbon, because carbon accounts for approximately 50% of the organic matter by mass. In contrast, the trace moieties examined correspond to elements (e.g., N and S) or functional groups (e.g., quinones and amino acids as determined by fluorescence spectroscopy) that are not present in every organic molecule within the fulvic acid fraction. Trace moieties can provide information on particular reactive functional groups and dominant biogeochemical processes [McKnight et al., 2003].

[15] Solid state 13C-NMR spectra were obtained to analyze fulvic acid carbon functional groups using a ramp cross-polarization magic angle spinning (CPMAS) pulse program and two pulse modulated decoupling on a Bruker DSX 300 NMR spectrometer, operating at a frequency of 75.48 MHz for 13C-NMR [Dria et al., 2002]. Functional groups were assigned following [Dria et al., 2002]. Practical constraints of sample quantity and instrument time prevented obtaining duplicate NMR spectra; however Dria et al. [2002] concluded that errors in intensity measurements under these analysis conditions were approximately five percent of the total intensity in the spectrum.

[16] Analyses for UV-Vis absorbance and fluorescence were conducted with 1-cm path length quartz cuvettes on an Agilent 5000 Spectrophotometer (Agilent) and Fluoromax-2 or Fluoromax-3 fluorometer (Jobin-Yvon Horiba), respectively. Fulvic acid solutions for spectroscopy were prepared by dissolving the lyophilized material in MilliQ water in glass amber bottles, stirring for 24 hours, and adjusting to pH 6–7 using 0.1 N HCl or NaOH. Specific UV absorbance values at 254 nm (SUVA254) were calculated for each sample by normalizing the absorption coefficients at 254 nm to the dissolved organic carbon (DOC) concentration. SUVA254 values provide a measure of the amount of chromophoric carbon [Weishaar et al., 2003].

[17] Excitation-emission matrices (EEMs) for whole water samples and fulvic acid solutions were collected with excitation range of 240–400 nm, emission 350–550 nm in reference beam mode, which corrects for first-order variation in the xenon lamp output. EEMs of MilliQ water were subtracted to remove Raman scattering and each EEM was then corrected for the wavelength-dependent contribution that instrumental components have on the measured signal using the emission and excitation correction files provided by the manufacturer [Cory and McKnight, 2005]. Intensities were corrected for the inner filter effect [McKnight et al., 2001] and converted to Raman units [Stedmon et al., 2003]. Excitation was incremented by 10 nm and emission by 2 nm.

[18] The fluorescence index (FI), which is an indicator of the contribution of microbial versus terrestrially derived precursor material to the fulvic acid pool of the DOM [McKnight et al., 2001], was obtained from the corrected EEM after verifying the presence of the FI peak between 460–480 nm at an excitation wavelength of 370 nm. The FI was calculated as the ratio of emission intensities at 470/520 nm [Cory and McKnight, 2005]. A FI of about 1.30 is typical for an allochthonous fulvic acid from the Suwannee River, while a microbially derived fulvic acid from Lake Fryxell has an FI around 1.80 [McKnight et al., 2001]. For comparison, we measured FI values of 1.25 and 1.78 for Suwannee River and Lake Fryxell fulvic acids, respectively. The standard deviation of the FI values for samples run in triplicate was ±0.01.

[19] Parallel factor analysis (PARAFAC) [Stedmon et al., 2003] was used to model the EEMs employing a “universal” model based upon 379 EEMs [Cory and McKnight, 2005]. PARAFAC is a statistical approach that identifies the fluorescing components having the greatest influence on a data set of EEMs without making assumptions on the shapes of the excitation and emission curves. The 13 components of the Cory and McKnight [2005] model represent groups of fulvic acid moieties described by their characteristic fluorescence spectra. Some of the 13 components are associated with known classes of organic compounds, for example, quinones and specific amino acids, owing to their similarity with the fluorescence spectra of these compounds (Table 2 [Cory and McKnight, 2005]). For example, components such as SQ1 and SQ2 have secondary absorbance peaks at 350–380 nm and appear to correspond to semiquinone fluorophores. The emission peak at higher wavelengths is broader for SQ1 than for SQ2, which is likely to reflect higher degrees of conjugation for quinones derived from lignins and plant pigments and supports the linkage of these components to terrestrial and microbial precursor organic matter (Table 2 [Cory and McKnight, 2005]). Thus the PARAFAC model provides chemical information on quinone-like fluorophores and the amino acid–like fluorophores and the relative importance of microbial versus terrestrial organic matter [Cory and McKnight, 2005; Stedmon et al., 2003; Yamashita and Tanoue, 2003]. The associations of each of the 13 components are listed in Table 2.

Table 2. Distribution of PARAFAC Components in the Arctic Fulvic Acids Shown as Percent Contribution of Each Component to the Total Modeled EEM
PARAFAC ComponentMolecular AssociationaSourcebStreamsLakesCoastal WatersStream MeancLake/Coastal Meanc
  • a

    Components are labeled and identified according to Cory and McKnight [2005].

