Terrestrially derived dissolved organic matter (DOM) can impact the fate of persistent organic pollutants (POPs) that are transported to the Arctic via global distillation. Interactions between DOM and POPs through hydrophobic binding processes may influence their photofate in arctic surface waters. We examined the DOM-mediated photodegradation of γ-hexachlorocyclohexane (lindane) and hexachlorobenzene (HCB) in arctic surface waters. These two halogenated organic compounds are commonly detected in the Arctic. We examined how different sources of DOM affect the indirect photolysis of these compounds. We conducted our study using DOM from arctic streams and lakes near the Toolik Lake Long-term Ecological Research Site. HCB or lindane was irradiated in the presence and absence of DOM from these sources, both at the surface of an arctic lake and at 10 cm depth, to investigate the indirect phototransformation of these two compounds and the depth dependence of the observed chemistry. In both artificial and natural sunlight, two of four DOM sources studied stimulated the photodegradation of HCB but not of lindane, suggesting that the indirect phototransformation is a selective process depending on the interactions between DOM and POPs. Through solubility studies, we found that HCB readily partitions to isolated Toolik Lake DOM, while lindane shows no affinity for DOM, findings that corroborate results previously reported in the literature. We demonstrate for the first time DOM's role as a sensitizer for photodegradation of some POPs under field conditions, thus confirming that this process may be an important control on the fate of POPs exhibiting a strong affinity for DOM.
 The fate of persistent organic pollutants (POPs) is of particular concern in arctic regions, due to the accumulation of these deleterious compounds in the region (via global distillation) and their subsequent bioaccumulation/biomagnification in food webs with potentially harmful consequences for human health [AMAP, 2009]. Global distillation facilitates the accumulation of POPs in the terrestrial and aquatic environments of the Arctic, where there are vast stores of organic carbon in soils [Schuur et al., 2008; Schuur et al., 2009] and in surface waters [Opsahl et al., 1999]. Organic carbon has long been recognized as a key compartment in which hydrophobic POPs can accumulate and react, so it is expected that the fate and ecological impact of POPs are linked to the transfer of organic carbon from land to water, an important facet of the Arctic carbon cycle that is susceptible to climate-change impacts [Freeman et al., 2001; Frey and McClelland, 2009].
 Given the importance of organic carbon as a key component in controlling POP fate e.g., as a photosensitizer, sorbent, etc., it has been hypothesized that altered patterns in snowmelt and ice out or shifting organic carbon dynamics due to thawing permafrost or the increased frequency of tundra fires [Romanovsky et al., 2002; Jones et al., 2009] may increase the uncertainty in predicting POP fate and its impact in arctic ecosystems [Wrona et al., 2006]. While evidence has been presented for microbially mediated degradation of POPs in arctic surface waters [Helm et al., 2000], POP photolysis (specifically indirect photodegradation mediated by DOM) has not been investigated.
 To date, studies of the influence of DOM on the photodegradation (specifically reductive photodehalogenation) of halogenated hydrophobic organic contaminants have been limited in scope. Indeed many of these studies have been conducted using commercial humic acids or International Humic Substances Society (IHSS) reference humic materials and artificial light sources [e.g., Burns et al., 1996; Burns et al., 1997; Blough, 1988; Latch and McNeill, 2006]. Because many of these halogenated compounds accumulate in the Arctic and are also designated as POPs we believe that DOM-mediated photoreduction has been largely overlooked. To our knowledge, there has been no study that examines how DOM native to arctic surface waters is able to bind and phototransform halogenated POPs under simulated and natural sunlight. Finally, halogenated POP fate may be particularly sensitive to shifts in the quality and quantity of DOM delivered from land to surface water associated with climate change in the Arctic. This sensitivity may occur because the amount and composition of DOM influences the degree of POP binding and photodegradation that occurs [Chin, 2003; Burns et al., 1996; Burns et al., 1997]. For example, DOM can enhance the degradation of hydrophobic contaminants through binding and partitioning processes by bringing them into close proximity to light-produced reactive intermediates in surface waters [Burns et al., 1996; Burns et al., 1997; Blough, 1988; Latch and McNeill, 2006].
 In arctic surface waters, plant and soil derived DOM is rich in chromophoric carbon moieties [Cory et al., 2007]. The relative importance of the DOM photosensitized degradation may vary with its chemical composition and physicochemical properties of the target pollutant [Guerard et al., 2009]. Cory et al.  demonstrated that arctic DOM from a range of sources is capable of initiating photochemical reactions including the production of hydroxyl radicals (OH•) at rates similar to DOM found in temperate waters [Grannas et al., 2006], which is important given that many POPs are highly susceptible to degradation by this pathway. Cory et al.  also showed that the chemical properties of the fulvic acid fraction of arctic DOM varied progressively with residence time in the surface water environment and that photoreactivity in particular decreased with residence time. Because DOM chemical properties can vary substantially even within a small geographical area such as the North Slope of Alaska, we anticipate the photosensitizing properties of DOM may vary across the Arctic region.
