Methane turnover and environmental change from Holocene lipid biomarker records in a thermokarst lake in Arctic Alaska

Abstract

Arctic lakes and wetlands contribute a substantial amount of methane to the contemporary atmosphere, yet profound knowledge gaps remain regarding the intensity and climatic control of past methane emissions from this source. In this study, we reconstruct methane turnover and environmental conditions, including estimates of mean annual and summer temperature, from a thermokarst lake (Lake Qalluuraq) on the Arctic Coastal Plain of northern Alaska for the Holocene by using source-specific lipid biomarkers preserved in a radiocarbon-dated sediment core. Our results document a more prominent role for methane in the carbon cycle when the lake basin was an emergent fen habitat between ~12,300 and ~10,000 cal yr BP, a time period closely coinciding with the Holocene Thermal Maximum (HTM) in North Alaska. Enhanced methane turnover was stimulated by relatively warm temperatures, increased moisture, nutrient supply, and primary productivity. After ~10,000 cal yr BP, a thermokarst lake with abundant submerged mosses evolved, and through the mid-Holocene temperatures were approximately 3°C cooler. Under these conditions, organic matter decomposition was attenuated, which facilitated the accumulation of submerged mosses within a shallower Lake Qalluuraq. Reduced methane assimilation into biomass during the mid-Holocene suggests that thermokarst lakes are carbon sinks during cold periods. In the late-Holocene from ~2700 cal yr BP to the most recent time, however, temperatures and carbon deposition rose and methane oxidation intensified, indicating that more rapid organic matter decomposition and enhanced methane production could amplify climate feedback via potential methane emissions in the future.

Keywords

carbon cycle climate change Holocene Thermal Maximum lipid biomarkers methane oxidation stable isotopes thermokarst lakes

Introduction

With at least 500 Gt (10¹⁵ g) of carbon stored in permafrost soils (Zimov et al., 2006), continued warming of the Arctic has the potential to provoke a positive feedback response, whereby once-frozen permafrost carbon is made available to decomposition, leading to greater emissions of major greenhouse gases such as methane from high-latitude wetlands. Arctic lakes and wetlands have been identified as substantial sources for the atmospheric methane budget (Brosius et al., 2012; Walter Anthony et al., 2006, 2007). In general, atmospheric methane concentrations are affected by abrupt climate change as recorded in ice cores that provide global scale responses of methane reservoirs (Brook et al., 2000; Fischer et al., 2008; Flückiger et al., 2002; Sowers, 2010). Although short-term studies monitoring methane emissions from individual Arctic lakes for up to several years have assessed seasonal differences (Walter Anthony et al., 2006, 2008), profound knowledge gaps remain regarding past methane dynamics and potential emissions from lakes and wetlands in the Arctic and its relationships to climate change. Predicting the contemporary climate change response in the Arctic requires a basic understanding of the relationship between methane dynamics and prior climatic and environmental variations (e.g. temperature and hydrology), especially during climate warming events such as the last deglaciation and the Holocene Thermal Maximum (HTM; Jones and Yu, 2010; Kaufman et al., 2004; MacDonald et al., 2006; Yu et al., 2013).

Variations in methane emissions from terrestrial ecosystems are closely linked to hydro climatic conditions. Much of the Arctic is covered in low-relief permafrost terrain (i.e. ground that remains frozen for several years) where local interactions between vegetation, hydrology, and geomorphology moderate how the regional climate affects these belowground conditions (French, 2007). At any given site, a range of terrestrial, wetland, and lake ecosystems and the biophysical conditions of their soils will cycle back and forth over millennia (Hinkel et al., 2003; Jorgenson and Shur, 2007). Surface water pools commonly coalesce into thermokarst lakes that expand and deepen by thaw-driven erosion and ground subsidence (Jorgenson and Shur, 2007). Eventually, lakes no longer freeze solid in winter, thus creating floating-ice conditions (Arp et al., 2011). Thermokarst lakes will often drain to lower areas, at which point their basins become wetlands again and their sediments refreeze (Jorgenson and Shur, 2007). How past climate change influences this cycle and how they both interact to affect methane emissions is a critical step to understanding how the Arctic landscape will respond to future warming. Multi-proxy paleo studies that incorporate climate, ecosystem, and methane dynamics can be a first step in assessing this issue.