  • b

    M denotes components associated with microbially derived organic matter; T denotes components associated with terrestrially derived organic matter [Cory and McKnight, 2005].

  • c

    Letters indicate a significant difference between the mean values for the stream and lake/coastal sites (t-test; p < 0.05 level).

C2quinone-like: Q2M2319191717211818151918
C4quinone-like: HQBoth19201917181716171619A17B
C5quinone-like: SQ1T6867356557A5B
C7quinone-like: SQ2M64443444344
C8amino acid: TrpM10021201311
C9quinone-like: SQ3M00010111201
C11quinone-like: Q1T1416141316151514171416
C12quinone-like: Q3M52549912964A9B
C13amino acid: TyrM1121334321A3B
C8 + C13Combined Amino AcidsM2123454452A4B

3.3. Elemental and Isotopic Analyses

[20] Dissolved organic carbon (DOC) concentrations were analyzed on a Shimadzu TOC 5000 analyzer as nonpurgeable organic carbon (NPOC) after acidification with phosphoric acid. Potassium-hydrogen phthalate solutions were used as standards for the DOC analyzer. Standard deviation in DOC concentrations for samples and standards analyzed in triplicate was ±0.5 mg-C L−1.

[21] Lyophilized fulvic acid samples were analyzed for CHN, O, S and ash content on an elemental analyzer by Huffman Laboratories (Golden, Colorado [Huffman and Stuber, 1985]). Elemental analysis data, including molar ratios, were calculated after correcting for ash content. Ash content ranged from 0.5 to 4% of the lyophilized sample mass. Owing to limited sample size, only one fulvic acid sample (TL) was analyzed in duplicate; the resulting standard deviations for C, H, N, O, and S contents were 0.08%, 0.04%, 0.01%, 0.01%, 0.04% respectively. The standard deviation for the C:N ratio was ±1.0.

[22] The stable isotopes δ13C and δ15N can provide information about the bulk (carbon) and trace (nitrogen) moieties in fulvic acids and can be indicative of the source of precursor organic matter [Amon and Meon, 2004; Hood et al., 2005; McKnight et al., 2003]. Fulvic acid samples were also analyzed for δ15N and δ13C by the Colorado Plateau Stable Isotope Laboratory in Flagstaff, Arizona. Samples were analyzed in continuous-flow mode using a Thermo-Finnigan Deltaplus Advantage gas isotope-ratio mass spectrometer interfaced with a Costech Analytical ECS4010 elemental analyzer. Helium flow rate was set at 110–130 mL/min. Oxygen flow rate is at 80 mL/min. A standard 3-meter GC column was used (set at 55°C) for peak separation, in combination with one quartz (combustion) tube filled with chromium oxide and silvered cobaltous/cobaltic oxide (set at 1020°C) and one quartz (reduction) tube filled with reduced copper (set at 650°C). Further details of the methods, standards used, and detection limits can be found at Duplicate analyses of the Oliktok Point fulvic acid sample differed by 0.02‰ for δ13C and 0.13‰ for δ15N.

3.4. Estimation of Microbial Contributions

[23] A simple mixing calculation (equation (1)) was used to estimate the preservation of nitrogen and to constrain the input of microbially derived nitrogen to the fulvic acid fraction.

equation image

Equation (1) estimates the “X” fraction of stream δ15N that should remain in the Arctic Ocean assuming that the dominant oceanic input comes from plankton having a δ15N signature of 7‰ [Schubert et al., 2001]. The mean stream δ15N value (−0.39‰; Table 2) was used to represent the Arctic ‘terrestrial’ signature and the Prudhoe Bay value of 1.7‰ was used to represent the coastal end-member value in equation (1). To estimate the corresponding difference in carbon content, we assumed a C:N ratio of 11 for a microbially derived fulvic acid [Brown et al., 2004].

3.5. Photodegradation Experiments

[24] Three whole water samples (Island Lake, Toolik Lake, and TWL stream) were exposed to natural sunlight for 12 hours at Toolik Lake Station in June 2002. The light-exposed and dark control water samples were placed in quartz cuvettes (1-cm path length) in air near Toolik Lake. The dark controls samples were wrapped in aluminum foil. Subsamples were collected after 2, 4, 6, and 12 hours of exposure, and fluorescence (but not DOC and absorbance) was monitored.

[25] Laboratory photodegradation experiments were conducted on seven of the nine fulvic acid samples (streams: TI, Oks, IM; lakes: TL, IL, CL; and coastal: PB). Fulvic acid solutions were prepared as described above and concentrations ranged from 10.4 to 12.9 mg-C L−1. The fulvic acid solutions were irradiated with an Oriel solar simulator (model number 91293; Oriel Instruments) equipped with a 1 kW ozone-free Xenon lamp. The solar simulator was set up as a batch reactor with a 250-mL water-jacketed glass beaker on a stir plate directly below the light source. The path length of the beaker was approximately 2.5 cm, as measured by the ferri-oxalate actinometer [Leifer, 1988]. The temperature was kept between 13°–18°C by circulating cold water. These temperatures are typical for surface waters in the region in summer ( Dark controls were covered in foil and kept on a stir plate in a cold water bath.