 In this paper we studied the role of the DOM present in arctic surface waters in the photosensitized transformation of two halogenated organic pollutants. The degree to which this can occur is dependent upon the chemical composition of the DOM phase and the physicochemical properties of the analyte. We hypothesized that DOM-mediated photodegradation of halogenated POPs is an important process in the fate of halogenated POPs in arctic surface waters. This DOM-mediated photodegradation may be an important sink for POPs in arctic surface waters due to (1) continuous solar irradiance impinging upon surface waters during the summer and (2) delivery of chromophoric DOM from terrestrial sources (e.g., soils) to shallow lakes via unshaded streams. To avoid the inherent uncertainties in using reference fulvic acids to evaluate the importance of processes expected to vary with DOM chemistry, we conducted experiments on DOM-mediated photo-degradation using whole water DOM collected from arctic surface water, as well as the hydrophobic acid (fulvic acid) fraction of DOM isolated from a number of arctic lakes, rivers, and coastal waters collected near the Toolik Lake Long-Term Ecosystem Research (LTER) site in the Alaskan Arctic. Further, we chose two halogenated POPs, hexachlorobenzene (HCB) and lindane (the γ isomer of hexachlorocyclohexane) which are prevalent in the arctic ecosystem [AMAP, 2009] and do not degrade by direct photolysis. HCB is very hydrophobic and can bind strongly to DOM, while lindane is much more water soluble and is not known to form complexes with DOM [Chiou et al., 1986; Chiou, 2002]. These are properties that may play an important role in the potential photodegradation of these two compounds. Additionally, we examined the potential depth dependence of the observed HCB phototransformation in an arctic lake by irradiating a set of samples at the surface and at 10 cm depth in Toolik Lake. Because we hypothesize that the interactions of POP with DOM are important in the observed photodegradation of POPs, we also investigated the partitioning of HCB and lindane to isolated Toolik Lake DOM.
2. Site Description
 The samples used in this study were either filtered surface waters (collected in summer 2003 from Toolik Lake, Alaska) or solutions of reconstituted fulvic acid isolates collected in 2002 from Toolik Lake, Alaska (Long-Term Ecological Research (LTER) Site, latitude 68°47′N), and several water bodies in the vicinity of the LTER including Imnavait Stream, Tussock Watershed Lower Stream, and Island Lake. Marine samples were taken from the Arctic Ocean near Prudhoe Bay (latitude 70°01′N). Full descriptions of these sites are presented in Cory et al. . These sites were selected for this comparative study because they were shown to exhibit a gradient in aromatic carbon content and associated light absorption and photolability [Cory et al., 2007]. We hypothesized that the extent of DOM-promoted photosensitization may vary with these differences in chemical properties among DOM sources.
 Hexachlorobenzene (99%) was purchased from Sigma-Aldrich (Milwaukee, Wisc.); lindane (+95%) from UltraScientific (North Kingstown, R.I.); hydrochloric acid (trace metal grade), methanol (Optima grade), standard pH buffers 4, 7, 10, and sodium hydroxide (reagent grade) from Fisher Scientific. All glassware was precombusted at 450°C in a muffle furnace for at least 4 h.
3.2. DOM Sources
 A complete site description can be found at http://ecosystems.mbl.edu/ARC/ and in [Kling et al., 2000]. For this study, we used two types of DOM: (1) whole water DOM from Toolik Lake is the bulk DOM in a filtered water sample and (2) the fulvic acid fraction of the DOM which is the hydrophobic acid fraction of DOM isolated by XAD-8 chromatography. Details of the DOM collection, filtration, isolation and characterization are provided in [Cory et al., 2007] and [Cory and McKnight, 2005]. All elemental, isotopic and spectroscopic characterization of the whole water DOM and the fulvic acid fraction of the DOM pointed to the importance of terrestrially derived precursor organic matter as the predominant source of organic matter in these arctic surface waters [Cory et al., 2007]. Many of the chemical proxy measurements for organic matter source that were measured on the arctic DOM, including δ13C, C:N ratios, aromaticity and fluorescence index values, overlapped with those of Suwannee River fulvic acid, the reference end-member sample for aquatic DOM derived from terrestrial sources. The Toolik Lake fulvic acid had an aromatic carbon content of 18%, which was intermediate between the values for surface waters with shorter residence times, e.g., Imnavait Creek fulvic acid at 23%, and longer residence times, e.g., Island Lake fulvic acid at 16%.
 Experiments with fulvic acid DOM were investigated at natural dissolved organic carbon (DOC) concentrations, which ranged from 3 to 12 mg C L−1 for the samples in this study [Cory et al., 2007], Table 1. Solutions of the fulvic acids were prepared to mimic the DOC concentration of the source water [Cory et al., 2007].
Table 1. DOM Properties and Experimental Conditions Studieda
DOC (mg C L−1)
Absorption Coefficient (a300; m−1)
Fulvic Acid % Aromatic C (13C NMR)
Estimated Rate of UVB Light Absorption (nmol photons m−3 s−1)
% HCB Degradation in Light
Instrumental error for DOC analysis was ±0.5 mg C L−1 and ±3 m−1 for absorption coefficients. Typical analytical error for 13C NMR is 2%–5%. See Cory et al.  for detailed chemical characterization of the fulvic acids discussed here.