Previous investigations have utilized the stable carbon isotope composition (expressed as δ¹³C values) of aquatic invertebrates (e.g. chironomid head capsules or cladoceran ephippia) preserved in lake and wetland sediments as a proxy for past methane availability and/or release (Eller et al., 2005; Van Hardenbroek et al., 2010; Wooller et al., 2009, 2012). The basis for this proxy is that incorporation of ¹³C-depleted methane carbon (δ¹³C < −40‰; Jones and Grey, 2011) via methanotrophic (methane-metabolizing) bacteria into the biomass of aquatic microbes and invertebrates yields δ¹³C values of sample specimen that are lower than other primary production sources, such as terrestrial plants or aquatic algae that use C₃ photosynthesis (δ¹³C = −20‰ to −36‰; Lambers et al., 1998). Such studies are of great importance because they indirectly serve as records of methane flux intensity change over time. However, a more direct proxy for methane availability and flux is the abundance and ¹³C-depletion of preserved lipid biomarkers derived from sedimentary aerobic methanotrophic bacteria that are the major players involved in methane oxidation in soils, lakes, wetlands, and peat (He et al., 2012; Le Mer and Roger, 2001; Whalen, 2005). For example, variations in the abundance of specific biomarkers of aerobic methanotrophs (e.g. diploptene, diplopterol) have been observed during the Holocene (Atahan et al., 2007; Kristen et al., 2011; Zheng et al., 2014), providing evidence for changes of past methanotrophic activity. And although these lipid biomarker proxies cannot be used as a quantitative measure of methane emissions they are strong indicators of an intensified methane cycle in former times (e.g. Hinrichs et al., 2003).

The stages of the permafrost landscape cycle are associated with different sediment facies and plant types, thus offering the potential to characterize a timeline of this cycle with plant remains and lipid biomarkers in the sedimentary record (Jones et al., 2012; Jorgenson and Shur, 2007; Pancost et al., 2002). Specific lipid biomarker compounds identified in such environments include n-alkanes from terrestrial and aquatic plants, including submerged mosses, and specific ratios thereof (Nichols et al., 2009, 2014; Pancost et al., 2002; Zheng et al., 2014), sitosterol from submerged mosses (Liebner et al., 2011) and triterpenoids of higher plants, including Ericaceae (Pancost et al., 2002). Additional paleoclimatic information is provided by analysis of branched glycerol dialkyl glycerol tetraethers (brGDGTs) as a proxy for annual mean air temperature (MAT; Peterse et al., 2012; Weijers et al., 2007) and mean summer air temperature (MST; Foster et al., 2016; Pearson et al., 2011), and stable hydrogen isotope analysis of n-alkanes as a proxy of hydroclimatic changes (Nichols et al., 2009, 2014). Combining these independent records can serve as an integrated climate-landscape narrative to compare with methane turnover indices from the same site.

To better understand the Holocene evolution of methane carbon turnover and environmental conditions in Arctic Alaska, we analyzed preserved lipid biomarkers in a radiocarbon-dated sediment core representing the last ~12,000 years from Lake Qalluuraq (Wooller et al., 2012), a thermokarst lake on the Arctic Coastal Plain. Our data set of source-specific biomarkers (i.e. methanotrophic bacteria, submerged mosses, higher plants, aquatic algae, and soil bacteria) gives insight into regulative controls of paleo methane dynamics, temperature change, and carbon deposition from a high-latitude lake environment in relation to important periods such as the HTM and the more recent neoglaciation. The HTM has been timed from ~11,300 to ~9100 cal yr BP for Central-eastern Beringia (Kaufman et al., 2004) and typifies key aspects of the nature of climatic change and ecological succession in Arctic regions under warmer conditions and may serve to develop a forecast for Earth’s future climate.

