Article pubs.acs.org/est
Complete and Partial Photo-oxidation of Dissolved Organic Matter Draining Permafrost Soils Collin P. Ward and Rose M. Cory* Department of Earth and Environmental Sciences, 2534 C.C. Little Building, 1100 North University Avenue, University of Michigan, Ann Arbor, Michigan 48109, United States S Supporting Information *
ABSTRACT: Photochemical degradation of dissolved organic matter (DOM) to carbon dioxide (CO2) and partially oxidized compounds is an important component of the carbon cycle in the Arctic. Thawing permafrost soils will change the chemical composition of DOM exported to arctic surface waters, but the molecular controls on DOM photodegradation remain poorly understood, making it difficult to predict how inputs of thawing permafrost DOM may alter its photodegradation. To address this knowledge gap, we quantified the susceptibility of DOM draining the shallow organic mat and the deeper permafrost layer of arctic soils to complete and partial photooxidation and investigated changes in the chemical composition of each DOM source following sunlight exposure. Permafrost and organic mat DOM had similar lability to photomineralization despite substantial differences in initial chemical composition. Concurrent losses of carboxyl moieties and shifts in chemical composition during photodegradation indicated that photodecarboxylation could account for 40−90% of DOM photomineralized to CO2. Permafrost DOM had a higher susceptibility to partial photo-oxidation compared to organic mat DOM, potentially due to a lower abundance of phenolic moieties with antioxidant properties. These results suggest that photodegradation will likely continue to be an important control on DOM fate in arctic freshwaters as the climate warms and permafrost soils thaw.
1. INTRODUCTION The amount of terrestrially derived dissolved organic matter (DOM) exported from land to freshwaters, and degraded therein, is comparable to the major net exchanges of C with the atmosphere, including uptake by oceans and land and the net flux of C between the atmosphere and land.1−5 For example, much of the CO2 released from rivers and lakes in northern latitudes results from the degradation of terrestrially derived DOM,5,6 and the flux of C from arctic and boreal freshwaters as CO2 to the atmosphere or as DOM to the ocean may account for up to 40% of net land surface C exchange with the atmosphere.7 Thus, understanding the controls on DOM fate as C that is either (1) completely oxidized to CO2 and returned to the atmosphere or (2) partially altered and exported to the ocean in organic form is critical to closing the terrestrial carbon budget. Bacterial respiration has been considered the primary control on the fate of DOM exported from land to water, but we recently showed that photochemical processing of DOM accounted for 70−95% of the total DOM degraded in the water column of arctic freshwaters.8 Photodegradation of DOM results in mineralization to CO2 by sunlight (complete photooxidation) as well as other products including oxidized DOM (partial photo-oxidation) and other altered moieties that can enhance or slow bacterial respiration of DOM.8−12 Both complete and partial photo-oxidation of DOM are important controls on DOM fate in arctic freshwaters in part due to high © 2016 American Chemical Society
concentrations of light-absorbing DOM that is labile to photodegradation.8,13−17 However, rates of DOM photodegradation to CO2 or to partially oxidized DOM may change as permafrost thaws and inputs of permafrost DOM to surface waters increase.18−24 This is because permafrost DOM has a different chemical composition compared to DOM draining the active soil layer.14,17,25−28 The chemical composition of DOM is expected to control its lability to photodegradation, and there are three main ideas on how DOM composition controls its degradation by sunlight to form CO2 or partially oxidized and altered organic matter, including the role of (1) aromatic carbon, (2) carboxyl carbon, and (3) phenolic carbon. First, aromatic C is the light-absorbing fraction of DOM and thus initiates the degradation of DOM through both direct and indirect photochemical reactions. For example, absorption of sunlight by aromatic C is likely responsible for the direct degradation of aromatic, high-molecular-weight compounds into less-colored, aliphatic, lower-molecular-weight compounds.13,17,29 Additionally, light absorption by aromatic C initiates the production of reactive intermediates or radicals including hydrogen peroxide, singlet oxygen, or hydroxyl radical30−32 or other reactive intermediates,33 which can Received: Revised: Accepted: Published: 3545
November 2, 2015 January 26, 2016 February 24, 2016 February 24, 2016 DOI: 10.1021/acs.est.5b05354 Environ. Sci. Technol. 2016, 50, 3545−3553
Article
Environmental Science & Technology
ICR MS), and solid-state 13C nuclear magnetic resonance spectroscopy (NMR) to assess the chemical composition of DOM. By comparing and contrasting the lability of different sources of DOM to photodegradation, we provide information on how the chemical composition of DOM controls its susceptibility to complete and partial photo-oxidation and how the photochemical processing of DOM in arctic surface waters may change with increased inputs of permafrost DOM.