Toolik Lake whole water, field
Toolik Lake fulvic acid, field
Toolik Lake fulvic acid, field, 10 cm in lake
Toolik Lake fulvic acid, lab
Imnaviat Stream fulvic acid, field
Island Lake fulvic acid, field
Tussock Watershed lower stream fulvic acid, field
Arctic Ocean fulvic acid, field
3.3. Hydrophobic POPs
 We prepared solutions of each analyte in water with or without DOM to investigate the role of DOM in the indirect photolysis of the POP. Solutions of dissolved-phase HCB or lindane were prepared by plating an aliquot of analyte stock solution (∼40 μg/mL in hexane) onto the walls of an amber bottle to yield a final concentration of approximately 15% their aqueous solubilities (HCB, ∼5 μg/L; lindane, ∼1 mg/L) [Schwarzenbach et al., 2003]. The hexane was allowed to evaporate and the desired amount of reagent grade laboratory water or filtered surface water was added. A quasi-equilibrium was established by allowing the solution to sit for several days until the solute fully dissolved (confirmed by replicate extractions of the aqueous solution and gas chromatography-electron capture detector analysis). Due to limited availability of the isolated fulvic acids, one concentration of HCB and lindane was studied (5 μg/L and 1 mg/L, respectively). We recognize these concentrations are higher than typically observed in natural arctic surface waters; however, limits of detection, conducting experiments in a remote field station, and sample size/extraction volume precluded the use of samples representing more environmentally relevant (<ng/L) concentrations. For experiments investigating direct photochemical degradation (i.e., in the absence of DOM), the solutions were used as is after equilibration (or quasi-equilibration) with reagent grade laboratory water. When preparing solutions of HCB or lindane in the presence of previously isolated fulvic acid [Cory et al., 2007], a mass of fulvic acid was added to the equilibrated POP solution (and shaken overnight at room temp) to mimic DOM concentrations of the source water from which the fulvic acid was isolated (Table 1). Reconstitution of the fulvic acid in this manner results in solutions with only slightly higher SUVA values (specific UV light absorption, a proxy for aromaticity) compared to whole water samples [Cory et al., 2007] and results in a solution with similar light absorption properties (and photoactive properties) to whole water. Because the POP solutions were not buffered, but instead were prepared in laboratory grade water with or without fulvic acid, or in filtered lake water, the pH of the final solution ranged from 4.8 to 5.0 or 7.2 to 7.4, for solutions made in the presence of fulvic acid or filtered lake water, respectively.
 Reaction mixtures were transferred into custom-made quartz reaction tubes (∼10 mL by volume with a 0.9 cm path length) that were sealed with O-rings wrapped in foil to mitigate potential sorption of the POPs to the O-rings. Degradation experiments were initiated by placing the reaction tubes into sunlight, or tightly wrapping in foil and deploying alongside irradiated samples for the dark control experiments. At each irradiation time point, POPs in the aqueous solution were extracted with hexane (1:5 (v/v), hexane:water with extraction efficiencies >90%) and immediately quantified on-site at Toolik Lake by gas chromatography using a Fisons 8000 GC equipped with an electron-capture detector and calibrated using external standards. Blanks were analyzed for potential organochlorine content using the same extraction and analysis method and no detectable levels were found in either ultrapure lab water or filtered Toolik Lake water used for these experiments.
3.4. Photochemistry Experiments
 We investigated the ability of DOM to sensitize the photochemical degradation of hydrophobic POPs by comparing the direct photodegradation of HCB or lindane (in the absence of DOM) to their degradation in solutions of whole water or fulvic acid DOM exposed to laboratory or natural sunlight. We chose irradiation times that resulted in approximately 50% reduction of the initial POP concentration (i.e., one half-life).
 All solutions were irradiated either in a solar simulator (Suntest CPS+) or under natural sunlight at Toolik Lake field station during May and June 2003. Dark controls were run alongside the experimental samples using the same quartz tubes wrapped in foil. For experiments in the field, all tubes were placed on a black anodized aluminum rack designed to minimize shading effects. The rack containing the experimental samples plus dark controls was exposed to ambient natural sunlight in air or in water in Toolik Lake at a depth of 10 cm. Ambient air temperatures ranged from −2.1°C to 21°C over the course of the experiment, while the measured surface water temperatures in Toolik Lake during the experiment were 5°C–7°C. Samples irradiated in the laboratory were kept between 13°C and 18°C [Cory et al., 2007].
 Our study goals included a comparison of the light absorbed by the DOM in the field and in the laboratory via chemical actinometry [Dulin and Mill, 1982], however analytical constraints at the remote field site prohibited measurements of the sunlight absorbed by the DOM via this approach. Instead, we compared the UVB output of the xenon lamp in the laboratory solar simulator, measured with a handheld UVB radiometer (Solar Light Co., Glenside, Penna.), to the UVB data logged at 15 min intervals by a ground-level radiometer at Toolik Field Station from 19 to 27 June 2004. At the field site, maximum UVB output occurred between 11:30 and 15:00 LT, with values ranging from 0.96 to 1.5 mW cm−2. The UVB output of the solar simulator was 2.2 ± 0.5 mW cm−2, i.e, on average about twice as high as the noontime maximum UVB output at the field site. Thus, we made a rough approximation to assume that 1 h of exposure in the solar simulator was equivalent to about 2 h of midday sunlight at Toolik Field Station (however, it should be noted that all data reported here are plotted as a function of exposure time to light source and are not scaled to any difference in solar simulator and field irradiation intensities). It is important to note that because this assumption did not consider the energy input from UVA or visible light, the equivalent Toolik sunlight time may be underestimated given that UVB light has been found to be a minor contribution to DOM photobleaching because the total energy output from the UVB is much less than the energy output of the UVA and visible region. Nonetheless, maximum quantum yields for many photochemical reactions sensitized by DOM occur in the UVB wavelength range [Moran and Zepp, 1997].