Materials and methods

Study area

The study site, Lake Qalluuraq, is located at 70°22.739′N, 157°20.861′W on the Arctic coastal plain in Alaska, where methane-derived carbon flow through aerobic methanotrophs into the broader microbial community is an important water column process (He et al., 2012, 2015). The lake water depth is ~2 m and the morphology of the lake basin indicates that Lake Qalluuraq was previously much larger and deeper than present. The lake was likely drained by lateral migration of the Meade River at an unknown time in the past (Wooller et al., 2012). The regional climate is Arctic in nature and characterized by relatively cold summers (~4°C) and warm winters (~−25°C) compared with inland sites because of its proximity to the ocean (Zhang et al., 1996). Nevertheless, Lake Qalluuraq is ice-covered during the winter period (October to May) and ice-free during the summer period (June to September). Sedimentation of organic matter primarily occurs during the summer when aquatic primary production is highest. In addition to algae, submerged mosses make up a substantial fraction of organic matter deposited (Wooller et al., 2012). The wetland habitat in the now-exposed surrounding lake basin is characterized by string- and flark-type peatland. The surrounding landscape has a high density of thermokarst lakes among a mosaic of moist tussock-sedge tundra with erect dwarf shrubs (Ericaceae, Betula nana, Salix spp.) on moderately drained uplands, and wet sedge tundra (Carex aquatilis, Eriophorum angustifolia) and Sphagnum spp. moss in drained lake basins and poorly drained lowlands. Willow shrubs (Salix spp.) sporadically grow on sandy floodplains and lake shoreline blowouts (Raynolds et al., 2006).

Sample collection

For our paleoecological biomarker study, a 160-cm sediment core was taken from the 2-m-deep western basin of Lake Qalluuraq (Wooller et al., 2012). The core was collected in a polycarbonate tube (7-cm internal diameter) using a percussion corer, split lengthwise, photographed, and described. The upper, 80-cm-long, organic-rich section of the core was sampled in 1-cm intervals for biomarker analysis; samples were stored at −20°C until extraction.

Core description and radiocarbon ages

Radiocarbon dates on seven plant macrofossil samples (wood, moss, graminoids) were previously used to construct an age–depth model for the sediment record (Wooller et al., 2012). The most distinct lithological transition in the core was from coarse, gray sand to dark peat at ~80-cm sediment depth (~12,300 cal yr BP) and marks the initiation of a wetland ecosystem at the site (Wooller et al., 2012). Two peat layers from the core at ~80 to 44 cm and 34 to 25 cm dated to last from ~12,300 to ~10,000 cal yr BP (peat layer I) and from ~6800 to ~3000 cal yr BP (peat layer II), respectively. These two peat layers have higher mean contents of total organic carbon (TOC; 12.4% and 8.6%, respectively) than the intervening sediments (~5.5%; see Figure 3 in Wooller et al., 2012). Over the most recent part of the record, TOC increased from ~2% to 12%. Sediment carbon isotope values (δ¹³C) are highest (−23.0‰) in peat layer I and continuously decrease to the top of the core (−26.7‰).

Lipid biomarker extraction, separation, and analysis

Biomarkers were extracted either from single-centimeter intervals (hydrocarbons, alcohols) or pooled samples encompassing 2- to 4-cm-thick sections (brGDGTs). According to the previously published chronology for the core (Wooller et al., 2012), this results in temporal resolution of 168–303 years during the emergent fen period ranging from ~12,300 to ~10,000 cal yr BP and of 650–915 years during the thermokarst lake period that started at ~10,000 cal yr BP. In both cases, freeze-dried and homogenized sediment (8–32 g) was extracted using a Dionex accelerated solvent extractor (ASE 200) with a mixture of dichloromethane (DCM):methanol (MeOH) 3:1 (v/v, four cycles of 5 min each) at 100°C and 7.6 × 10⁶ Pa. Before extraction, known amounts of n-hexatriacontane, behenic acid methylester, n-nonadecanol, and 2-methyl-octadecanoic acid were added as internal standards. All total lipid extracts (TLEs) were dried using a Turbovap LV (Zymark Corp.) at 35°C under a nitrogen stream.

Dried TLEs of single-centimeter horizons were separated in DCM-soluble asphaltenes and n-hexane-soluble maltenes. The maltenes were separated by solid phase extraction (Supelco LC-NH₂ glass cartridges; 500 mg sorbent). Four fractions of increasing polarity (hydrocarbons, ketones, alcohols, and fatty acids) were obtained by elution with 4 mL n-hexane, 6 mL n-hexane:DCM 3:1 (v/v), 7 mL DCM:acetone 9:1 (v/v), and 8 mL 2% formic acid in DCM (v/v). For brGDGT analyses, dried TLEs of pooled sample horizons were separated into apolar and polar fractions by column chromatography on silica (3 g) using 15 mL n-hexane:DCM 95:5 (v/v) and 15 mL DCM:MeOH 1:1 (v/v), respectively. Apolar fractions were stored for future compound-specific radiocarbon age determinations, whereas polar fractions were subjected to analysis of brGDGTs. All biomarker fractions were dried under a gentle stream of nitrogen and stored at −20°C until analysis.