[26] Subsamples for spectroscopic analysis were collected at various time points for up to 100 min and were analyzed within 2 hours. Absorbance data for two of the seven experiments were not obtained owing to instrument failure. Subsamples for DOC analysis were collected from the initial and final samples. Dissolved oxygen (DO) concentration was measured before and after irradiation and was constant over that interval at 8 mg L−1. The intensity of the light source was monitored with a UVA+UVB radiometer (Solar Light Co. model number PNA 100) and varied from 22 to 25 mW cm−2 over the course of the experiments.

[27] The daily variation in natural sunlight UVB output at Toolik Lake Field Station was monitored by radiometer measurements for UVB from 19 to 27 June 2004 every 15 min. Maximum UVB output occurred between 11:30 and 3 pm with values ranging from 1.5 to 0.96 mW cm−2. The laboratory solar simulator had a UVB output of 2.2 mW cm−2. The ratio of the solar simulator UVB output to midday natural sunlight UVB output at the station ranged from 1.5 to 2.3. Therefore, considering only the UVB output, 1 hour in the solar simulator is approximately equivalent to 1.5 to 2.3 hours of midday sunlight at Toolik Field Station. Because the output of UVA light at the station was not measured, the assumption of equivalent Toolik natural sunlight time does not consider the energy input from UVA light, and the equivalent Toolik sunlight time may be underestimated. Although the maximum quantum yields for many photochemical reactions occur in the UVB wavelength range [Moran and Zepp, 1997], UVB light has been found to be a minor contribution to the total photobleaching because the total energy output from the UVB is at least 10 times less than the energy output of the visible region [Molot et al., 2005; Reche et al., 2000].

[28] The effects of light on the fulvic acid optical properties (e.g., photobleaching) were quantified in two ways: loss of absorbance and fluorescence as a function of light-exposure time. To quantify the loss of absorbance for each fulvic acid, absorbance values at each wavelength (240–600 nm) were first converted to absorption coefficients (aFA, m−1) according to

equation image

where A is the measured absorbance and L is the cuvette length in meters. Next, photobleaching; Δa(m−1) was quantified as a function of wavelength by

equation image

Finally, to quantitatively compare loss of absorbance at a given wavelength among samples with different initial absorbance properties, photobleaching values (Δa; equation (2)) for each sample were normalized to the amount of light absorbed at each wavelength (Qa,λ) calculated by equation (4) [Hu et al., 2002].

equation image

where E(0) is the irradiance of the incident light at the surface of the reaction vessel, at is the total light absorption of the sample (absorption by fulvic acid and water [Smith and Baker, 1981]), S is the cross section of the reaction vessel (0.0056 m2), L is the path length (0.025 m) and t is the exposure time in seconds. Here aFAgm is the geometric mean of fulvic acid absorption at the start and end of the time period examined and was used to estimate fulvic acid absorption between two sampling times [Miller and Zepp, 1995]. In summary, photobleaching values for each fulvic acid sample obtained by equation (2) were normalized to the moles of photons absorbed according to equation (4) to give wavelength specific photobleaching per mole of photons absorbed.

[29] The loss of fluorescence as a function of light exposure time for each sample was quantified by calculating the magnitude of change for each of the 13 groups of fluorophores. The magnitude of change for each fluorophores was calculated as percent loss according to equation (5). The percent loss of fluorescence per fluorophore group was not normalized to light absorbed because it is unknown what fraction of light absorption results in emission for each PARAFAC component in each fulvic acid sample.

equation image

where Fmax represents the relative concentration of a PARAFAC component in Raman units [Stedmon et al., 2003] at the start (Fmax1) and end (Fmax2) of the experiment.

4. Results

4.1. Chemical Characteristics of Whole Water DOM

4.1.1. Bulk Characteristics

[30] The proportion of the DOM isolated from the whole water samples as fulvic acid ranged from 22 to 47% (Table 3). The IM River, a highly colored stream, had the highest percent fulvic acid and Island Lake, a clear, shallow lake had the lowest percent fulvic acid. Generally, the stream waters containing higher DOC concentrations had higher fulvic acid contents than the lakes and coastal ocean sites, except for Campsite Lake (46% fulvic acid), which is clear and shallow (similar to Island Lake).

Table 3. Characteristics of Whole Water and the Fulvic Acid Fraction of DOMa
 StreamsLakesCoastal WatersStream MeanaLake/Coastal Meana
  • a

    Letters indicate a significant difference between the mean values for the stream and lake/coastal sites (t-test; p < 0.05) level.

  • b

    SUVA254 denotes specific absorbance at 254 nm.

  • c

    FI denotes fluorescence index (Ex = 370; Em 470/520 nm).

  • d

    Percent amounts are given for amino acid–like fluorophores tryptophan (Trp) and tyrosine (Tyr); see Table 2.

  • e

    All elemental data are presented on an ash-free basis.

  • f

    Values are molar ratios.

  • g

    Peak area is given as a percentage of total spectrum area of quantitative 13C-NMR analysis.