 Spectra of the solar simulated or total direct natural sunlight, obtained from Suntest or SMARTS (http://www.nrel.gov/rredc/smarts/) [Gueymard, 2001], respectively, were converted to photon flux densities, and then each were normalized to the photon flux density in the UVB region to get a model solar spectrum (in mol photons m−2 s−1 nm−1). Model solar spectrum of the laboratory or natural sunlight multiplied by the respective measured UVB output provided estimated photon flux densities in the UVB region for our experiments.
 Near-surface rates of light absorption by DOM in the field and in the laboratory were estimated according to
where aDOM,λ is the Naperian absorption coefficient of DOM at a wavelength λ (m−1 nm−1) and Eλ,z is the estimated photon flux density at wavelength λ (in mol photons m−2 s−1 nm−1) [Hu et al., 2002] at the surface.
 We tested whether DOM was participating in photochemical reactions by evaluating photobleaching of the DOM in solutions containing only DOM (whole water or fulvic acid without POP) analyzed from experiments conducted alongside the solutions containing POP + DOM. Photobleaching was quantified as the fractional loss of absorbance or fluorescence, relative to the initial absorbance or fluorescence, as a function of time exposed to light. The fractional loss of absorbance was measured at 300 nm because maximum photobleaching had been observed by Cory et al.  to occur between 300 and 320 nm for DOM at these sites. Photochemical loss of fluorescence was measured as the change in emission intensity at an excitation/emission pair of 370/450 nm, respectively. Absorbance values and fluorescence intensities were measured on all samples, as previously described in detail [Cory et al., 2007].
 POP degradation profiles are reported here as plots of time versus the concentration of analyte (C) relative to the initial concentration (Co). To perform comparative statistical significance tests for the dark control experiments and corresponding irradiation experiments, data were fitted to a first-order rate equation. The linearized slopes (and 90% confidence intervals) were compared for light and dark controls using ANOVA for each experimental condition. For the dark control samples, the entire range of data was considered. For the irradiated samples, data points from the initial 6 h of irradiation were considered, as the observed loss often leveled off over time (which would skew the linear fits), likely due to effects of DOM photobleaching, and/or changes in light intensity at the field site. In all cases, at least three time points (each analyzed in duplicate or triplicate) were included to conduct the statistical analyses/comparisons of the initial rates.
3.5. Solubility Studies
 The POP-fulvic acid partition coefficient (KDOC in L/kg) can be determined as a function of DOC concentration (in kg/L as carbon) using equation (2),
where Sw* and Sw are the analyte aqueous solubility in the presence and absence of DOC [Chiou et al., 1986]. Toolik Lake DOM used for solubility studies was concentrated from Toolik Lake inlet water via reverse osmosis and saturated with sodium through cation exchange (sample courtesy of Dr. E. M. Perdue). An appropriate volume of Toolik Lake RO concentrate was diluted with Milli-Q water in 40 mL vials to yield solutions having DOC levels of 0, 7, 14, 27, and 35 mg C L−1 for HCB experiments and 0, 5, 9 and 16 mg C L−1 for lindane experiments. The pH of all samples was adjusted to 7 (no more than ±0.04 pH units) with 2N HCl and the specific conductance was adjusted in each sample to 600 (±22) μS/cm with NaCl. Solubility enhancement experiments were conducted according to the method presented by Uhle et al. . Excess HCB or lindane (determined by water solubility) was plated out on the walls of 8 mL amber glass vials. Vials were filled (no headspace) with the premade DOM solutions, covered with the dull side of aluminum foil, and capped with Teflon-lined polyphenol screw caps. All samples were prepared in triplicate. Vials were placed in a box and placed on a shaker bath at 25°C for at least 24 h to equilibrate.
 To extract samples, ∼1 mL of solution was removed from the 8 mL vial and transferred to a 2 mL vial using a transfer pipette. The mass of solution was weighed in order to calculate the exact volume transferred. One mL of hexane was added to the 2 mL vial and weighed again. Samples were mixed on a vortex mixer for 1 min. Hexane extracts were transferred to 2 mL autosampler vials and analyzed via GC (Fisons 8000 equipped with on-column injector and an electron-capture detector), calibrated using external standards. The concentration of analyte in the various aqueous DOM solutions was then back-calculated from the amount of analyte quantified by GC and assuming the densities of hexane (0.659 g/L) and water (1.000 g/L) at the temperature (25°C) used for the solubility study equilibration period.