Hydrocarbon and alcohol fractions were analyzed via gas chromatography and flame ionization detection (GC-FID) and GC coupled to mass spectrometry (GC-MS). For quantification by GC-FID, hydrocarbons were dissolved in 500 µL n-hexane and spiked with squalane as injection standard. Alcohols were converted to trimethylsilyl (TMS) ethers using bis(trimethylsilyl)trifluoracetamide in pyridine before analysis. They were dissolved in 500 µL n-hexane and quantified relative to the internal standard n-nonadecanol. GC-FID analysis was performed using a ThermoQuest Trace GC equipped with a 30-m Restek Rxi-5MS fused silica capillary column (0.25-mm ID, 0.25-µm film thickness). Biomarker fractions were injected in splitless mode via an autosampler (1 µL) at 310°C and the carrier gas was He with a flow rate of 1 mL min⁻¹. The GC temperature program used was injection at 60°C, 1 min isothermal; from 60 to 150°C at 10°C min⁻¹; from 150 to 310°C at 4°C min⁻¹; 30 min isothermal giving a total run-time of 80 min.

Peak identification was based on relative retention times in parallel to mass spectrometric identification via GC-MS. The GC-MS (ThermoFinnigan Trace GC coupled to a Thermo Scientific DSQ II MS) was operated in electron impact (EI⁺)-mode at 70 eV with a full mass scan range of m/z 40–900 and 1.5 scans per second. The interface temperature was set to 310°C and He was used as carrier gas at a constant flow rate of 1 mL min⁻¹. Injection mode, capillary column, and the temperature program were identical to that of GC-FID analysis.

The average chain length (ACL) of n-alkanes (C₂₁ to C₃₁) was calculated based on quantified compounds according to

ACL = \frac{(21 * C_{21} + 23 * C_{23} + 25 * C_{25} + 27 * C_{27} + 29 * C_{29} + 31 * C_{31})}{(C_{21} + C_{23} + C_{25} + C_{27} + C_{29} + C_{31})}

Carbon isotopic compositions of hydrocarbons and alcohols were determined in duplicate using a Trace GC Ultra coupled via a GC combustion interface III to a Delta plus XP isotope ratio mass spectrometer (GC-IRMS, all ThermoFinnigan). The GC was equipped with an identical capillary column and programmed in the same way as used for GC-FID and GC-MS analysis. Injection was in splitless mode and the flow rate of He carrier gas was 1 mL min⁻¹. Stable carbon isotope values are reported in the delta-notation as δ¹³C values relative to Vienna Pee Dee Belemnite (VPDB) with a precision of better than 1‰. All δ¹³C values of hydrocarbons were referenced to the co-injected standard squalene and its known isotope value (−19.9‰). δ¹³C values of sitosterol, analyzed as TMS ether in alcohol fractions, were corrected for additional carbon atoms introduced during the derivatization reaction and referenced to the internal standard n-nonadecanol (δ¹³C = −28.4‰).

Stable hydrogen isotopic compositions of hydrocarbon biomarkers were measured in duplicate with the same GC-IRMS setup and parameters described above but with the exception of a pyrolysis furnace operated at 1440°C that quantitatively converts eluting compounds to graphite, H₂, and CO. The contribution of $H_{3}^{+}$ was determined daily by injecting pulses of H₂ gas with a constant isotope value and peak heights varying over a 10-fold range, and automatically implemented during data acquisition via Isodat NT 2.0. Stable hydrogen isotope values are reported in the delta-notation as δD relative to the Vienna Standard Mean Ocean Water (VSMOW) with a precision of better than 3‰. All δD values were referenced to the co-injected standard squalene and its known isotope value (−169‰). Daily precision was checked by injecting a lab-based n-alkane mixture (C₂₀ to C₂₉, C₃₁, C₃₃) with concentrations varying over a fivefold range.