  • h

    Ar/Al-1 denotes ratio of aromatic (Ar) to aliphatic (Al-1) carbon.

Whole Water DOM
DOC, mg-C L−114.512.
Fulvic Acid, %41473839342246292541A31B
SUVA254,c m2 g−18.87.688.55.35.964.5NA8.3 A5.4 B
%C8d(Trp)00not measured21222112
%C13d(Tyr)32not measured2285642A5B
Fulvic Acid Fraction
SUVA254,cm2 g−110.610.49.598.
%Aliphatic (Al-1, 0–60 ppm)g33.934.844.632.544.645.643.741.940.636A43B
%Carbohydrate (Al-II, 60–90 ppm)g16.114.913.518.813.512.616.715.716.91615
%Anomeric (90–110 ppm)g7.
%Aromatic (Ar, 110–160 ppm)g2223.318.223.617.716.417.516.919.922A18B
%Carboxyl (160–190 ppm)g16.416.114.213.815.216.313.316.213.81515
%Aldehyde/ketone (190–230 ppm)g4.

[31] The whole water samples from the streams had higher DOC concentrations and specific UV absorbance values (SUVA254) compared to the samples from the lakes and coastal sites (Figure 1 and Table 3). There is a strong relationship between DOC concentration and SUVA254 for the whole water samples (r2 = 0.7; regression not shown).

Figure 1.

Chemical characteristics of whole water (x axis) and fulvic acid fraction (y axis). Dotted line is 1:1 line. Slope p value (95% confidence level) is shown on graph. (a) SUVA254 (m2 g-C−1). (b) FI (fluorescence index, Ex = 370, Em 470/520 nm). (c) Amino acid fluorescence: percent contribution from tyrosine and tryptophan-like fluorophores (%). (d) Percent contribution from SQ1 fluorophores (%).

4.1.2. Trace Moieties

[32] The fluorescent fraction of DOM and fulvic acid in particular is currently understood as measuring trace moieties (e.g., quinones and fluorescent amino acids). The whole water DOM fluorescence index (FI) values were uniformly low, ranging from 1.35 to 1.40, indicative of input of plant-derived precursor material (Figure 1 and Table 3). There was no significant difference between the stream, lake and coastal mean FI values for the whole water samples (Table 3). However, significant differences were observed in the distribution of the 13 fluorophores identified by PARAFAC in the whole water DOM between the stream, lake and coastal samples (full data set not shown). For example, one key distinction was the percentage of the amino acid–like fluorophores identified as tryptophan and tyrosine-like. The stream mean value of the combined percent amount of the amino acid fluorophores was significantly lower than the lake and coastal mean (Figure 1 and Table 3).

4.2. Chemical Characteristics of the Fulvic Acid Fraction

4.2.1. Bulk Characteristics

[33] The fulvic acid samples had between 50 and 53% carbon by mass (Table 3). There were no significant differences in percent carbon among the stream, lakes and coastal samples. The stream fulvic acids had a significantly higher oxygen content compared to the lake/coastal fulvic acids, and the difference between the sample means was 2% (p < 0.05). The stream fulvic acids had significantly lower H:C ratios relative to the lake/coastal fulvic acids (p < 0.05; Table 3). There were no significant differences among water bodies for the O:C ratios with the Oks Seep and PB coastal fulvic acids having the highest ratios, while the OP coastal fulvic acid had the lowest value.

[34] All the Arctic fulvic acids had δ13C values of about −28‰ (Table 3) indicating that the organic matter precursors are derived predominantly from terrestrial plants [McKnight et al., 2003]. There was no significant difference in δ13C values across the wide range of sampled water bodies (Table 3).

[35] The aromatic carbon content and aromatic to aliphatic carbon ratio (Ar/Al; Table 3) was significantly higher (p < 0.05) for the stream fulvic acids compared to the lake and coastal fulvic acids. Additionally, IM River, Oks Seep and TI Stream fulvic acids had greater anomeric carbon and less aliphatic carbon compared to the lake/coastal fulvic acids (Table 3), while no clear differences were observed between the stream and lake/coastal fulvic acids for the other moieties, e.g., carboxyl, and ketone/aldehyde groups. Although the TWL Creek fulvic acid was included in the stream group for statistical analysis, this sample had a consistently different 13C-NMR spectrum than the other stream fulvic acids (Table 3).

[36] SUVA254 values for the fulvic acids were uniformly higher than the respective whole water values of the same sample (Figure 1), consistent with the strong light-absorbing properties of the fulvic acid fraction of the DOM relative to the nonfulvic acid fraction [McKnight and Aiken, 1998]. The stream fulvic acids had higher SUVA254 values compared to the lake and coastal fulvic acids, consistent with the greater aromatic content of the stream fulvic acids (Table 3). Indeed, there was a positive correlation between SUVA254 values and aromatic carbon content of the fulvic acids (r2 = 0.6; data not shown) as found previously for fulvic acids from different sites worldwide [Chin et al., 1994; McKnight et al., 1997; Weishaar et al., 2003].

