Source- and substrate-specific export of dissolved organic matter from permafrost-dominated forested watershed in central Siberia



[1] Terrestrial and aquatic dissolved organic matter (DOM) was characterized to trace the likely processes of DOM formation and stream export in a permafrost-dominated watershed in central Siberia. Stream samples were collected in spring (May–June 2003) and summer (July–August 2003) at both low flow and stormflow. Dissolved organic matter was analyzed by pyrolysis/gas chromatography/mass spectrometry, and identified pyrolysis products were simultaneously analyzed for compound-specific isotope ratios by isotope ratio mass spectrometry. Pyrograms of terrestrial and stream DOM contained a similar series of pyrolysis products, suggesting a terrestrial origin for DOM in the small stream draining our study catchment. However, despite the overall similarity of chemical composition of stream DOM at different seasons, we also observed distinct differences in isotopic fingerprint between seasons and hydrologic phases (stormflow versus low flow). This variation appears to be due to the changing origin of stream DOM from different soil layers and the catchment sources following permafrost thawing during the frost-free period. In general, chemical and isotopic composition of stream DOM was similar to DOM produced in soils of colder north facing slopes (P < 0.01) with a shallow active layer. South facing slopes with deeper active layers produce little DOM that enters the stream, suggesting that DOM produced in the active layer is retained and stabilized in underlying, unfrozen mineral soils. Climate change that results in additional seasonal thawing of permafrost-dominated landscapes will decrease the amount of DOM exported to riverine systems and change its chemical composition.

1. Introduction

[2] Dissolved organic matter (DOM) plays an important role in the global carbon cycle, with export of stored carbon from terrestrial sources via creeks and streams to the world oceans [Neff and Asner, 2001; Dittmar and Kattner, 2003; Cole et al., 2007]. Production and release of DOM in nonpermafrost forest ecosystems are well documented [Kalbitz and Solinger, 2000; McDowell, 2003; Michalzik et al., 2003]. In general, observed DOM concentrations and fluxes represent the balance between processes that release DOM, such as leaching and desorption [Guggenberger et al., 1994], and those that remove DOM, such as adsorption and biodegradation [Jardine et al., 1989; Kalbitz and Solinger, 2000]. There is an important role of local carbon stocks in the DOM export to riverine systems [Aitkenhead and McDowell, 2000], while temperature and moisture [Christ and David, 1996] exert controls on production and mobilization of DOM in terrestrial ecosystems. Properties of soils modulate the terrestrial DOM flux to surface runoff [Guggenberger and Kaiser, 2003]. However, for permafrost-affected ecosystems of Siberia, there is far less known about controls of DOM fluxes in soils and rivers [Kawahigashi et al., 2004; Prokushkin et al., 2005].

[3] Arctic and boreal forest permafrost soils contain a significant portion of the world's soil organic carbon [Hobbie et al., 2000] that can be subjected to substantial release under global warming [Schulze and Freibauer, 2005]. In northern boreal ecosystems, organic carbon is mainly stored in the upper soil as plant debris, which is a major source of DOM. Since permafrost prevents deep seepage of soil solutes, retention and mineralization of DOM in the shallow active layer varies dramatically depending on the depth to which permafrost thaws in the summer [MacLean et al., 1999]. Under current conditions, Arctic and sub-Arctic river basins release large amounts of terrigenous DOM to northern seas [Hansell et al., 2004; Striegl et al., 2005]. Because of warming at higher latitudes the amount and composition of terrigenic DOM exported to riverine systems are expected to change drastically [Dittmar and Kattner, 2003; Striegl et al., 2007].

[4] In permafrost-free basins, contributions of groundwater and soil water originating in riparian and hillslope ecosystems to rivers vary depending on the season and the proportion of wetlands [Hinton et al., 1998]. In permafrost terrains, progressively thawing permafrost (from May to September) causes increasingly deeper infiltration and correspondingly a longer residence time of solutes in soils resulting in DOM sorption and/or mineralization. Thus DOM pathways and sources differ substantially between seasons that produces changes in both DOM quality and quantity [MacLean et al., 1999; Prokushkin et al., 2001; Striegl et al., 2005, 2007; Neff et al., 2006].

[5] Dissolved organic matter is a very heterogeneous and complex mixture of organic molecules with different chemical characteristics including a polymeric structure of major constituents [Schulten and Gleixner, 1999]. The thermolytic pyrolysis technique is a useful tool for generating products that give information about the original mixture that composed the initial natural organic matter. In particular, recent application of pyrolysis/gas chromatography/mass spectrometry (Py/GC/MS) revealed distinct spatial [White et al., 2002] and temporal [da Cunha et al., 2000] heterogeneity of organic matter in riverine ecosystems. However, even more detailed information can be obtained from examination of compound-specific isotope ratios acquired by isotope ratio mass spectrometry (IRMS). Measurements of the stable isotopic composition of DOM provide an important tool to investigate the biogeochemical processes occurring in peat, soil, and sediments [Kracht and Gleixner, 2000]. The shift from analysis of bulk isotopic composition to compound-specific isotopic content allows molecular insights into soil carbon turnover [Gleixner et al., 2002]. In particular, Kracht and Gleixner [2000] reported distinctive isotopic signatures associated with the various biogeochemical processes that occur during the process of humification (e.g., biological degradation, selective preservation, trophic level effects, and introduction of microbial compounds). They showed that the combined analysis of structural and isotopic contents of source solid and dissolved organic matter by Py/GC/MS/IRMS is a powerful technique for elucidating the sources and fates of DOM in soils.