indirectly break down DOM to form CO2 or partially oxidized compounds.9,33−35 Because light absorption by aromatic C is the first step in the direct and indirect degradation of DOM, several studies have proposed that the content of lightabsorbing aromatics in DOM governs the lability of DOM to photodegradation.17,36,37 DOM draining deeper permafrost layers of arctic soils contains less light-absorbing, aromatic C compared to DOM draining the active, organic mat.14,17,25−28 Therefore, permafrost DOM has been proposed to be less susceptible to photodegradation compared to organic mat DOM.17 However, DOM from deeper permafrost soils has been found to be more labile to photodegradation, quantified as a loss of chromophoric DOM, than expected on the basis of only concentrations of light-absorbing aromatic C.14 This result is likely due to differences in the chemical composition of permafrost DOM compared to organic mat DOM, which may influence the pathways and moieties involved in photodegradation following the absorption of sunlight by aromatic C. For example, carboxyl C within DOM has been proposed to be labile to complete photomineralization to CO2.38−42 Miles and Brezonik (1981)40 reported that chemical conversion of carboxyl C within DOM to less reactive functional groups (i.e., esters) decreased photochemical oxidation of DOM to CO2 by 50%, thereby suggesting that carboxyl C may account for a substantial fraction of CO2 produced upon exposure of DOM to sunlight. However, it was not possible to quantify the extent to which carboxyl C within DOM was converted to lessreactive, ester C or how the chemical treatment used to convert carboxyl to ester C may have changed the composition of other moieties in the DOM also involved in photochemical degradation. To address these limitations, others have investigated photodecarboxylation by quantifying changes in carboxyl C content during photodegradation by titration.42 However, quantification of carboxyl C by titration of the organic acids within DOM suffers from the wide range of acidic functional groups (and pKa values), including carboxyl, phenolic, or other moieties, that contribute to the total acidity of DOM in natural waters.43 Third, in addition to absorption of light by aromatic C or the photochemical conversion of carboxyl C to CO2, phenolic moieties within the DOM pool may also play a pivotal role in suppressing photodegradation by quenching indirect reactions involved in the complete or partial photo-oxidation of DOM. For example, a recent study proposed that phenolic C, a subset of the aromatic fraction of DOM, acts as an antioxidant by quenching reactive oxygen species (ROS)44 involved in the photochemical oxidation of DOM.9,33−35 Given the antioxidant properties of phenolics,44 phenolic C may slow the photooxidation of DOM by quenching ROS before they can modify DOM. However, no study has tested the relationship between phenolic C content and susceptibility of DOM to photodegradation. Thus, the objectives of this study were to test the role of (1) aromatic, (2) carboxyl, and (3) phenolic carbon in the photodegradation of DOM. We quantified the susceptibility of DOM draining the shallow organic mat and the deeper permafrost layer of arctic soils to complete and partial photooxidation by measuring apparent quantum yields (AQYs) for each process. We related differences in AQYs to the chemical composition of DOM before and after sunlight exposure using UV−visible absorbance, fluorescence spectroscopy, Fourier transform ion cyclotron resonance mass spectrometry (FT-