4.2.2. Trace Moieties

[37] The Arctic fulvic acid FI values were consistently lower than the whole water values (Figure 1), ranging from 1.17 for the Oks Stream and 1.35 for the OP coastal ocean fulvic acid (Table 3 and Figure 1). Terrestrially derived fulvic acids have previously been shown to have lower FI values compared to their respective whole water sample likely due to the higher FI of the nonfulvic acid fraction in the whole water DOM [Schwede-Thomas et al., 2005]. There was no significant difference in FI values between the Arctic fulvic acids from streams sites and lake/coastal sites (Table 3). The uniformly low fulvic acid FI values indicate that the predominant source of the Arctic fulvic acids is terrestrially derived organic matter.

[38] Significant differences among the stream, lake and coastal fulvic acids were identified for some of the PARAFAC-identified fluorophores associated with terrestrial and microbial end-member organic matter (Table 2). Components that did not differ significantly are not discussed in the text. Stream fulvic acids had significantly lower amounts of components C3 and Q3, both of which have been associated with microbial organic matter while Q3 has further been identified as an oxidized quinone-like component (Table 2 [Cory and McKnight, 2005]). The stream fulvic acids had significantly higher amounts of components C1, C10 and SQ1, components indicative of a ‘terrestrial’ signature (Table 2 [Cory and McKnight, 2005]). Lastly, the stream fulvic acids had greater amounts of partially and fully reduced quinone-like components SQ1 and HQ compared to the lake and coastal fulvic acids (Table 2). The partially reduced quinone-like component of microbial origin, SQ2, was not significantly different among the fulvic acid samples (Table 2).

[39] The combined amount of the amino acid fluorophores, tyrosine and tryptophan, was greater in the whole water samples compared to the fulvic acid fraction, likely owing to the fact that any free amino acids contributing to the whole water DOM fluorescence would be lost during the fulvic acid isolation procedure (Figure 1). Similar to the patterns observed for the whole water DOM, there was a significant difference in the combined amount of the amino-acid fluorophores among the stream, lake and coastal fulvic acids (Figure 1 and Table 2). There was a weak correlation between the sum of the percent contribution of the tyrosine-like and tryptophan-like fluorophores to the water residence time (r2 = 0.4; p = 0.06; data not shown). A positive correlation was found between the combined amounts of the tryptophan and tyrosine-like components and nitrogen content of the fulvic acid, as well as between the amount of aliphatic carbon content and the tyrosine-like component (Figure 2).

Figure 2.

Correlations between amino acid fluorophores and other fulvic acid chemical parameters. Slope p value (95% confidence level) is shown on graph. (a) Percent tyrosine-like fluorohore (Tyr) versus percent aliphatic carbon. (b) Percent contribution from both tyrosine and tryptophan-like fluorophores versus percent nitrogen (ash-free).

[40] The combined percent mass due to nitrogen and sulfur was less than two percent for all the fulvic acid samples (Table 3), and falls within the range observed for many other fulvic acids [Brown et al., 2004; McKnight et al., 1997]. The stream fulvic acids had significantly lower nitrogen and sulfur contents compared to the lake/coastal mean values (p < 0.05; Table 3). We observed a positive correlation between water residence time and the nitrogen content of the fulvic acids (Figure 3). Lower nitrogen contents of the stream fulvic acids resulted in significantly higher C:N ratios compared to the lake and coastal fulvic acids (p < 0.05; Table 3). No clear differences in the S:C ratios were discerned. The PB sample had the highest S:C ratio, while TWL Creek fulvic acid had the lowest (Table 3).

Figure 3.

Relationships between fulvic acid character data and log of water residence time. Slope p value (95% confidence level) is shown on graph. (a) Percent nitrogen (ash-free). (b) The δ15N.

[41] The stream fulvic acids and the Toolik Lake fulvic acid were depleted in δ15N, with values ranging from −0.16‰ for the Oks Seep to −0.66‰ for the TI Stream, consistent with other terrestrially influenced DOM samples (Table 3) [Amon and Meon, 2004; Hood et al., 2005]. Island Lake and Campsite Lake had values of 0.4‰ and 0.7‰, while the PB and OP coastal fulvic acids had the highest values of 1.3‰ and 1.7‰, respectively (Table 3). There was a strong, positive correlation between the δ15N of the fulvic acids and the estimated water residence time of the source waters (Figure 3). Further, there was an inverse correlation between the C:N ratio and the δ15N of the fulvic acids (r2 = 0.5; data not shown).

4.3. Effects of Light

4.3.1. Whole Water DOM

[42] The whole water showed a decrease in total fluorescence as a function of exposure to natural light, while there was no significant change in fluorescence for the dark control samples as shown for Island Lake whole water sample in Figure 4. Examining the magnitude of change for each fluorophore component identified by PARAFAC provides information on the relative photolability of the fluorophores in a given sample. The loss of fluorescence in the whole water samples varied by fluorophore and ranged from 6 to 100% (data not shown). The average percent loss of fluorescence, calculated from the percent loss of the total modeled fluorescence, ranged from 25% to 32% for the whole water DOM samples (data not shown). Groups of fluorophores having excitation maxima overlapping the output of sunlight (i.e., > 280 nm) showed the greatest loss of fluorescence. These fluorophores include semiquinone-like components SQ1 and SQ2. Of these two components, SQ2 always decreased to a greater extent compared to SQ1 (shown for Island Lake whole water in Figure 4). Generally, component SQ2 showed the greatest percent decrease for every sample compared to all other components (data not shown).