4.1. Photolysis of DOM
 The estimated near-surface rate of UVB light absorption by DOM (equation 1) ranged from 28 nmol photons m−3 s−1 to 111 nmol photons m−3 s−1 for the whole water and fulvic acid DOM samples irradiated in the laboratory and in the field (Table 1). For the fulvic acid DOM photolyzed at the near-surface in the field, the estimated rates of UVB light absorption were positively correlated with aromatic carbon content (r2 = 0.7, p < 0.05; aromatic carbon as reported in Cory et al. [2007, Table 1]). A positive correlation between the rate of light absorption and aromatic carbon content is expected given that aromatic carbon comprises most of the chromophoric fraction of DOM. Toolik Lake whole water DOM and its fulvic acid fraction had similar rates of UVB light absorption in the field, consistent with the finding that the fulvic acid fraction of DOM was previously found to be the predominant light absorbing fraction of DOM in these waters [Cory et al., 2007]. Because the estimated rate of UVB light absorption by Toolik Lake fulvic acid was about 2.5 times greater in solar simulated light in the laboratory compared to natural sunlight in the field (Table 1), the rate of light absorption was higher in the laboratory compared to the field.
 We observed loss of absorbance for all DOM samples exposed to light in the lab and in the field (Figure 1), confirming that DOM was initiating photochemical reactions leading to its degradation [Del Vecchio and Blough, 2002; Grannas et al., 2006; Cory et al., 2007]. The absorbance coefficient at 300 nm (a300,DOM) decreased by 10%–20% upon exposure to light for DOM solutions exposed to light in the lab and in the field (Figure 1), while there was no significant change in the absorbance coefficient of the dark control samples over the same time period.
 The samples studied had similar fluorescence index values indicative of plant/soil precursors from the DOM (range 1.17 to 1.35) [Cory et al., 2007]. In the irradiation experiments, fluorescence intensities decreased by 35%–70% of the original values for all samples studied (Figure 2), with no significant loss observed in the dark controls. The greater loss of fluorescence intensity compared to absorbance is consistent with many previous studies [Del Vecchio and Blough, 2002; Ma and Green, 2004; Cory et al., 2007].
4.2. Photolysis of HCB and Lindane
 We did not observe direct photolysis of lindane or HCB exposed to solar simulated light in the laboratory for this study (Figure 3), consistent with previous studies observed for similar compounds [Burns et al., 1997; Fu et al., 2004]. Concentrations of lindane in the presence of DOM and light differed from the initial concentration by 13%, but there was no clear trend in lindane concentrations as a function of exposure to light and there was a similar variability in lindane concentrations for the dark control experiment. During the dark control experiment for lindane (e.g., aqueous lindane in the presence of DOM but not exposed to light), there was a 15%–20% increase in lindane concentrations over the exposure period, suggesting that lindane concentrations may not have reached a stable concentration prior to the start of the experiment or an analytical artifact existed in the analysis of these particular samples. There was no statistically significant difference between the direct and indirect photolysis results for lindane, illustrating lindane's overall photostability in the presence or absence of DOM. For direct photolysis of HCB, there was no statistically significant difference between the direct photolysis and the dark control experiments, illustrating HCB's stability in the absence of photosensitizers.
 HCB underwent indirect photolysis upon exposure to artificial or natural sunlight in the presence of DOM (Figures 3–4). There was a clear trend in decreasing concentrations of HCB in the presence of DOM and light for both laboratory and field experiments, and loss of HCB was clearly greater than variability in the dark controls, as illustrated for HCB in the presence of Toolik Lake fulvic acid in Figure 3. Because the observed kinetics did not clearly fit expected pseudo first order, we did not fit a rate expression over the entire data range. In order to make statistical comparisons between experiments, we attempted to fit a pseudo first-order rate expression on the basis of the initial rates (first 6 h) to enable comparison of irradiated samples to the dark controls (Figure 5). The photon flux in the field over the initial 6 h interval was fairly consistent, relative to the rest of the photo-exposure period.
 Calculating the cumulative percent loss of HCB as the amount of HCB at the final time point of the experiment relative to the initial concentration of HCB, we found that there was about a 50% loss of HCB exposed to solar simulated light in the laboratory and a similar loss of HCB exposed to natural sunlight in the field, when HCB was photolyzed in the presence of Toolik Lake fulvic acid (Figures 4a and 4b). Further, the percent loss of HCB due to indirect photolysis did not differ significantly among Toolik Lake fulvic acid compared to Toolik Lake whole water DOM (Figures 4a and 4c), which is consistent with the similar rates of light absorption of Toolik Lake fulvic acid compared to Toolik Lake whole water DOM (Table 1).