2. METHODS 2.1. Experimental Design. Site location and sample collection were previously described in detail,27 and the experimental design is presented in Figure S1. Briefly, triplicate soil pits were dug on a hillslope within the Imnavait Creek watershed in Arctic Alaska. The soils are classified as moist acidic tundra and are composed primarily of tussock cottongrass (Eriophorum vaginatum), the major vegetation species in Arctic Alaska. The average thaw depth in late summer (i.e., mid-August) in the Imnavait Creek basin is ∼44 cm, with a maximum of ∼52 cm; therefore, the deeper permafrost layer is well below the deepest thaw for this location.45 DOM was leached in triplicate from soils collected from the shallow organic mat (5−15 cm) and deeper permafrost soil layer (95−105 cm) of each pit. The pH of the leachates were mildly acidic (organic mat = 5.7 ± 0.1; permafrost = 5.8 ± 0.1),27 similar to the pH of Imnavait Creek (∼5.4), the headwater stream that drains these soils.8 Thus, the results of this study are relevant for arctic freshwaters characterized by mildly acidic pH, but different results may be expected in near-neutral waters, given that DOM photodegradation may vary with pH.46 2.2. Measurement of Complete and Partial Photooxidation. Photochemical CO2 production and O2 consumption were quantified as previously described.8,47 Photochemical CO2 production (μM) was quantified as the lightexposed minus dark-control difference in dissolved inorganic carbon (DIC) concentration, while photochemical O2 consumption (μM) was quantified as the dark minus light difference in the dissolved O2 concentration. Inner filter effects were accounted for by normalizing the photochemical CO2 production or O2 consumption to light absorption by each DOM source in the experimental phototubes.48 Apparent quantum yields for photochemical CO2 production or O2 consumption were assumed to decay exponentially from 280 to 600 nm.49 The contribution of H2O2 production to photochemical O2 consumption was quantified as the change in photochemical O2 consumption upon the addition of catalase (200 units mL−1; Sigma-Aldrich).9 Additional details of the photodegradation experiments are described in Figures S2 and S3. 2.3. Analysis of DOM Photodegradation from WholeWater and Solid-Phase Extraction. Following sunlight exposure, aliquots of dark-control and light-exposed DOM were taken for optical spectroscopy and FT-ICR MS. The remaining leachates were extracted on PPL solid phase50 (SPE) to concentrate DOM and minimize inorganic impurities in preparation for high-resolution FT-ICR MS and solid-state 13C nuclear magnetic resonance spectroscopy (NMR). DOC recovery of the solid-phase extractions ranged from 54−61% for organic mat DOM and 60−76% for permafrost DOM. As expected,27 similar changes to the chemical composition of organic mat and permafrost DOM after sunlight exposure were observed for the whole-water (0.45 μm filtered soil leachate) 3546
DOI: 10.1021/acs.est.5b05354 Environ. Sci. Technol. 2016, 50, 3545−3553
Article
Environmental Science & Technology
DOM to sunlight produced more CO2 than did permafrost DOM (organic mat =161 ± 10 μM-CO2, permafrost =34 ± 1 μM-CO2; Figure S2). The production of CO2 by sunlight consumed 5.2 ± 0.5% and 3.4 ± 0.3% of the initial C in organic mat and permafrost DOM. When normalized to mol of photons absorbed by each DOM source (Table S1), there was no detectable difference in the apparent quantum yield for photochemical CO2 production at 350 nm (Figure 1; organic mat = 0.90 ± 0.06 mmol CO2 mol photons−1, permafrost = 1.25 ± 0.12 mmol CO2 mol photons−1; p = 0.08).