[6] We applied Py/GC/MS/IRMS analysis to DOM sampled from a forested watershed in a continuous permafrost region of Siberia. We have attempted (1) to examine the molecular and isotopic composition of terrestrial and aquatic DOM, (2) to characterize the alteration of DOM on its passage from forest floor to stream, and (3) to trace the terrestrial sources of DOM in the stream during different seasons and hydrologic phases and to estimate likely changes of DOM fluxes under global warming perspective in permafrost terrain of Siberia.

2. Material and Methods

2.1. Study Site

[7] Kulingdakan watershed (64°17–19′N, 100°11–13′E) is located northeast of central Siberia (Figure 1) and encompasses about 4,100 ha. It consists of forested hills with table-like tops and boggy valley bottoms. The watershed area (elevation from 132 to 630 m) is typical for the Syverma plateau. The region has a cold continental climate. The average temperature in January (the coldest month) is −36°C (Tura Meteorological Station, 1929–1998), and the average temperature for July (the warmest month) is 16°C. Average annual temperature is −9.5°C. Long-term average annual precipitation for the region is 354 mm. About 30–40% of annual precipitation falls as snow. The watershed is located on the western edge of the zone of continuous permafrost distribution in Siberia.

Figure 1.

Location of the research area (open square) on the territory of the Russia. DOM samples were collected in the Kulingdakan stream catchment (right) located approximately 5 km north of the town of Tura. Solid dots are sites for soil solution sampling on the north and south facing slopes, respectively. Open dot is sampling site of stream water.

[8] Study plots (10 m × 10 m each) were selected at midelevation on north and south facing slopes of 12–13° inclination, referred to as north and south slope, respectively. Sites were separated by a stream valley which collected the DOC input from both slopes. Vegetation is typical for central Siberian deciduous coniferous forests [Abaimov, 2005]. The overstory of both plots is formed by larch (Larix gmelinii (Rupr.) Rupr.) regenerated after a ground fire in 1902. Canopy closure, defined as the percentage of ground area shaded by overhead foliage, is 0.24 on the north slope and 0.53 on the south slopes. The ground vegetation is dominated by dwarf shrubs (Ledum palustre L., Vaccinium vitis-idaea L., and Vaccinium uliginosum L.) and mosses (Pleurozium schreberi (Brid.) Mitt., Hylocomium splendens (Hedw.) B.S.G., and Aulocomnium palustre (Hedw.) Schwaegr.) with patches of lichens (Cladina spp. and Cetraria spp.). Mosses (largely P. schreberi) form an acidic organic layer (pH 3.8–5.0) of 6.4 ± 2.5 cm (mean ± standard deviation, n = 6) thickness on the south facing slope and 13.0 ± 3.5 cm (n = 6) thickness on the north facing slope. Depth of the active soil layer, defined as the portion of the upper soil that thaws in summer, typically reaches a maximum in September of up to 0.4 m on north facing slope and about 1.0 m on south facing slope.

2.2. Materials

2.2.1. Forest Floor Leachates

[9] Solutes percolated through the entire organic layer which accumulated on the surface of mineral soil (including both living moss and decomposing organic material (O layer)) were collected by zero tension lysimeters (n = 6, area 95 cm2). The design of lysimeters was similar to those used by Hongve [1999, Auxiliary materials]. Each lysimeter was connected to 1 L PET bottles preconditioned sequentially with 0.1 M HCl, distilled water, and finally with leachate. For analyses of the chemical and isotopic composition of the leachates, combined samples from six independent replicates (on south and north slopes) yielding a total volume of 1.5 L were taken from one percolation event on 5 August 2003 following a moderate rainfall (Table 1). To simplify handling of sample names, we introduced a scheme describing source (O, organic layer), season of sampling (Su, summer), date (3, 5 August 2003) and location (N and S, north and south facing slopes, respectively). Thus abbreviation of forest floor leached DOM samples collected for analysis is as follows: O-Su-3-N and O-Su-3-S.

Table 1. Characteristics of Sampling Conditions and Organic Solutes Taken for Analysis by Py/GC/MS/IRMSa
DateAbbreviationStream DOM Phase DescriptionDOC, mgC L−1Q, m3 s−1P, mm
20 May 2003S-Sp-HSpring stormflow (snowmelt, descending limb)23.34.62
11 Jun 2003S-Sp-LSpring low flow24.00.490
18 Jul 2003S-Su-1-HSummer intense stormflow23.33.2427.1
24 Jul 2003S-Su-2-LSummer low flow14.90.250
5 Aug 2003S-Su-3-ISummer intermediate stormflow16.91.536.2
DateAbbreviationTerrestrial DOM Sample DescriptionDOC, mgC L−1P, mmP*, mm
  • a

    Q is stream discharge, P is precipitation, and P* is precipitation percolated through forest floor.

5 Aug 2003O-Su-3-SLeachate of forest floor (south facing slope)55.4 ± ± 0.2
5 Aug 2003O-Su-3-NLeachate of forest floor (north facing slope)40.1 ± ± 0.3
5 Aug 2003B-Su-3-NSoil solution (north facing slope)16.0 ± 1.36.2

2.2.2. Soil Solution

[10] Soil water that had passed through the mineral soil (B horizon) and traveling laterally above the permafrost (38 cm depth) was sampled on 5 August 2003 on the north slope site (B-Su-3-N) using zero tension lysimeters (n = 3). Soil solutions were sampled and combined up to give a total volume of 2 L for further analysis.