An aliquot (10%) of each polar lipid fraction from the pooled sediment samples was dissolved in 200 µL n-hexane/isopropanol (99:1, v/v) and separated with a Prevail Cyano column (2.1 mm × 150 mm, 3 µm; Grace, Germany) maintained at 35°C in an Agilent 1200 series high performance liquid chromatography (HPLC) instrument. Using a flow rate of 0.2 mL min⁻¹, the gradient of the mobile phase was first held for 5 min with 100% eluent A (n-hexane/isopropanol, 99:1 (v/v)), followed by a linear gradient to 91% A and 9% B (n-hexane/isopropanol, 90:10 (v/v)) in 30 min. Reconditioning of the column was achieved by back-flushing eluent B for 10 min. After analysis, the column was equilibrated with 100% A at 0.2 mL min⁻¹ for 15 min. Detection was achieved with an Agilent 6130 MSD single quadrupole mass spectrometer coupled to the HPLC instrument via a multimode ion source operated in atmospheric pressure chemical ionization (APCI) mode. APCI settings were nebulizer pressure of 60 psi, vaporizer temperature of 250°C, drying gas (N₂) flow of 6 L min⁻¹, drying gas temperature of 200°C, capillary voltage of 2 kV, and corona current of 5 µA. Using Chemstation software, the detector was set for two signals: signal one was a scan from m/z 500 to 2000 and signal two was set to selective ion monitoring (SIM) of [M + H]⁺ ions (m/z 1018, 1020, 1022, 1032, 1034, 1036, 1046, 1048, and 1050; fragmentor voltage of 190 V).

To assess the entire temperature variation in our record, we applied two temperature reconstructions, one representing MAT and the other MST. Based on SIM peak areas of extracted ions, the cyclisation ratio (CBT) and the methylation index (MBT) of brGDGTs were calculated and trends in MAT were estimated according to Weijers et al. (2007)

MAT = \frac{(MBT - 0.122 - 0.187 * CBT)}{0.02}

The analytical errors of the MBT index and the CBT ratio were determined at ±0.01 units, leading to a variance in MAT estimates of ±0.6°C. MST estimates, by contrast, were calculated based on an expanded global recalibration of fractional abundances of specific brGDGTs, previously defined by Pearson et al. (2011), according to Foster et al. (2016)

MST = 32.5 + (85.4 \times GDGT - Ib) - (43.1 \times GDGT - II) - (32.1 \times GDGT - III)

Proxy temperature validity check

Both proxy temperature determinations were checked against modern weather data collected between 1981 and 2010 at Barrow, located north of Lake Qalluuraq (accessible at http://www.currentresults.com/Weather/Alaska/average-annual-temperatures.php#e). For our youngest sample, a reconstructed MAT of −7.0°C is comparable to today’s measured regional MAT (between −8 and −14°C). The reconstructed MST of 3.8°C for the same sample coincides with the instrumental mean of June to August temperatures (3.7°C).

Although there are additional calibration models for brGDGT-based temperature reconstructions for soils (Peterse et al., 2012) and lakes (e.g. Loomis et al., 2012; Pearson et al., 2011), only the Weijers et al. (2007) and Foster et al. (2016) models resulted in reasonable MAT and MST reconstructions, respectively. The applicability of the original Weijers et al. (2007) calibration model to high-latitude environments has been validated earlier by Peterse et al. (2014) who studied lakes in the Canadian and Siberian Arctic. The other calibration models either gave much higher or even inverse temperature profiles, identifying that there is no general calibration model and that more regional calibrations have to be developed. Nonetheless, for MAT reconstructions, we apparently do not observe a strong warm-temperature bias that is sometimes found in soils, which may result from a combination of the insulating effect of vegetation and the heat capacity of soil water (Weijers et al., 2011). The slight positive divergence in MAT compared with weather data from Barrow is likely the result of in situ production of specific brGDGT compounds, for example GDGT-III, in lakes during the summer seasons as inferred from tropical (Loomis et al., 2011; Tierney and Russell, 2009) and other Arctic lakes (Pearson et al., 2011; Shanahan et al., 2013a). On the other hand, extraction of characteristic brGDGTs and application of the Foster et al. (2016) model obviously resulted in reasonable reconstructed MST. Accordingly, since both temperature proxy reconstructions are partly based on different brGDGTs and calibration models, we presume that relative temperature variations in our record are valid.