and SPE-DOM fractions by any method used (i.e., optical spectroscopy and FT-ICR MS). We present the mass spectra of SPE-DOM because these spectra were less susceptible to interfering species (e.g., salts)27 and to directly compare to 13C NMR spectra, which can only be collected from SPE-DOM. 2.4. Effect of Sunlight on the Chemical Composition of Organic Mat and Permafrost DOM. UV−visible absorbance and fluorescence spectra were acquired using a Horiba Scientific Aqualog, as previously described.27 Naperian absorption coefficients were calculated by multiplying absorbance (A) by 2.303 and dividing by the path length (m) of the quartz cuvette. Spectral slope ratio (SR), specific UV−visible absorbance as 254 nm (SUVA254; L mg-C1− m−1), and fluorescence index (FI) were calculated following previously reported methods.29,51,52 High-resolution mass spectra were acquired using a 12 T Bruker SolariX FT-ICR mass spectrometer. Sample preparation, acquisition parameters, and formula assignment criteria were previously described in detail.27 The mass spectra were analyzed for differences in relative intensity of the peaks common to light-exposed versus dark-control spectra. Formulas detected in each FT-ICR mass spectrum were categorized as photodegraded or photoproduced. Formulas were categorized as degraded versus produced by light if their intensity decreased or increased after light exposure, respectively, using the 95% confidence interval of the mean of experimental replicates (N = 3) to determine if a change in peak intensity was significantly different from zero. The coefficient of variation of peak intensities within instrumental replicates was 12%, similar to the coefficient of variation reported by Sleighter et al. (2012).53 Aromatic or aliphatic character of formulas produced or degraded by light was determined using the aromaticity index (AIMOD).54 Formulas were categorized as tanninlike on the basis of their chemical composition (0.6 ≤ O/C ≤ 1.2, 0.5 ≤ H/C ≤ 1.5, AIMOD < 0.67).36 There are three main analytical limitations of FT-ICR MS as a quantitative tool for the characterization of DOM; each are well-documented. Briefly, FT-ICR MS is optimized for the detection of higher- versus lower-mass formulas (>250 m/z),55 depends strongly on the ionization affinity of compounds within the DOM pool,27,56,57 and does not yield structural information about the isomeric diversity within each m/z detected.58 To minimize these limitations of FT-ICR MS, we used Orbitrap mass spectrometry to detect low-mass formulas and 13C NMR to assess functional-group distribution. The production of low-molecular-weight formulas (100−250 m/z) after sunlight exposure was determined by comparing Orbitrap mass spectra (Exactive, Thermo Scientific) of lightexposed and dark-control organic mat and permafrost DOM.55 Details of collection and processing of Orbitrap mass spectra are described in Table S4. 13 C NMR spectra of freeze-dried SPE-DOM were acquired using a Varian Infinity CMX 300 MHz spectrometer at a spinning rate of 14 kHz. Sample preparation, acquisition parameters, and spectra processing details were previously reported.27 The instrumental error associated with each integration region was quantified by acquiring replicate resonance spectra of Suwannee River Fulvic Acid (International Humic Substances Society) and was less than 1%.
Figure 1. Apparent quantum yield at 350 nm for photochemical oxidation (x-axis) and mineralization (y-axis) of organic mat (blue squares) and permafrost (red circles) DOM. Error bars show standard error of analytical duplicates and, in some cases, are smaller than the symbol.
Similar to CO2 production, organic mat DOM consumed more O2 due to light exposure compared to permafrost DOM (organic mat = 146 ± 23 μM-O2, permafrost = 55 ± 4 μM-O2; Figure S2). However, in contrast to CO2 production, apparent quantum yields for photochemical O2 consumption at 350 nm were significantly lower for organic mat DOM compared to permafrost DOM (organic mat =0.82 ± 0.09 mmol O2 mol photons−1, permafrost =1.94 ± 0.13 mmol O2 mol photons−1, p < 0.001; Figure 1). It follows that the ratio of photochemical O2 consumption to CO2 production was significantly lower for organic mat DOM compared to permafrost DOM (Figure 2, S2; organic mat = 0.90 ± 0.04, permafrost = 1.60 ± 0.12; p < 0.001). There was no detectable production of hydrogen peroxide (H2O2) from exposure of organic mat or permafrost DOM to sunlight (Figure S3). Although photoexposure of DOM is expected to result in H2O2 production,9,30 no increase in dissolved O2 after the addition of catalase suggests that photochemical production of H2O2 from organic mat and permafrost DOM was below the limit of detection (∼1 μM). 3.2. Effects of Sunlight on the Light-Absorbing and Emitting Fractions of Organic Mat and Permafrost DOM. As expected, sunlight exposure degraded the aromatic fraction of organic mat and permafrost DOM, quantified as loss and shifts in the chromophoric and fluorescent fraction of DOM (CDOM and FDOM; Table S1; Figures S4 and S5). For example, after sunlight exposure, there was a 27 ± 3 and 4 ± 1 decrease in CDOM (per m), quantified as the decrease in the absorption coefficient at 305 nm (a305) for organic mat and permafrost DOM, respectively. When normalized to mol photons absorbed by each DOM source from 280−600 nm, the loss in a305 was similar for each DOM source (organic mat = 4.5 ± 0.4 m−1 mol photons m−2; permafrost = 4.2 ± 0.8 m−1 mol photons m−2). Sunlight exposure also decreased the
3. RESULTS 3.1. Photomineralization and Photo-oxidation of Organic Mat and Permafrost DOM. Exposing organic mat 3547
DOI: 10.1021/acs.est.5b05354 Environ. Sci. Technol. 2016, 50, 3545−3553
Article
Environmental Science & Technology
Figure 2. Loss (gray squares) and gain (black squares) in relative peak intensity after sunlight exposure of formulas that were common to darkcontrol and light-exposed organic mat or permafrost DOM. Changes in peak intensities were calculated using the 95% CI of experimental triplicates. Dotted red line corresponds to CV of peak intensities in instrumental duplicates (12%).