Figure 4.

Changes in fluorescence intensities for (a) Island Lake filtered whole water in natural sunlight at Toolik Lake LTER to (b) Island Lake fulvic acid in the solar simulator. Fluorescence intensities (Fmax values) are in Raman Units (RU).

[43] All whole water samples showed a decrease in the fluorescence index (FI) as a function of exposure to sunlight as shown for Island Lake whole water in Figure 4 while no significant decrease was found for the dark control samples. The magnitude of the decrease was on the order of about 0.1 units, which is significantly greater than instrument error. Because the variation in the FI has been shown to be attributed to the relative amounts of SQ1 and SQ2 in a sample [Cory and McKnight, 2005], the decrease in FI can be explained by the different photolability of components SQ1 and SQ2, i.e., greater loss of SQ2 compared to SQ1 during photolysis results in a lower FI.

[44] The amino acid–like fluorophores showed the smallest percent decrease compared to all the other fluorophores for the whole water samples, as shown for Island Lake whole water sample in Figure 4. For the other whole water samples, there was an increase in the amount of one or both amino acid–like fluorophores (Figure 5).

Figure 5.

Percent contribution from the amino-acid fluorophores (tyrosine and tryptophan-like fluorophores) before and after exposure to light for (a) Arctic whole water and (b) fulvic acid samples.

4.3.2. Fulvic Acid Fraction

[45] The fulvic acid samples showed no detectable decrease in DOC concentration after irradiation (data not shown). Both the DOC concentrations for the irradiated samples and the dark controls were within the analytical error of the original unirradiated concentration. There was a loss in absorbance for all irradiated fulvic acid samples for all wavelengths between 240 and 650 nm, while there was no loss in absorbance beyond analytical error for any dark control (data not shown). The fulvic acids showed the maximum loss of absorbance around 300–310 nm.

[46] Patterns in the fulvic acid loss of fluorescence as a function of light exposure were very similar to what was observed for the whole water samples in natural sunlight as illustrated for Island Lake fulvic acid in Figure 4. For example, all fulvic acid samples showed a decrease in total fluorescence, with the component SQ2 showing the greatest percent decrease, as demonstrated in Figure 4. Accordingly, the fulvic acids all exhibited a decrease in the FI, but to a smaller extent than the whole water samples (Figure 4). Like the whole water samples, the amino acid fluorophores changed the least during photolysis relative to the dark controls (Figure 4). In addition, most fulvic acids had a greater percent contribution of the tryptophan-like and tyrosine-like fluorophores after exposure to light (Figure 5). This was due to a lower percent loss of these fluorophores relative to all other fluorophores or an actual increase in the relative concentration of these fluorophores (Figure 5).

[47] Measures of the extent of photobleaching of fulvic acid samples were correlated with water residence time of the source water. For example, the loss of absorbance of the fulvic acids at 310 nm (the wavelength where the greatest loss of absorbance was observed for all samples), normalized to moles of photons absorbed by each sample, was correlated with water residence time (Figure 6). Further, the percent loss of the most photochemically labile fluorophore, SQ2, was also correlated with water residence time (Figure 6).

Figure 6.

Correlations between photobleaching parameters and log of water residence time for fulvic acid fraction. (a) Loss of absorbance at 310 nm (m−1) normalized to mols of photons absorbed according to Hu et al. [2002]. (b) Percent loss of SQ2.

4.4. Estimation of Microbial Contributions to Fulvic Acid Character

[48] The mixing equation (equation (1)) estimates the fraction of stream δ15N that should remain in the Arctic Ocean assuming a microbial end-member δ15N signature of 7‰ [Schubert et al., 2001] and a terrestrial end-member δ15N value of −0.39‰ (stream mean value; Table 2). Plugging the end-member values into the mixing equation, we calculated that 70% of nitrogen in the coastal fulvic acids could have originated from nitrogen in fulvic acid molecules flushed into the streams from the tundra, with the remaining 30% originating from autochthonous input of N-containing fulvic acid molecules in the rivers or coastal waters. A 30% contribution of new fulvic acid N from degradation of microbial material corresponds to approximately 6% contribution of new fulvic acid carbon, based on the difference in C:N ratios between terrestrial and microbial end-member fulvic acids [Brown et al., 2004; McKnight et al., 1997].