Results

Lipid biomarker data and distribution

Hydrocarbon concentrations

The hydrocarbon fractions from Lake Qalluuraq are characterized by substantial amounts of vascular plant wax-derived long-chain odd C₂₁ to C₃₃ n-alkanes, higher plant-derived triterpenoids (taraxer-14-ene, friedel-3-ene), and bacterial hopanoids (e.g. hop-17(21)-ene; Figure 1, see Table 1 for biomarker source assignments). Lipid biomarker concentrations are in the µg g⁻¹ dry weight (dw) range with highest amounts of up to 138 µg g⁻¹ dw for n-C₂₇ found at the surface of the core (Table 2). Concentrations tend to decrease downcore but increase slightly between 28 and 36 cm (~4400 and ~7600 cal yr BP), corresponding to peat layer II (Figure 2). Distributions of n-alkanes peak at n-C₂₇ with higher amounts of n-C₂₃ in samples younger than ~8800 cal yr BP (Figures 1a, 2d), resulting in values of ACL from 25.6 to 26.0 (Table 2). Older samples show higher ACL values between 26.3 and 26.9. Triterpenoid and hopanoid hydrocarbon concentrations are in general lower than those of the n-alkanes, ranging from 0.11 to 12.9 µg g⁻¹ dw, but show a similar downcore distribution with increased amounts in peat layer II. We also observe a diunsaturated highly branched isoprenoid (HBI) with 25 carbon atoms (C_25:2) that is more prominent at the base of the core (Table 2, Figure 1b).

Figure 1.

Representative GC-FID chromatograms of hydrocarbons in Lake Qalluuraq samples at (a) 16-cm sediment depth in the late Holocene (~1900 cal yr BP) and (b) 68-cm sediment depth in the early Holocene (~11,300 cal yr BP).

Table 1.

Biomarker compounds and their input sources studied in Lake Qalluuraq sediments.

Biomarker(s)	Input source(s)	References
Highly branched isoprenoids (i.e. C_25:2)	Freshwater diatoms	Volkman et al. (1994); Belt et al. (2001)
Long-chain n-alkanes (n-C₂₁ to n-C₂₇, maximizing at n-C₂₃)	Aquatic submerged plants, including mosses	Ficken et al. (2000); Mügler et al. (2008)
Long-chain n-alkanes (n-C₂₃ to n-C₃₃, maximizing at n-C₂₇)	Terrestrial vascular plants, including sedges	Eglinton and Hamilton (1967)
Triterpenoids (Taraxer-14-ene, Friedel-3-ene)	Higher plants, including Ericaceae	Ageta and Arai (1983); Pancost et al. (2002)
Hopanoids (e.g. Hop-17(21)-ene)	Aerobic bacteria, including methanotrophs	Birgel and Peckmann (2008)
Phytosterols (i.e. Sitosterol)	Higher plants, including mosses; aquatic algae	Volkman (1986); Liebner et al. (2011)
Branched glycerol dialakyl glycerol tetraethers	Anaerobic soil bacteria; aquatic bacteria	Weijers et al. (2007); Tierney et al. (2010)

Table 2.

Downcore abundance of hydrocarbon biomarkers in µg g⁻¹ dw and n-alkane-derived ACL in Lake Qalluuraq sediments. Dashed line separates samples of the emergent fen (before ~10,000 cal yr BP) from the thermokarst lake period.

Sample depth (cm)	Age (cal yr BP)	C_25:2	C₂₁	C₂₃	C₂₅	C₂₇	C₂₉	C₃₁	ACL	Taraxer-14-ene	Hop-17(21)-ene	Friedel-3-ene
1	64	0.81	43.5	118	115	138	62.0	33.5	25.6	11.8	5.28	8.96
4	433	0.26	18.4	49.9	45.8	53.5	26.6	16.1	25.7	5.20	4.58	4.06
8	926	0.21	15.7	43.6	38.2	46.2	24.5	15.6	25.7	3.77	3.47	3.15
12	1418	0.24	17.6	48.3	40.9	49.6	26.3	15.6	25.7	4.25	3.78	3.16
16	1911	0.16	8.82	24.6	22.1	26.1	13.8	8.29	25.7	3.61	2.52	3.38
24	2896	0.17	9.31	26.2	23.2	27.7	14.2	8.63	25.7	3.84	2.88	2.81
28	4371	0.40	21.0	55.3	44.7	57.2	33.1	19.9	25.7	7.65	5.28	5.06
32	5987	0.54	14.6	40.7	36.0	45.5	28.1	18.7	26.0	12.9	3.57	6.71
36	7603	0.41	8.31	24.4	28.0	35.2	19.5	10.8	26.0	6.26	3.43	4.01
40	9218	0.24	3.30	10.9	11.6	14.4	10.0	7.75	26.4	9.70	3.19	3.47
44	10,134	0.18	0.50	2.69	3.73	5.41	4.38	2.71	26.9	0.91	1.74	0.73
48	10,824	0.43	1.62	5.22	8.11	8.52	7.19	5.97	26.8	2.18	5.94	1.69
52	10,878	0.39	0.87	3.60	6.26	6.25	4.46	3.15	26.6	0.88	3.74	0.31
56	10,932	0.38	1.11	5.10	9.92	9.08	6.60	4.67	26.6	0.70	6.15	0.24
68	11,266	1.27	0.96	4.31	6.79	7.01	4.98	2.53	26.4	0.65	4.23	0.20
72	11,597	2.22	1.83	6.67	10.2	11.5	7.53	3.65	26.3	0.84	6.42	0.37
80	12,258	2.89	0.71	2.86	3.97	4.38	4.04	2.37	26.7	0.61	2.47	0.11