 To evaluate how the degree of association between the target POP and DOM may influence DOM's ability to sensitize the photodegradation of the POP, we measured the binding constant between Toolik Lake fulvic acid for HCB and confirmed literature observations that lindane has no affinity for DOM due to its large aqueous solubility [Chiou et al., 1986; Chiou, 2002]. As expected, lindane showed no affinity for the DOM under the conditions tested, whereas HCB was observed to bind readily to Toolik Lake DOM (Figure 6). We report a HCB log KDOC of 3.87, which compares favorably to the value for 2,2′,4,5,5′-PCB (log KDOC of 4.10 for Suwannee River fulvic acid), a compound with an octanol-water partition coefficient that is nearly identical to HCB (log KOW of 6.11 for the PCB). The small differences in KDOC between these two compounds can be attributable to differences in DOM composition and to a lesser degree the structure of the analytes. These results corroborate findings by Chiou et al.  and Chiou , who also observed no binding of lindane to either Aldrich humic acid (a commercial humic material with an extremely high partitioning capacity for hydrophobic compounds [see Chin et al., 1997]) or to International Humic Substances Society (IHSS) reference fulvic acids, while compounds such as the PCBs and DDT (with a log Kow value of >5.5) readily partitioned into these materials.
4.3. Effect of Surface Water Light Regime on DOM-Sensitized HCB Photolysis
 We qualitatively evaluated the influence of light scattering and light screening in the DOM-sensitized degradation of HCB by carrying out degradation experiments (including light and dark controls) at a depth of 10 cm in Toolik Lake. After 6 h of exposure to natural sunlight, the loss of HCB exposed to DOM at the surface of Toolik Lake was about twice that of HCB loss in the sample at a depth of 10 cm (30% loss versus 15% loss at the surface and 10 cm, respectively; Figure 7). Also, as shown in Figure 5, when taking initial loss rates, the in lake degradation rate is statistically significantly less than degradation at the surface (at the 90% confidence level), and in fact becomes statistically indistinguishable from the dark control experiment. This difference shows that light screening can affect DOM sensitized photodegradation of HCB. Further, these results suggest that the rate of indirect photolysis of HCB likely varies with depth in the water column due to light screening, as expected based on the much lower estimated rate of light absorption at 10 cm compared to the estimated near-surface rate of light absorption (Table 1). However, after 25 h of exposure, the percent loss of HCB at the surface versus 10 cm depth did not differ significantly when comparing these experiments.
4.4. Evidence for DOM-Sensitized Photolysis for Other Arctic DOM Samples
 In addition to Toolik Lake fulvic acid and Toolik Lake whole water DOM, Island Lake fulvic acid plus natural sunlight clearly sensitized the indirect photolysis of HCB (Figure 4e and Figure 5). For other fulvic acid samples isolated from arctic surface waters, the indirect photolysis pathway for HCB loss was not as clearly demonstrated (Imnavait Stream and Tussock Watershed Lower Stream, Figures 4f and 4h and Figure 5). Unlike the Toolik Lake and Island Lake fulvic acids, the other fulvic acids studied exhibited variable or decreasing concentrations of HCB in the dark. For HCB in the presence of stream fulvic acids (Imnavait and Tussock Watershed Lower Stream), mean concentrations of HCB were always lower in solutions exposed to light compared to dark controls, however at some time points there was overlap in the standard deviation between HCB concentrations in the light exposed and dark control samples and the decay constants were not statistically significantly different at the 90% confidence level (Figure 5). While concentrations of HCB in the presence of Imnavait or Tussock Watershed Lower stream fulvic acids kept in the dark exhibited larger standard deviations around the mean compared to the same solution exposed to light, there was a statistically significant trend (i.e., slope statistically significantly different from zero at the 90% confidence level) of decreasing HCB for these samples in the dark as a function of time. A similar result was observed for HCB in the presence of fulvic acid isolated from the coastal waters of the Arctic Ocean, however the rate of indirect photolysis was statistically significantly greater than the dark loss at the 90% confidence interval. These results may suggest that there was a dark interaction between fulvic acid and HCB leading to loss of HCB. We hypothesize that our method of reconstituting the fulvic acids may have resulted in a small fraction (e.g., <10%, based on absorbance and DOC analysis of similar samples in our experience) of fulvic acid that did not dissolve in solution, and that HCB may have preferentially associated with this carbon. It is also possible that interactions between DOM and HCB influenced the extraction and detection of HCB, or more likely that HCB was lost to the sample vials after transfer of the original solution to the quartz tubes at the start of the experiment through interaction with undissolved or coagulated DOM nanoparticles. Alternatively, these results may have been influenced by starting the experiments prior to full equilibration of HCB in solution after the addition of fulvic acids.
5.1. The Case for Indirect Photolysis
 Of the two pollutants that we investigated, the more hydrophobic HCB, with its stronger affinity for DOM (relative to lindane), exhibited greater degradation in the presence of light and DOM from two arctic lakes (Figures 3–5). Neither HCB nor lindane exhibited direct photolysis, consistent with previous work. This observed difference between HCB and lindane reactivity with DOM adds to prior studies quantifying the importance of the DOM–POP interaction for indirect photolysis [Burns et al., 1996; Burns et al., 1997; Lambrych and Hassett, 2006].