2.2.3. Stream Water

[11] Stream water was sampled and discharge was measured daily using previously published methods [Prokushkin et al., 2001, Auxiliary materials]. To analyze variability in the properties of riverine DOM under different hydrologic conditions and different seasons, five samples (2.5 L each) were taken from Kulingdakan stream during spring and summer periods (Table 1). Spring samples include a high-discharge sample taken during snowmelt (May 20 (S-Sp-H)) and a relatively low but stable discharge phase (June 11 (S-Sp-L)). Summer samples were taken during a period of intense stormflow (July 18 (S-Su-1-H)), low flow (July 24 (S-Su-2-L)), and intermediate stormflow (August 5 (S-Su-3-I)). Immediately after sampling in the field, all water samples were filtered using precombusted glass fiber filter (GF/F, Whatman International Ltd., Maidstone, U.K.). After a rinse of 100 mL, 200 mL were stored in a refrigerator at 3°C for further chemical analysis. For analysis of the chemical and isotopic composition of DOM, 1 L of leachate, 1.5 L of soil solution, and 2 L of stream water were reduced in a water bath at 40°C to a final volume of 20 mL, frozen, freeze-dried, homogenized, and kept solid in desiccators prior to analyses.

2.3. DOM Analysis

[12] Subsamples of 2–10 mL (depending on DOM concentration) were dried on water bath (40°C). Organic carbon content in dried material was determined by oxidation with added dichromate/sulfuric acid mixture (4.9 g dm−3 K2Cr2O7 and concentration H2SO4, 1:1 volume/volume) with colorimetric detection of the reduced Cr3+ [Vanchikova et al., 2006]. Light absorption of the solutions was measured using a colorimeter (KFK-3, ZOMZ, Zagorsk, Russia) at a wavelength of 590 nm [Prokushkin et al., 2001; Vanchikova et al., 2006] against a blank solution. For calibration, standard sucrose solutions (n = 10, 0.1–1.0 mg C, three replicates) were used. The standard deviation of the organic carbon analysis for sucrose standard was 0–1.6% (R2 = 0.998, p < 0.001) and 2.5–6.9% for samples analyzed in three replicates.

2.4. Pyrolysis/Gas Chromatography/Mass Spectrometry/Isotope Ratio Mass Spectrometry

[13] From four to six replicates of circa 2–3 mg of freeze-dried solid DOM samples were pyrolyzed in a Curie point pyrolyzer (type 0316, Fischer, Germany) using a ferromagnetic sample tube at 590°C for 9.9 s. The pyrolysis products were transferred online to the gas chromatography (GC). Following split injection (ratio 10:1, flow rate 2.0 mL min−1), the pyrolysis products were separated on a BPX 5 column (60 m × 0.32 mm, film thickness 1.0 μm, Scientific Glass Engineering, Weiterstadt, Germany) in a GC HP 5890 (Waldbronn, Germany) using a temperature program of 36°C for 5 min, then 5°C min−1 to 270°C followed by a jump (30°C min−1) to a final temperature of 300°C. The injector temperature was set to 250°C. The column outlet was coupled to a fixed splitter (ratio approximately 1:9). The larger split stream was transferred to a combustion furnace converting the pyrolysis products to CO2, N2, and H2O (CuO, NiO, and Pt, set at 960°C), and the δ13C values were determined using an isotope ratio mass spectrometer (DeltaplusXL, Finnigan MAT, Bremen, Germany). Dynamic background correction by ISODAT version 7.0 software was applied. The threshold intensity for peak detection at the IRMS was set to 200 mV. This value was found to optimize the number of peaks used for isotope determination and ion source linearity to minimize background effects and to reduce the isotopic influence of peaks that were not entirely separated at baseline.

[14] The smaller split stream of the GC eluates was transferred to an ion trap mass spectrometer (GCQ, ThermoQuest, Egelsbach, Germany). The transfer line was heated to 270°C and source temperature was held at 180°C. Pyrolysis products were ionized by electron impact with 70 eV ionization and identified by comparison with reference spectra using GCQ identification software version 2.2 and Wiley 6.0 mass spectra library on the basis of the mass range of m/z 25–450 and mass spectra obtained from Gleixner et al. [2002].

2.5. Calculations and Data Interpretation

[15] Relative abundance of single pyrolysis products was calculated using the m/z 44 trace of the IRMS calibrated against an external caffeine standard (VWR International, Germany). This carbon-related peak area was normalized to the pyrolyzed amount of sample. The δ13C values of individual substances were externally calibrated against the working standard caffeine (−51.8‰). Standard runs with caffeine were done before and after every sample. Linear correlation coefficients (Pearson r) were calculated for both the relative abundances and isotopic content of pyrolysis products identified in all samples (STATISTICA r.6, StatSoft, Inc.).

3. Results

3.1. Characterization of Dissolved Organic Matter by Py/GC/MS/IRMS

[16] Pyrolysis of polymeric organic matter produces monomers as follows: polysaccharides yield furans and carbonyl compounds, proteins produce pyrroles and indoles, polyhydroxy aromatics generate a range of phenolic substances as well as benzene derivatives, and lignin is precursor of methoxyphenols [Gleixner et al., 2002; Steinbeiß et al., 2006].