Figure 2.

Downcore profiles of lipid biomarker and other parameters from Lake Qalluuraq on the Alaska North Slope: (a) reconstructed MAT based on brGDGTs and atmospheric methane concentration according to Brook et al. (2000), (b) reconstructed MST based on brGDGTs and δD values of n-C₂₃ and n-C₂₇, (c) concentration and δ¹³C values of hop-17(21)-ene, (d) concentration of n-C₂₃ and n-C₂₇ andn-alkane-based ACL (C₂₁ to C₃₁), (e) amount of TOC and concentration of sitosterol, and (f) δ¹³C values of TOC and sitosterol. TOC and its δ¹³C values are from Wooller et al. (2012). Red bar indicates the HTM for Central-eastern Beringia (~11,300 to ~9100 cal yr BP) based on proxy data compiled by Kaufman et al. (2004).

Hydrocarbon stable carbon isotopes

Hydrocarbon δ¹³C values are shown in Table 3. δ¹³C values of odd carbon numbered n-alkanes range from ~−27‰ to ~−32‰. Exceptions are higher values around −26.4‰ found for n-C₂₇ and n-C₂₉ in the 36-cm horizon at ~7600 cal yr BP and lower values between −34.7‰ and −36.2‰ found for n-C₂₁ in samples younger than ~4400 cal yr BP. Taraxer-14-ene and friedel-3-ene are consistent in δ¹³C values ranging from −26.9‰ to −29.9‰. δ¹³C values of the C_25:2 HBI could only be determined in the three oldest samples and are ~−23‰. Hop-17(21)-ene is the most ¹³C-depleted hydrocarbon with δ¹³C values < −42‰. Lowest values, down to −55.9‰, are found in samples older than ~10,000 cal yr BP, corresponding to peat layer I (Figure 2c).

Table 3.

δ¹³C values in ‰ of hydrocarbon biomarkers in Lake Qalluuraq sediments. Dashed line separates samples of the emergent fen (before ~10,000 cal yr BP) from the thermokarst lake period.

Sample depth (cm)	Age (cal yr BP)	C_25:2	C₂₁	C₂₃	C₂₅	C₂₇	C₂₉	C₃₁	Taraxer-14-ene	Hop-17(21)-ene	Friedel-3-ene
1	64	nd	−35.3	−30.4	−29.7	−29.2	−29.0	−30.0	−29.7	−48.4	−27.8
4	433	nd	−36.2	−30.2	−29.1	−28.3	−27.8	−28.9	−29.9	−51.5	−27.7
8	926	nd	−35.6	−29.5	−28.9	−27.9	−27.9	−28.8	−28.9	−48.6	−27.7
12	1418	nd	−36.0	−29.8	−29.4	−28.4	−28.3	−28.9	−28.7	−49.8	−27.5
16	1911	nd	−34.7	−31.0	−30.4	−29.4	−29.4	−30.2	−28.5	−50.8	−28.5
24	2896	nd	−36.0	−30.2	−29.6	−28.5	−28.3	−29.5	−29.0	−48.6	−27.4
28	4371	nd	−35.2	−29.4	−28.6	−27.8	−27.8	−28.8	−28.7	−45.5	−27.5
32	5987	nd	−32.2	−28.7	−29.3	−27.4	−27.5	−28.7	−27.6	−42.3	−27.4
36	7603	nd	−30.6	−28.0	−27.4	−26.3	−26.5	−29.0	−28.4	−48.0	−27.5
40	9218	nd	−30.9	−29.2	−29.3	−28.6	−28.5	−28.5	−28.8	−42.3	−27.4
44	10,134	nd	−27.3	−27.3	−28.2	−28.0	−27.9	−28.9	−28.4	−49.6	−27.2
48	10,824	nd	−29.5	−30.0	−31.4	−31.3	−30.8	−29.2	−28.4	−50.7	−28.2
52	10,878	nd	−28.0	−28.9	−30.9	−30.6	−29.7	−28.4	−26.9	−50.9	nd
56	10,932	nd	−26.8	−28.7	−30.8	−30.5	−29.7	−28.1	nd	−52.9	nd
68	11,266	−24.0	−29.8	−27.3	−29.8	−29.6	−28.4	−32.5	nd	−54.7	nd
72	11,597	−23.0	−30.6	−28.8	−31.6	−31.2	−31.9	−32.1	nd	−54.8	nd
80	12,258	−22.5	nd	−31.1	−32.2	−31.5	−31.9	−33.4	−27.2	−55.9	nd