5. Discussion

5.1. Comparison Between Whole Water DOM and Fulvic Acid Fraction

[49] In general, the parameters that can be measured on both the whole water and fulvic acid samples showed strong agreement between the sample types, demonstrating that the chemical characteristics of fulvic acid fraction of the DOM provide insight into the nature of DOM in the system (Figure 1). This suggests that interpretations based on the trends in the chemical characteristics of the fulvic acid fraction may apply to the whole water DOM as well. For example, while the fulvic acid SUVA254 values were higher than their whole water DOM counterpart SUVA254 values, the trend in among the stream, lake and coastal mean values was the same: the stream mean SUVA254 values were significantly higher than the lake and coastal means for both the fulvic acid and whole water DOM (Figure 1 and Table 3). The consistently higher fulvic acid SUVA254 values compared to the whole water values (Figure 1), suggests that the bulk of the chromophoric carbon content in DOM is found in the fulvic acid fraction of the DOM.

[50] Similarly, the distribution of fluorophores was generally well-correlated between the whole water DOM and fulvic acid samples, implying that DOM fluorescence may be strongly associated with the fulvic acid fraction of the DOM. We would expect that fluorophores previously identified as ‘humic/fulvic’ [Coble et al., 1990] should not differ significantly between the fulvic acid fraction and whole water DOM. This was indeed the case for SQ1, a terrestrial fluorophore in the humic/fulvic region identified by Coble et al. [1990]. The percent amount of SQ1 varied linearly and close to the 1:1 line between the whole water and fulvic acid samples (Figure 1), showing that the contribution of SQ1 did not differ between the fulvic acid fraction and the whole water DOM. In contrast, while there is a strong correlation between fulvic acid and whole water DOM amino acid fluorescence, the points fell below the 1:1 line (Figure 1). This is likely due to the loss of free amino acids not associated with the fulvic acid fraction of DOM during the fulvic acid isolation procedure.

[51] The fluorescence index (FI) was an exception to the general correlations between the whole water DOM and fulvic acid values: no correlation exists between the whole water and fulvic acid FI values. This is likely due to the limited range of values shown by both the whole water and fulvic acid samples (Table 3 and Figure 1). For both the whole water and fulvic acid FI values, there was no significant difference between the stream, lake and coastal mean values (Table 3). The whole water FI values were consistently higher the fulvic acid fraction, suggesting that autochthonous organic matter that may contribute to the higher FI values of the whole water sample is not associated with the fulvic acid fraction of DOM.

5.2. Understanding Trends in Fulvic Acid Chemical Characteristics

5.2.1. Influence of Microbial Processes

[52] Of the chemical characteristics that we measured on the fulvic acids, the range of the δ13C, δ15N, FI, SUVA254 and C:N values provide the most information on the dominant source of the organic matter (i.e., microbial versus terrestrial). Overall, the Arctic fulvic acids had low δ13C, δ15N and FI values and high SUVA254 and C:N values. The range of the latter chemical characteristics is consistent with other values for higher plant/terrestrial DOM samples. For example, the δ13C values of all the Arctic fulvic acids are similar to ultrafiltered DOM from the Arctic Ocean [Amon and Meon, 2004] and within the range of values obtained for lignin from C3 plants (−25 to −31‰, respectively [Wedin et al., 1995]). These results provide strong evidence for the predominately terrestrial nature of the DOM in the Arctic surface waters and support the importance of the aquatic-terrestrial linkage in the Arctic.

[53] We expected to measure an increase in the δ13C and FI with a concurrent decrease in the C:N between the streams and the lake/coastal mean values, indicating an increased contribution from microbially derived organic matter into the fulvic acid pool. This is because the δ13C signature, FI value and C:N ratios are expected to shift with a greater contribution from microbially derived carbon [McKnight et al., 2003, 2001]. For instance, microbially derived organic matter has a higher FI value (∼1.8; [McKnight et al., 2001]) and as such we would expect to see an increasing FI with longer water residence time if microbial processing of the DOM pool was significant relative to other controlling processes. No trend in the δ13C and FI values of the fulvic acids across water residence times constrains the influence of microbially derived organic matter in the DOM pool.

[54] Results from the mixing calculation (equation (1)) provide further semiquantitative constraint on the extent of microbial input to the Arctic fulvic acid pool. The mixing calculation results demonstrated that up to 70% of the nitrogen in the coastal fulvic acids could have originated from nitrogen in fulvic acid molecules flushed in to the streams. The remaining 30% of the nitrogen could be due to an input of N-containing fulvic acid, which is consistent with the mean stream nitrogen content being 20% less than the lake/coastal mean nitrogen content (Table 3). We expect the microbially derived fulvic acid to have a nitrogen content of about 5 times greater than that of the stream fulvic acid [Brown et al., 2004]. Thus a 30% contribution of new fulvic acid N from degradation of microbial material would correspond to only a 6% contribution of new fulvic acid carbon, which we hypothesize would not be expected to result in a readily detectable change in the FI or the δ13C signal. Therefore a 30% contribution of microbially derived fulvic acid indicated by the change in δ15N is not inconsistent with the stability of the fulvic acid FI and the δ13C values across the gradient of water residence times. The mixing calculation results suggest that the addition of microbially derived organic enriched in nitrogen compared to plant material is significant to the nitrogen signature of the fulvic acid pool only. Thus the mixing calculation provides an explanation for the reason we observed the expected trends indicative of increased microbial contribution over greater water residence times for the fulvic acid nitrogen signature only, as measured by δ15N and C:N ratios.