[17] Pyrograms of terrestrial and aquatic samples of DOM contained almost identical series of pyrolysis products (Figure 2) and suggested similar chemical composition of source materials. In the current study, 33 single compounds and combinations of different chemical origin were identified in the pyrolysates (Table 2). However, the number of substances determined in an individual sample varied from 25 to 32 (Table 3). The lowest number of pyrolysis products was found in forest floor leachate DOM. On the other hand, substances like 3-furaldehyde (see Table 2) and 5-methyl-2-furaldehyde were specific markers of DOM originated in forest floors. In stream water these two substances were found only during intense hydrologic “flushes” like snowmelt and intense stormflow. Both substances are polysaccharide breakdown products [Kracht and Gleixner, 2000; da Cunha et al., 2000; Steinbeiß et al., 2006]. In contrast, substances from the end of the pyrogram were attributed to the soil solution and stream samples, i.e., methylmethoxyphenol, methylindenes, ethylbenzofuran, indan-1-one, methylnaphthalene, and dimethylnaphtalin. These compounds are typical for pyrolysis products of degraded lignin and humic acids [Van Smeerdijk and Boon, 1987; Schulten and Gleixner, 1999].

Figure 2.

Pyrochromatograms of analyzed samples. Relative intensity of pyrolysis products was normalized to the peak area (mV) of toluene (7). Peak numbers follow identified substance numbers listed in Table 2. B-Su-3-N, soil solution (north facing slope); O-Su-3-S, leachate of forest floor (south facing slope); S-Sp-H, spring stormflow; S-Su-3-I, summer intermediate stormflow.

Table 2. Observed Retention Times and Characteristics of Pyrolysis Products Identified in Samples of Terrestrial and Aquatic DOM
No.Retention Time, minSubstanceMolecular WeightBase Peak, m/zCharacteristic Ions, m/zPrecursorsaFractionb
  • a

    Ps, polysaccharides; Pr, protein; Lg, lignin; Pp, polyphenol, nonlignin phenol; Ha, humic acids; Chitin, (C8H13O5N)n.

  • b

    PO, polysaccharides; Phe, phenolic compounds; AH, aromatic hydrocarbons; Ha, humic substances; N, nitrogen-containing compounds.

110:43Isocyanoethane555543  PrN
211:183-methylfuran82818253 PsPO
516:52Dimethyl-1,3-cyclopentadiene9479–10077–9091–55 PsPO
617:411H-pyrrol6767–10041–4050–15 Chitin/PrN
819:47Trimethylfuran11011010995 PsPO
920:133-furaldehyde9695–10067–1693–14 PsPO
1222:04C2-benzenes10691–100106–45  PpAH
1323:11Ethylbenzen + styren106 + 10491 + 10478–60  Lg/PrAH
1423:752-methyl-2-cyclopenten-1-one96679596 PsPO
1524:22Substitute phenol12212280107 PpPhe
1726:12Methyl-cyclopenten-one969581  PsPO
1826:36Phenol949466  Ps/Lg/PrPhe
1926:75C3-benzenes12010512079 Biodegraded ligninAH
2027:83C3-benzenes12010512079 Biodegraded ligninAH
2128:532,3-dimethylcyclopenten-1-one 1106795 PsPO
2228:952-methylphenol108637779108Biodegraded ligninPhe
2430:212-methoxyphenol12481109124 LgPhe
2530:95Dimethylphenol + C1-indol122 + 13113112277107Pp/Lg + PrPhe
2631:64Methylmethoxyphenol + C2-phenol148 + 122124/107   LgPhe
2732:10C2-phenol + C2-phenol122107122  LgPhe
2832:52Methylindene130129130115 HaHa
2932:71Methylindene + methylindene130129130115 HaHa
3036:97Indan-1-one13213278104 HaHa
3137:29Methylnaphthalene + C2-indol142 + 117142/117   Ha + PrHa
3237:50Ethylbenzofuran146131146  PsPO
3340:77Dimethylnaphthalin156141156  HaHa
Table 3. Observed Relative Abundance of Pyrolysis Products of DOM of Terrestrial and Aquatic Origin as Determined by Py/GC/MS/IRMSa
  • a

    Identification was performed by parallel Py/GC/MS measurements; S-Sp-H, spring stormflow; S-Sp-L, spring low flow; S-Su-1-H, summer intense stormflow; S-Su-2-L, summer low flow; S-Su-3-I, summer intermediate stormflow; O-Su-3-S, leachate of forest floor (south facing slope); O-Su-3-N, leachate of forest floor (north facing slope); B-Su-3-N, soil solution (north facing slope).

  • b

    These have coeluting peaks.

13Ethylbenzen + styren1.040.260.920.400.760.200.310.070.470.021.370.411.490.430.510.05
15Substitute phenol1.050.210.490.080.710.
25Dimethylphenol + C1-indol0.560.100.420.100.420.
26Methylmethoxyphenol + C2-phenol0.470.070.640.120.400.050.480.070.690.130.330.
27C2-phenol + C2-phenol0.740.120.790.200.690.110.650.
29Methylindene + methylindene0.760.130.690.110.570.020.410.110.670.
31Methylnaphthalene + C2-indol0.400.
 Total amount:33.8 31.8 29.9 15.2 24.0 39.8 43.7 23.5 

[18] Identified pyrolysis products were divided into five classes of precursors: polysaccharides (PO), aromatic hydrocarbons (AH), phenols (Phe), nitrogenous compounds (N), and humic substances (Ha). Breakdown products of polysaccharides were the most abundant fraction of pyrolysis products in all samples analyzed (Figure 3). The proportion of polysaccharides relative to all pyrolysis products varied from 51% in forest floor leachates to 39% in stream water and decreased at low stream discharges. A similar trend was found for aromatic hydrocarbons, which decreased from 21% at snowmelt to 15% during summer low flow. Derivatives of phenol were the second most abundant group of pyrolysis products (23–30%). In contrast to previous groups the proportion of phenols was found to increase at lower discharges. The group of humic substances followed the same trend. The proportion of nitrogenous compounds remained relatively constant in aquatic samples and was approximately 1.5 times lower than that of forest floor leachates.