 Overall, an important strength of our study was the field demonstration of this process, measuring HCB loss under natural sunlight in the presence of the DOM actually present in these waters. Studies to date that have investigated DOM-mediated photolysis have been conducted in the laboratory using solar simulators and typically with Aldrich humic acid [Burns et al., 1996; Burns et al., 1997] as the DOM source, despite the well known limitations of this material as a model DOM [Malcolm and MacCarthy, 1986]. Further, as recently highlighted by Guerard et al. , different sources of DOM can vary widely in photosensitizing properties, and thus it is important to investigate DOM-mediated photolysis with DOM collected from the field site. Although several predominantly lab-based studies report photochemical degradation pathways of some organic pollutants in water/ice/snow under arctic conditions (including direct pathways and indirect pathways promoted by photochemically generated hydroxyl radicals) [Matykiewiczová et al., 2007a; Matykiewiczová et al., 2007b; Weber et al., 2009; Kahan and Donaldson, 2010; Rowland et al., 2011] we are not aware of any other field study investigating the indirect photolysis of hydrophobic POPs mediated by DOM in arctic surface waters.
 Despite the complications of a potential dark interaction from some of the fulvic acids studied, the results overall show that DOM was participating in photochemical reactions during most experiments involving exposure to light. First, this was shown by the loss of absorbance and fluorescence observed for all samples (Figures 1–2). Taking initial rates of photochemical loss into account, ANOVA results show statistically significant indirect photochemical loss (compared to dark controls) at the 90% confidence interval for 5 of the 8 experiments conducted (Figure 5). In addition, it was previously shown that for the DOM samples in this study, hydroxyl radical production increased as DOM was photobleached [Grannas et al., 2006], which provides direct evidence for the production of reactive intermediates capable of degrading HCB when HCB was exposed to light in the presence of DOM. For two of the four different sources of DOM a statistically significant loss of HCB occurred. For the other two sources of DOM differences in the mean HCB concentration as a function of exposure time were not significantly different from dark controls (Figure 4). It follows that indirect photolysis was the most likely pathway for loss of HCB in the presence of DOM and light for the DOM samples for which a loss occurred.
 Because the evidence for DOM-mediated indirect photolysis varied among the DOM sources studied, it appears that DOM source is a factor and that the chemical characteristics of the DOM may influence the importance of this process across surface waters. The case for indirect photolysis of HCB was strongest for Toolik Lake whole water DOM and fulvic acid, which clearly promoted the photochemical loss of HCB in the laboratory and in the field, with little significant concurrent loss in dark control experiments (Figures 4–5). Island Lake and Arctic Ocean DOM also showed statistically significant photochemical loss in comparison to the dark controls. The photochemical loss of HCB in the presence of DOM from Tussock Watershed Lower stream and Imnavait stream was not statistically significantly different from the dark controls.
 A challenging aspect of the interpretation of these results was the fact that the loss of HCB in the presence of DOM and light as a function of exposure time did not follow the kinetic model previously advanced by Burns et al.  to describe the effects of hydrophobic partitioning to DOM on the basis of second-order DOM-mediated kinetics in aqueous solution. This model predicts a pseudo first-order rate constant for pollutant decay varying with DOM concentration, light absorption, and binding affinity of the pollutant for DOM. Moreover, these investigators only used Aldrich humic acid, which can be inexpensively obtained in large quantities but is a very poor surrogate for aquatic DOM, as Malcolm and MacCarthy  and Chin  have previously discussed. Although our study included variability in DOM concentration and light absorption among the different DOM samples investigated (Table 1), we did not specifically vary these parameters in a manner that would have enabled a rigorous test of the kinetic model previously put forward due to the lack of DOM available for this study. Assuming that the loss of HCB in the presence of all DOM samples studied here was due to indirect photolysis, the variability in the percent loss of HCB was qualitatively consistent with the estimated variability in rate of light absorption and measured DOM concentrations, and expected variability in binding affinity for HCB with the different fulvic acids studied [Grandbois et al., 2008].
 Experiments conducted to compare the rates of surface versus in-lake HCB degradation indicate that light screening can have a significant effect on DOM-sensitized loss of HCB, as anticipated based on the much lower rate of light absorption at 10 cm depth in the water body compared to the estimated near-surface rate of light absorption. This result suggests that it may be possible to couple a model of HCB degradation to a model for photodegradation of DOM, such as the model developed for an alpine lake with a similarly snowmelt driven hydrologic regime [Miller et al., 2009]. In this model, the physical flushing of the lake through the spring snowmelt period and the summer was quantified based on 18O isotopic data and the change in DOM concentration and SUVA was simulated to determine lake-scale first-order degradation rate coefficients for different DOM fractions.
 It is likely that sorptive partitioning of HCB to any undissolved or post experiment precipitated fulvic acids or to glass surfaces contributed to the observed loss of HCB in the dark control samples and to the variability in HCB concentrations in the dark controls and light exposed samples. The aromaticity of the fulvic acids correlates reasonably well with the average percent loss of HCB in the dark, as well as to the mean standard deviation observed for the data points collected from the dark controls (r2 = 0.80 and 0.62, respectively, p < 0.05; data not shown; aromaticity was determined by 13C-NMR) [Cory et al., 2007]. Because lyophilized fulvic acids were added to aqueous solutions of HCB in pure water just prior to the start of the experiment, HCB may have preferentially partitioned into the fine particulate fraction of fulvic acid that may not have initially dissolved. Both the size of the fine particulate fraction of fulvic acid and its affinity for HCB depend on the aromaticity of the fulvic acid, especially at the pH range of the HCB + fulvic acid solutions (4.8–5.0), where many carboxyl groups may be protonated [Ritchie and Perdue, 2003]. This dependence on aromaticity thus favors a greater fraction of fine particulates to facilitate subsequent sorptive loss of HCB in some of our samples. Similarly, aggregation of dissolved fulvic acids to form nano-sized to sub-micron-sized particles or gels [Chin et al., 1998] as the experiment proceeded could also result in additional sorptive losses of the analyte. Previous work has shown that the aromaticity of different humic and fulvic acids explained the variability in their sorption capacities for ∼1000 different hydrophobic organic contaminants [Niederer et al., 2007], which is consistent with our hypothesis that fulvics with the highest aromatic carbon content may have facilitated the largest sorptive loss of HCB from aqueous solution.