Figure 3. Van Krevelen diagrams of formulas that decreased (gray) or increased (black) in abundance after sunlight exposure. Bubble size is proportional to the change in relative peak intensity after sunlight exposure. Changes in peak intensities were calculated using the 95% CI of experimental triplicates. Peak intensities decreased by 1 to 80% and increased by 1 to 200% in organic mat DOM. Peak intensities decreased and increased by 1 to 60% in permafrost DOM. The CV of peak intensities in instrumental duplicates was 12%.
average molecular weight of each DOM source, as measured by an increase in the slope ratio (SR). The absolute increase in SR was 1.5-fold higher for organic mat compared to that for permafrost DOM; however, the light-normalized increase in SR was significantly higher for permafrost compared to organic mat DOM (p < 0.001). 3.3. Effect of Sunlight on High-Resolution Mass Spectra of Organic Mat and Permafrost DOM. There was strong overlap between peaks in mass spectra of darkcontrol and light-exposed treatments of each DOM source, which is as expected given that less than 10% of the DOM was consumed or altered during light exposure (section 3.1). On average, 69−79% of peaks identified in dark-control treatments of organic mat DOM were also detected in light-exposed treatments (Table S2). Similarly, 72−81% of peaks were common to replicate dark-control and light-exposed treatments of permafrost DOM (Table S2). However, photochemical production of CO2, consumption of O2, and the concurrent loss and alteration of CDOM and FDOM suggested substantial alteration of DOM during light exposure. To test the magnitude of this photochemical alteration, we used Spearman’s rank correlation to identify the degree to which peak intensities were ranked in the same order between dark-control and light-exposed treatments of each DOM source. The rank analysis indicated that the distribution of the common peak intensities was significantly altered after photodegradation. Spearman’s rank correlation coefficients between the intensities of the peaks common to dark-control
and light-exposed organic mat DOM spectra ranged from 0.77−0.91, less than coefficients within triplicate dark-control spectra (0.96−0.99) or triplicate light-exposed spectra (0.95− 0.99; Table S3). Spearman’s rank correlation between the intensity of peaks in mass spectra of permafrost DOM exhibited a similar decrease after photodegradation, as observed for organic mat DOM (Table S3). Thus, the relatively short-term photoexposure did not result in the complete loss of many peaks present initially in the dark control or the production of many new peaks (unique to light-exposed DOM), but it did significantly alter the distribution of the intensities of the peaks common to both dark-control and light-exposed DOM. Photodegraded formulas had (i) higher molecular weight, (ii) higher O/C ratios, and (iii) lower H/C ratios compared to photoproduced formulas. The molecular weight of formulas in organic mat and permafrost DOM that were photodegraded was on average 106 and 62 Da higher, respectively, than formulas that were photoproduced (Figure 2). Thus, for both DOM sources, higher-molecular-weight formulas were photodegraded, while lower molecular weight formulas were photoproduced. Analysis of organic mat and permafrost DOM in van Krevelen space revealed that photodegraded formulas exhibited higher O/C and lower H/C ratios compared to photoproduced formulas (Figure 3). In each DOM source, there were two clusters of photodegraded formulas. One cluster of photodegraded formulas had high O/C ratios (>0.5) and spanned a range of H/C from 0.5−1.2, while another cluster had lower 3548
DOI: 10.1021/acs.est.5b05354 Environ. Sci. Technol. 2016, 50, 3545−3553
Article
Environmental Science & Technology