Figure 3.

Relative proportions of pyrolysis fractions in analyzed samples. Pyrolysis product fractions: polysaccharides (PO), phenolic compounds (Phe), aromatic hydrocarbons (AH), humic substances (Ha), and nitrogen-containing compounds (N). S-Sp-H, spring stormflow; S-Sp-L, spring low flow; S-Su-1-H, summer intense stormflow; S-Su-2-L, summer low flow; S-Su-3-I, summer intermediate stormflow; O-Su-3-S, leachate of forest floor (south facing slope); O-Su-3-N, leachate of forest floor (north facing slope); B-Su-3-N, soil solution (north facing slope).

3.2. Terrestrial DOM

[19] Forest floor leachates contained highest total amount of pyrolysis product concentrations (Table 3). Surprisingly, slightly higher concentrations and total amount of pyrolysis products were measured in the north slope (43.7) relative to the south slope (39.8), which contrasts the DOM concentrations (40.1 and 55.4 mg C L−1, respectively). This disagreement is only partly explained by coeluting and unidentified peaks in the beginning of the pyrograms, which were omitted from the analysis (e.g., acetic acid (south slope forest floor leachate, retention time 740–780 s, Figure 2)). On the other hand, south slope DOM contained higher amounts of “fresher” (2-furaldehyde, 5-methyl-2-furaldehyde, and 2-methoxyphenol) and humified materials (e.g., methylmethoxyphenol and dimethylnaphthalin) comparatively to DOM from north slope. Despite this fact, DOM collected on north and south slopes demonstrated high Pearson correlation (r = 0.95, p < 0.01, Table 4) of the relative abundances of identified pyrolysis products suggesting similar chemical characteristics of sources of DOM on these climatically contrast slopes.

Table 4. Correlation Coefficients Obtained by Linear Correlation Analysis (Pearson r) for Relative Abundances of Identified Pyrolysis Products of DOM Samples (Significant (p < 0.01) for r ≥ 0.45)a
  • a

    S-Sp-H, spring stormflow; S-Sp-L, spring low flow; S-Su-1-H, summer intense stormflow; S-Su-2-L, summer low flow; S-Su-3-I, summer intermediate stormflow; O-Su-3-S, leachate of forest floor (south facing slope); O-Su-3-N, leachate of forest floor (north facing slope); B-Su-3-N, soil solution (north facing slope).

S-SP-L 0.840.920.930.500.610.91
S-Su-1-H  0.730.740.730.830.78
S-Su-2-L   0.980.400.470.96
S-Su-3-I    0.410.480.95
O-Su-3-S     0.950.46
O-Su-3-N      0.53

[20] The isotopic content of pyrolysis products from the leachates showed overall depletion of 13C in most substances of south slope forest floor leachate relative to the ones from the northern slope (Table 5). In contrast, δ13C values from 2-furaldehyde, 5-methyl-2-furaldehyde, 2-methoxyphenol, and benzene indicated an isotopic enrichment between 1.6–3.4‰ in the south slope. Moderate correlation between isotopic content of DOM collected on north and south slope (r = 0.69, Table 6) suggests specific decomposition processes producing a unique isotopic fingerprint, despite DOM being derived from almost identical source matter. Therefore more detailed analyses are needed to identify processes controlling production and release of DOM depending on local conditions.

Table 5. Isotopic Content (‰ Vienna Peedee Belemnite) of Pyrolysis Products of DOM of Terrestrial and Aquatic Origin as Determined by Py/GC/MS/IRMSa
  • a

    Identification was performed by simultaneous Py/GC/MS measurements; S-Sp-H, spring stormflow; S-Sp-L, spring low flow; S-Su-1-H, summer intense stormflow; S-Su-2-L, summer low flow; S-Su-3-I, summer intermediate stormflow; O-Su-3-S, leachate of forest floor (south facing slope); O-Su-3-N, leachate of forest floor (north facing slope); B-Su-3-N, soil solution (north facing slope).

  • b

    Mean calculated for substances that appeared in all analyzed samples.

3Methyl-1,3-cyclopentadiene−24.00.4−25.10.2−24.00.5−24.70.5−24.70.3  −22.61.1−25.80.5
4Benzene−26.40.7−24.80.1−27.41.0  −26.10.3−20.90.3−24.30.5−27.20.8
93-furaldehyde−25.40.9−26.61.4−24.90.6    −24.20.6−24.20.6  
13Ethylbenzen + styren−26.10.4−24.90.9−25.70.4−24.61.1−25.50.3−25.50.2−22.71.0−26.00.6
15Substitute phenol−24.90.2−21.91.1−21.80.7−28.60.3−21.91.0−25.30.7−24.70.6  
165-methyl-2-furaldehyde−23.80.8  −21.80.6    −22.50.4−23.50.4  
25Dimethylphenol + C1-indol−23.71.0−28.31.1−27.90.4−27.30.7−25.50.4−24.60.5  −26.50.5
26Methylmethoxyphenol + C2-phenol−26.20.8−25.60.4−24.30.3−26.80.8−25.70.7−23.60.4  −26.80.9
27C2-phenol + C2-phenol−26.01.0  −27.30.4−26.00.5−26.00.5−25.50.7  −27.70.8
28Methylindene−26.20.2−27.80.5−26.22.8−30.71.0−23.50.2    −24.30.5
29Methylindene + methylindene−25.20.8−27.90.6−25.52.6−27.60.4−23.70.3    −25.10.3
30Indan-1-one−23.90.3−25.60.3−26.50.7−25.90.8−25.10.4    −26.20.2
31Methylnaphthalene + C2-indol−24.20.2    −25.70.4−25.00.8    −25.80.2
32Ethylbenzofuran  −25.80.7  −25.50.1−24.90.7    −25.70.2
33Dimethylnaphthalin−25.60.4−27.40.4−26.60.5−25.30.2−24.90.6−25.91.6  −26.60.5
 Meanb−26.1 −26.0 −25.8 −26.0 −24.9 −26.0 −24.7 −26.4 
Table 6. Correlation Coefficients Obtained by Linear Correlation Analysis (Pearson r) for Isotopic Composition of Identified Pyrolysis Products of DOM Samples (Significant (p < 0.01) for r ≥ 0.54)a
  • a