 Variability in HCB concentrations in the dark and light exposed fulvic acid solutions may also be due in part to disequilibrium between HCB and fulvic acid interactions. For example it is possible that lyophilized fulvic acids added to aqueous solutions of HCB did not reach equilibrium prior to the start of the experiment. Although scavenging of HCB to fine particulate fulvics or glass surfaces may have contributed to variability in HCB concentrations, it is likely this source of variability and sorptive loss process were similar for dark and light exposed treatments of the same sample at the same experimental time point, such that differences between the dark and light exposed samples were most likely due to photochemical processes mediated by DOM. The pH and ionic composition of the dark and light exposed samples were identical, and the time from experiment start to HCB extraction was the same for both dark and light exposed samples (at a given time point). Although photodegradation preferentially removes the aromatic fraction of fulvic acid [Cory et al., 2007], which should lower its sorption capacity, it is unlikely that the particulate aromatic carbon content decreased enough over the time scale of our photolysis experiments to alter the sorptive capacity of the fulvic acid relative to the dark control.
 Microbially mediated loss of HCB can be effectively ruled out in these experiments considering that fulvic acid isolates or sterile-filtered (0.22 μm) water was used as the DOM sources, and microbial degradation of α-hexachlorocyclohexane has been suggested to occur with half-lives of 0.61 to 1.44 years in arctic waters [Helm et al., 2000], much slower than the dark loss observed here for HCB. Additionally, abiotic reduction of HCB by reduced DOM moieties is highly unlikely because these reactions are typically very slow (days for the most highly reactive halogenated organic compounds such as hexachloroethane) [Kappler and Haderlein, 2003].
 Although we do not at present have the data to distinguish between sorptive loss of HCB to fine particulate organics, glass surfaces, or other potential loss processes in dark controls, hydrophobic POPs such as HCB have such a strong affinity for particulate organic matter and thus accumulate in soils and sediments where they are protected from sunlight. Processes such as thawing permafrost are expected to increase delivery of particulate and dissolved organic matter enriched in hydrophobic aromatic carbon to sunlit surface waters [McGuire et al., 2009], where sorptive interactions and indirect photodegradation mediated by DOM probably act together to influence the fate of HCB. Thus our results on DOM fulvic acid character, coupled with the variability in the dark control experiments, suggest that both processes are important for POP fate.
5.2. Biogeochemical Influences on DOM's Photosensitizing Properties
 The timing of the delivery of aromatic DOM to arctic surface waters roughly coincides with the maximum solar output of the summer season, and thus based on a qualitative comparison in the estimated UVB action spectrum of “fresh” (unexposed) DOM in June versus bleached DOM in August (Figure 8), indirect photolysis of HCB or similar POPs would be expected to be greatest in early summer, possibly occurring largely as the DOM is exported from the soil flowing overland into shallow pools and streams. Additionally, POPs accumulated over winter months in snow will be released in a spring pulse [Meyer and Wania, 2008] concurrently with newly exposed DOM. As for the overall photobleaching of DOM, we expect indirect photolysis of POPs to occur predominantly within the top ≤1 m of the water column, due to the screening of UV light by DOM below these depths in the water column [Morris et al., 1995; Wrona et al., 2006, and references therein]. Nonetheless, given that DOM photodegradation in lake surface waters has been observed and has been quantitatively modeled [Miller et al., 2009], we would expect indirect photolysis to also be an important sink for POPs in many arctic lakes.
6. Conclusions and Implications
 This study provided a qualitative assessment that indirect photolysis of POPs may be an important component of POP fate in the Arctic and susceptible to changes in the Arctic due to the dependence of this process on DOM and light. Overall, the importance of DOM-sensitized photolysis of POPs may be enhanced with climate-driven changes to the arctic hydrologic and carbon cycles that alter the timing, quantity or quality of fresh DOM exported from the tundra to surface waters [Wrona et al., 2006]. Our results suggest that it is important to consider that DOM-mediated photodegradation of POPs may be one of many pathways influencing POP fate that may be affected by climate related disturbances to the hydrologic or carbon cycles in the Arctic.
 This research was funded by NSF's Arctic Natural Sciences Program (Grant OPP-0097142). We gratefully acknowledge the staff of the University of Alaska's Institute of Arctic Biology Toolik Field Station. We also thank Kevin Wheeler and Chris Jaros for their field support during the summers of 2002 and 2003.