mat and permafrost DOM as indicated by an increase in the SR and by the preferential degradation of higher-molecular-weight formulas in the FT-ICR mass spectra for both DOM sources (Figure 2; Table S1). In addition, sunlight decreased CDOM and FDOM, the light-absorbing and emitting fractions of DOM, respectively, suggesting that aromatic moieties within the DOM pool were degraded (Table S1; Figures S4 and S5). Consistently, the ratio of aromatic to aliphatic C within each DOM source decreased after exposure to sunlight, as measured by 13C NMR (Figure 4). However, aromatic C content is likely not the only control on DOM susceptibility to photodegradation. Permafrost DOM had less aromatic C27 but was more susceptible to photodegradation compared to organic mat DOM, quantified as the sum of apparent quantum yields of complete and partial photooxidation (Figure 1). This finding is consistent with previous studies showing that aromatic C content was not a strong control of the susceptibility of freshwater and marine DOM to photodegradation.14,59−61 Although light-absorbing aromatic C is needed to initiate photochemical reactions, there are likely many subsequent, indirect reactions taking place following light exposure that break down DOM, yielding CO2 and partially oxidized or degraded DOM. For example, exposure of DOM to sunlight produces reactive oxygen species (ROS) and initiates chargetransfer reactions that completely and partially oxidized DOM.9,31−35,62−64 These indirect photochemical reactions taking place following light absorption may depend strongly on the composition of aromatics or other moieties susceptible to degradation by reactive intermediates or on the composition of moieties that can quench reactive intermediates. By relating changes in DOM composition following photoexposure to CO2 production (complete photo-oxidation) and to partial photooxidation, we provide new insight into other moieties along with aromatic C that may be controlling the complete and partial photo-oxidation of DOM. 4.2. Photomineralization of Organic Mat and Permafrost DOM: Evidence for Photodecarboxylation. The results strongly support photodecarboxylation as an important pathway for the complete photo-oxidation of DOM to CO2. Generally, photodecarboxylation is described as the breakdown of organic acids (R−COOH) to hydrocarbons (R−H) and CO2 by sunlight:62−64
O/C ratios (0.3−0.5) and spanned a range of H/C from 1.0− 1.5. Compared to photodegraded formulas, photoproduced formulas exhibited lower O/C and higher H/C ratios. For instance, compared to formulas degraded by sunlight, the average O/C of formulas in organic mat and permafrost DOM that increased in intensity after photodegradation was 0.11 and 0.23 lower, respectively, while the average H/C of photoproduced formulas in each DOM source was 0.14 higher. Despite a large overlap in the composition of photodegraded formulas in organic mat and permafrost DOM, the composition of photoproduced formulas differed between the DOM sources (Figure 3). Photoproduced formulas in organic mat DOM formed one cluster ranging in O/C from 0.4 to 0.7 and H/C from 0.6 to 1.5. In contrast, photoproduced formulas in permafrost DOM formed two clusters: one ranging in O/C from 0.4 to 0.6 and in H/C from 1.2 to 1.6 and a second ranging in O/C from 0.05 to 0.15 and in H/C from 0.75 to 1.1. 3.4. Effect of Photodegradation on Functional Group Distribution. Integration of resonance spectra of DOM according to dominant functional or structural groups revealed compositional differences between dark-control and lightexposed treatments of organic mat and permafrost DOM (Figure 4). In both DOM sources, sunlight exposure decreased
Figure 4. Functional group distributions of dark-control and lightexposed organic mat DOM (dark and light blue bars) and permafrost SPE DOM (dark and light red bars) determined using 13C NMR. Instrumental error for each functional group region was