    S-Sp-H, spring stormflow; S-Sp-L, spring low flow; S-Su-1-H, summer intense stormflow; S-Su-2-L, summer low flow; S-Su-3-I, summer intermediate stormflow; O-Su-3-S, leachate of forest floor (south facing slope); O-Su-3-N, leachate of forest floor (north facing slope); B-Su-3-N, soil solution (north facing slope).

S-SP-L 0.630.600.570.490.590.57
S-Su-1-H  0.380.610.480.750.69
S-Su-2-L   0.200.430.610.42
S-Su-3-I    0.430.500.86
O-Su-3-S     0.690.54
O-Su-3-N      0.70

[21] There was nearly a twofold decrease of the total relative abundances of pyrolysis products in soil solution compared to forest floor leachate sampled on the same day (August 5) on the north slope. When combined with the overall lower correlation (r = 0.53) of the relative abundances, it suggests that selective changes in the concentrations of particular substances occurred after contact of forest floor leachate with the mineral soil matrix.

[22] For instance, 2-furaldehyde demonstrates the highest decrease in relative concentration in soil solution (3.8-fold) as well as one of the highest depletions of δ13C values (3.5‰). The isotopic content of pyrolysis products indicated that initially, heavier substances, mainly attributed to the polysaccharide fraction, were depleted, whereas lighter compounds that originated from aromatic precursors were enriched in 13C (i.e., 2-methoxyphenol, Figure 4). This suggests preferential degradation of plant-derived markers and likely introduction of DOM from microbial biomass synthesized from soil CO2 that is depleted by 13C [Kracht and Gleixner, 2000]. In contrast to the selective removal of the majority of plant compounds in soil the δ13C values of isocyanoethane and 2,4-pentadienal demonstrate the preservation of the plant precursors of these molecules in soil, whereas the decrease of their total relative abundances suggests utilization by soil biota.

Figure 4.

(a) Dependences between δ13C values of pyrolysis products of soil (B-Su-3-N) and respective forest floor DOM products (O-Su-3-N), and (b) differences between δ13C values of soil and forest floor DOM pyrolysis products (B-Su-3-N and O-Su-3-N, respectively) plotted against the isotopic content of pyrolysis products of forest floor (O-Su-3-N) sampled on the north facing slope.

3.3. Stream Water DOM

[23] Total relative abundances of pyrolysis products of stream water DOM and their corresponding isotopic values varied significantly between seasons and hydrologic conditions. Highest relative contents were observed in spring. Both snowmelt and spring low-flow samples showed similarly high relative abundances of pyrolysis products of DOM with only slight decreases (circa 5%) in low-flow phase. However, the correlation analysis of isotopic content of pyrolysis products indicated a relatively low similarity between snowmelt and spring low-flow DOM (r = 0.60, p < 0.01).

[24] Summer stream samples showed 1.2 to 2.2-fold less total relative abundance of pyrolysis products compared to spring values. The lowest amount of pyrolysis products was found in summer low flow (15.2), while under increased discharges the sum of relative abundances almost doubled. This is supported by data showing positive relationships between discharge and stream DOM concentrations found earlier [Schiff et al., 1997; MacLean et al., 1999]. Linear correlation analysis of relative abundances of identified substances demonstrates either moderate (r = 0.74, intense stormflow versus low flow) or strong (r = 0.98, p < 0.001), intermediate stormflow versus low flow) correlation of summer stream DOM composition. The overall content of pyrolysis products under intense stormflow conditions increased twofold relative to low-flow value, and concentrations of toluene, 2-furaldehyde, 2,4-pentadienal, and phenol showed threefold to fourfold increases. These findings suggest that during runoff events the source of DOM entering the stream changes to material that is more enriched by plant-derived compounds. This observation is further supported by the low similarity between DOM at low flow and stormflow demonstrated by linear correlation of isotopic content (Table 6).

4. Discussion

4.1. Biogeochemical Processes Controlling DOM Formation in Soil

[25] The warmer location on south facing slopes has higher decomposition rates resulting in thinner organic layers accumulated on soil surface [Prokushkin et al., 2005] and almost doubled CO2 emissions from the soil (12.2 ± 3.9 and 7.0 ± 2.0 μM CO2 m−2 s−1, respectively for south and north facing slopes) (A. S. Prokushkin, unpublished data, 2005). Compounds from the end of the pyrogram of south slope DOM (i.e., methylmethoxyphenol and dimethylnaphtalin) as well as the higher concentrations of 2-methoxyphenol are indicative of more complete decomposition of plant material in the warmer environment. However, the average isotopic composition of pyrolysis products of DOM leached from forest floors on north slope demonstrates overall enrichment (1.3‰) of DOM by 13C relative to south slope values, suggesting that DOM leached from forest floors on colder north facing slopes originates from more degraded material [Kracht and Gleixner, 2000]. Moreover, the ratio of 2-furaldehyde/pyrrole considered as a mineralization index [da Cunha et al., 2000] was higher in DOM leached from the forest floor on south slope (7.3 versus 5.6), corroborating the lower degradation of organic matter there. Such conflicting information regarding decomposition rates and degradation conditions is likely due to the period of sampling. Midsummers in continental parts of Siberia are usually characterized by droughts, likely impeding microbial activity in upper soil of warm and well-drained sites like south facing slopes (topsoil moisture drops to 15–23%). In particular, Šantručková et al. [2003] demonstrated the decline of microbial communities within the permafrost zone of central Siberia at temperatures above +5°C. Our earlier findings also showed that DOC concentrations in forest floor leachates of south facing slope declined with increasing litter layer temperature in the range of 7–13°C [Prokushkin et al., 2005]. In contrast, DOC production in forest floor of cooler north facing slopes positively affected by increasing temperatures as a consequence of increasing microbiologic activity in moist environment.