5.2.2. Influence of Photochemical Processes

[55] There was no detectable loss of DOC over the course of the fulvic acid photolysis experiments, suggesting that photo-mineralization of fulvic acids may be slow in situ. However, correlations between chemical characteristics of the fulvic acids and water residence time suggests that in situ reactivity such as photochemical degradation of the DOM significantly influences the chemical character of the strongly ‘terrestrial’ Arctic fulvic acids. First, two measures of fulvic acid photobleaching, loss of absorbance at 310 nm and loss of SQ1 fluorescence, were found to be inversely correlated with water residence time, likely owing to depletion of the photolabile DOM over increasing exposure times in the natural environment (Figure 6). Second, exposure of the samples to light in the field and laboratory caused shifts in the optical properties of the whole water and fulvic acids consistent with the initial differences observed between the stream, lake and coastal samples in this study.

[56] For example, this study shows that photodegradation can lower the specific absorbance at 254 nm (SUVA254) owing to greater loss of absorbance compared to DOC. Loss of absorbance is due to the breakdown of the chromophoric fraction, which is mainly attributed to the amount of aromatic fraction in DOM [McKnight and Aiken, 1998]. Accordingly, the stream fulvic acids had higher aromatic carbon contents relative to the lake and coastal fulvic acids (Table 3). Thus loss of absorbance due to exposure to light is consistent with the observed differences in the specific absorptivity and aromatic carbon content of the Arctic fulvic acids derived from sources with different water residence times.

[57] The lack of a relationship between the FI of the fulvic acids and water residence time may reflect differential susceptibility among fluorophores to photodegradation. The FI has been explained as the relative amount of two components, SQ1 (terrestrial) and SQ2 (microbial) [Cory and McKnight, 2005]. We observed that exposure to light results in preferential removal of the ‘microbial’ SQ2 component relative to the loss of SQ1 for both the whole water and fulvic acid samples, resulting in a decrease in the FI. Thus a uniform FI among the stream, lake and coastal fulvic acids may reflect a balance between some continual but small microbial production of new fulvic acid material and preferential photochemical removal of microbially derived organic matter in the surface waters.

[58] In addition, the results from the photodegradation experiments provided evidence for role of sunlight in the nitrogen composition of the fulvic acids. The amino acid–like fluorophores, tyrosine and tryptophan, contributed to a greater percentage of the total fluorescence after photolysis for most samples (Figure 5). Therefore the data suggest that amino acids are not readily photodegraded by either direct or indirect mechanisms, and as a consequence, the relative contribution of the amino acid–like fluorophores increased after photolysis owing to the preferential loss of other fluorophores. This selective preservation of photochemically recalcitrant nitrogen containing moieties may contribute to the observed positive correlation between the fulvic acid nitrogen content and water residence time and is consistent with the evidence for selective preservation shown by the mixing calculation results.

6. Conclusions and Implications

[59] The results presented here indicate that the biogeochemical transformation of the DOM released from the tundra during snowmelt begins in the freshwater streams and lakes and continues in the coastal waters of the Arctic Ocean. The chemical characteristics of the fulvic acid fraction along a gradient of increasing water residence time provided evidence that photochemical transformations are more important than microbial processes in controlling the chemistry, but not the quantity, of the nonmineralized (remaining) fulvic acid fraction of DOM in Arctic surface waters. This result is important because many studies have shown that the ability of microorganisms to uptake and respire DOM changes when the DOM undergoes photodegradation [Moran et al., 2000; Tranvik and Bertilsson, 2001]. Therefore our results imply that the photochemical “pretreatment” that the stream DOM undergoes during export into lakes and coastal zones may regulate the ability of microorganisms to mineralize the DOM and influence the ultimate fate of DOM (organic matter versus carbon dioxide) exported to the ocean.

[60] Results from this study also suggest that the fate of Arctic DOM may be particularly sensitive to changes in the light regime brought about by climate change or shifts in the springtime ozone hole in the Arctic [Solomon, 1999]. The Arctic ozone hole may enhance photochemical processes during the late spring in the Arctic surface waters [Gibson et al., 2000; Pienitz and Vincent, 2000]. Enhanced photochemical processes could clearly cause a shift in the quality of DOM, which we predict would affect the microbial cycling of carbon in the Arctic stream, lake and ocean food webs. Indeed, Gibson et al. [2000] concluded that changes in the quantity or quality of DOM could cause variations in biological UV exposure in the Arctic Ocean to a greater extent than direct ozone-related controls on UV flux.


[61] We greatly appreciate the field support provided by the staff at Toolik Lake LTER. We thank Kevin Wheeler and Amanda Grannas for help with sample collection. Pat Hatcher and Karl Dria (Ohio State University) kindly provided assistance with 13C CPMAS-NMR data for the fulvic acids. This work was funded by NSF grants OPP-0097182 and OPP-0097142.