[26] Despite overall 13C depletion of south facing slope DOM, most abundant substances (2-furaldehyde, 5-methyl-2-furaldehyde, and 2-methoxyphenol, known to originate from fresher plant tissues [Gleixner et al., 2002; Steinbeiß et al., 2006]) were enriched in 13C. Therefore source plant material on south slope might be enriched in 13C because of local environmental conditions, like water stress, that decrease discrimination against 13C in photosynthesis. Additionally, plant species might differentially contribute to the entire organic layer. Organic matter in the O layer of the south facing slope is composed to a larger extent of vascular plant litter (e.g., larch and ericaceous dwarf shrubs), while mosses contribute proportionally more to soil organic matter in the colder north facing slope. Acetic acid formed by the cleavage of acetyl groups from woody hemicelluloses [Sjostrom and Reunanen, 1990] were found in the south slope forest floor leachate (continuous large area peak (retention time 740–780 s, Figure 2)) providing an evidence of larger input of organic matter originating likely from larch litter.

[27] Soil solution DOM from north facing slopes is depleted in 13C (−26.4‰) relative to DOM from the forest floor (−24.7‰), presumably owing to preferential decomposition or adsorption of 13C-enriched substances [Schiff et al., 1997; Ludwig et al., 2003]. The latter process is thought to be unlikely because adsorption in soil occurs predominantly for hydrophobic compounds (i.e., phenol derivatives) with depleted 13C rather than 13C-enriched hydrophilic substances, which are mainly carbohydrates [Kaiser et al., 2001]. Despite a decrease in their relative abundance, 13C-depleted compounds demonstrate minor changes of isotopic content suggesting biochemical discrimination against more recalcitrant DOM in decomposition processes and preferential adsorption to soil. Nevertheless, isotopic examination of soil solution DOM demonstrates further enrichment of 2-methoxyphenol by 13C and preservation of 2,4-pentadienal suggesting specific pathways of utilization of their plant precursors in this environment. The introduction of humic material to soil solution constituting about 11% of the total amount of identified substances is another key feature of soil DOM formation. The narrow range of δ13C values in humic compounds of soil DOM suggests relatively similar chemical “history” of precursor molecules. Moreover, higher average values of δ13C in dissolved humus compared to the mean values of other compounds in soil solution (+0.87‰) are supported by earlier findings [Kaiser et al., 2001] that indicated higher 13C in DOM on its passage through the soil profile. The latter observation suggests that either trophic enrichment of precursors or isotopic fractionation during DOM decomposition is occurring. Thus further studies are necessary to trace initial composition of precursor molecules in site-specific vegetation and to understand the processes of DOM production, mobilization, and stabilization in permafrost-affected soils.

4.2. Reconstruction of DOM Sources in Stream

[28] In permafrost terrains the flow of snowmelt water is restricted to shallow upper organic layers of soil because of impermeability of frozen ground [Woo and Winter, 1993]. Respectively, DOM at snowmelt has shorter hydraulic residence and derives from upper organic-rich horizons composed from different organic materials, i.e., plant debris, lysed microbial cells, and humified material not adsorbed to soil [Jardine et al., 1989]. The presence of specific chemical markers of forest floor DOM in stream as well as isotopic composition of those substances (e.g., 2-furaldehyde and 5-methyl-2-furaldehyde) clearly demonstrates the importance of organic horizons as a source of stream DOM at this season. Correlation analysis of isotopic fingerprints, however, indicates also that this DOM is derived from multiple terrestrial sources. In particular, the presence of acetic acid in pyrolysates of snowmelt DOM indicates the introduction of fresh vascular plant-derived hemicelluloses. These findings support our earlier hypothesis suggesting the importance of litter fall from the previous autumn in driving DOC dynamics the following spring in larch-dominated systems [Prokushkin et al., 2005]. Likewise, δ13C values of pyrolysis products of snowmelt DOM are intermediate among forest floor and soil solutions suggesting a mixing of original and reworked material. Since soil microbial activity in this period is insufficient to produce high-DOM inputs to the stream, it is likely that only DOM abiotically produced in the freeze/thaw cycle is exported to the stream. This is supported by earlier findings showing that less-degraded organic matter enters riverine ecosystems during spring floods [Striegl et al., 2005, 2007; Neff et al., 2006].

[29] Concentrations of DOM in the stream during spring low flow remained high, but water flow paths varied by location. The main source of water in the stream during spring low flow is the melting of ice in upper soil. In valleys and north facing slopes the thawing process occurs in organic soil layers having extremely high moisture content (450–960%). In contrast, on south facing slopes the active layer thickness has already reached a depth of 25–30 cm of mineral soil, where soil moisture content is relatively low (<40%). Since higher water retention occurred in clayey soils, the contributions of south facing sites to the stream is likely negligible. The lower correlation of relative abundances and isotopic contents of DOM pyrolysis products between south slope and stream DOM during spring low flow provides further support for this hydrologic scenario.

[30] Despite significant differences between spring and midsummer low flows in hydroclimatic conditions, active layer thickness, and DOM concentrations, there is a similarity in DOM composition among these samples. This finding suggests a similar source of organic matter for stream DOM at lower discharges. Since stream low flow demonstrates higher correlations (relative abundances and isotopic contents) with soil solution of cooler north facing slopes, lateral hydrologic flow paths in locations with shallow active layers (i.e., north and east facing slopes) likely provide the main source of water and dissolved organic matter in stream.

[31] Rainfalls of intermediate intensity (e.g., 6 mm) cause the passage of water through the forest floor and ultimately through the entire soil profile down to the permafrost table. The lack of forest floor chemical markers in the stream and overall low correspondence of isotopic composition of DOM between forest floor leachates and stream support this hydrologic interpretation. The highest correlation between terrestrial DOM and stream DOM under these conditions is also shown for soil solution sampled on north facing slope. It supports our earlier idea that atmospheric water from lower-intensity rainstorms is retained in the soil of dry south facing slopes and then lost by evapotranspiration. In contrast, colder and wetter sites release Corg to streams.

[32] Therefore, under global warming perspective, DOM produced in the forest floor might be preserved and stabilized in deepen active layer of soil because of low summer precipitation. Higher accumulation of Corg in mineral soil in south slope site compared to north slope site (2.36 versus 1.72 Mg C m−3 in upper 0.5 m) may support this. On the other hand, the increased residence time of DOM in soil results in its increased microbial mineralization and consequently in reduced export of terrigenous Corg in riverine systems [Carey, 2003; Neff et al., 2006]. δ13C enrichment of stream DOM compared to soil solution (1.5‰) suggests a release of microbially altered organic matter. In an earlier study, Kawahigashi et al. [2004] demonstrated the increase of concentrations of microbial products in stream DOM with the shift from continuous to discontinuous permafrost. Therefore the increased depth of active soil layer promotes the retention of plant-derived DOM in soil and favors the export of microbially derived organic compounds.

[33] A heavy rainstorm (>20 mm) provoked significant flush of stream discharge (3.24 m3 s−1). Under such hydrologic conditions the main portion of water in stream passed through the upper organic-rich forest floor only. The presence of a specific chemical fingerprints (e.g., 5-methyl-2-furaldehyde) unique to forest floor DOM in stream DOM provides clear evidence of this flux. Interestingly, the isotopic composition of these substances is similar to forest floor leachates of the south facing slope, despite the fact that correlation analysis showed higher correspondence of stream DOM to the north slope sources: soil solution and forest floor leachate (Tables 4 and 6). These findings emphasize that multiple sources of DOM enter the stream during large pulses of discharge and the crucial role that is played by rainstorm intensity in modulating the partitioning of DOM fluxes in the stream among different landscape sources.

5. Conclusions

[34] The results of combined structural and isotopic analyses indicate that terrestrial and stream DOM contained a similar series of pyrolysis products, suggesting a terrestrial origin for DOM in the stream draining our study catchment. However, despite the overall similarity of chemical composition of stream DOM at different seasons, we also observed distinct differences in isotopic fingerprint between seasons and hydrologic phases (stormflow versus low flow). This variation appears to be due to the changing origin of stream DOM from different soil layers and the catchment sources following permafrost thawing. This fact clearly demonstrates that the quantitative analysis of DOM pyrolysis products used to fingerprint the origin of stream DOM in earlier investigations [e.g., da Cunha et al., 2000; White et al., 2002] may not be sufficient enough to fully elucidate stream DOM sources because of significant (bio)chemical alteration of many of the identified compounds. Thus examination of compound-specific isotope ratios is crucial for understanding the processes in the terrestrial environment that control DOM production, mobilization, and export to aquatic ecosystems. In addition, use of Py/GC/MS/IRMS overcomes the limitations of isotope applications for DOM tracer studies because of the previously identified problems associated with low spatial and temporal variation in the 13C signal of bulk DOM [Schiff et al., 1997].

[35] Terrestrial and aquatic DOM demonstrated enrichment by polysaccharide and phenolic structures, originating from lignocellulose decomposition products. This is attributed to relatively fresh organic matter entering the riverine system because of rapid surface runoff of terrestrial DOM in permafrost terrains. Deeper infiltration of organic solutes and higher retention time in soil cause a decrease of DOM concentrations in stream and an alteration of its chemical composition. In particular, isotopic composition of DOM suggests higher input of microbially derived material.

[36] On the other hand, under current conditions, composition and functional characteristics of stream DOM are mainly controlled by shallow soils developed in colder and wetter positions in landscape. Deeper soils found in warmer and well-drained southern aspects effectively retain both water and DOM, resulting in lower export of both to rivers. Such findings suggest that warming and consequent deepening of the active layer will change not only the chemical composition of DOM entering rivers in permafrost terrains but also significantly decrease the summer export of water and dissolved organic matter from forested watersheds underlain by continuous permafrost in Siberia.


[37] This research was supported by a stipend of the Deutscher Akademischer Austauschdienst (DAAD) for A. Prokushkin to visit the Max Planck Institute for Biogeochemistry, Jena (Germany) to perform Py/GC/MS/IRMS analyses and by the Russian Fund of Basic Research (grant 05-05-64208) for field investigations and sample collection. We acknowledge and appreciate the work of two anonymous reviewers, who provided critical and very helpful comments.