nd: not determined.

Hydrocarbon stable hydrogen isotopes

δD values of odd n-alkane hydrocarbons are between −204‰ and −251‰ (Table 4), with both the highest and lowest values of n-C₂₁ (−204‰ to −219‰) and n-C₂₇/n-C₃₁ (−243‰ to −251‰), respectively, found in samples older than ~10,000 cal yr BP. δD values of taraxer-14-ene and friedel-3-ene were determined in samples younger than ~11,000 cal yr BP and are even more D-depleted, ranging from −253‰ to −288‰. The HBI C_25:2 showed δD values between −211‰ to −238‰, but were only determined in the three oldest samples. Hop-17(21)-ene is most D-enriched in the youngest sample (−230‰) and becomes progressively D-depleted with age down to around −263‰.

Table 4.

δD values in ‰ of hydrocarbon biomarkers in Lake Qalluuraq sediments. Dashed line separates samples of the emergent fen (before ~10,000 cal yr BP) from the thermokarst lake period.

Sample depth (cm)	Age (cal yr BP)	C_25:2	C₂₁	C₂₃	C₂₅	C₂₇	C₂₉	C₃₁	Taraxer-14-ene	Hop-17(21)-ene	Friedel-3-ene
1	64	nd	−239	−237	−233	−231	−229	−236	−265	−230	−269
4	433	nd	−240	−239	−235	−236	−234	−241	−264	−245	nd
8	926	nd	−242	−236	−228	−231	−233	−236	−267	−234	−275
12	1418	nd	−245	−237	−234	−231	−230	−235	−265	−231	−273
16	1911	nd	−243	−239	−234	−234	−234	−241	−276	−250	−274
24	2896	nd	−241	−237	−233	−231	−229	−238	−266	−244	−271
28	4371	nd	−242	−236	−236	−235	−238	−243	−281	−253	−281
32	5987	nd	−239	−231	−228	−229	−229	−240	−284	−254	−277
36	7603	nd	−234	−231	−224	−225	−228	−240	−275	−253	−269
40	9218	nd	−236	−240	−235	−236	−237	−245	−288	−263	−277
44	10,134	nd	−211	−212	−223	−236	−226	−238	−253	−253	−257
48	10,824	nd	−219	−234	−245	−249	−245	−247	−270	−260	−265
52	10,878	nd	−208	−222	−241	−249	−242	−249	nd	−251	nd
56	10,932	nd	−214	−224	−246	−247	−239	−247	nd	−256	nd
68	11,266	−222	−207	−227	−245	−250	−242	−245	nd	−255	nd
72	11,597	−211	−208	−226	−238	−248	−249	−251	nd	−256	nd
80	12,258	−238	−204	−214	−227	−243	−242	−245	nd	−259	nd

nd: not determined.

Sitosterol concentration and stable carbon isotopes

Sitosterol concentrations range from 2.38 to 76.3 µg g⁻¹ dw, peaking in peat layer II and are at a minimum in samples older than ~10,000 cal yr BP (Figure 2e). Following peat layer II, concentrations are at a minimum again at ~3000 cal yr BP but increase up to 58.8 µg g⁻¹ dw at the top of the core. δ¹³C values of sitosterol widely range from −23.6‰ to −30.8‰ (Figure 2f) and show a minimum at the beginning of the HTM that is offset from TOC by up to ~6‰ and become progressively more positive toward the top of peat layer I (−23.6‰ at ~10,000 cal yr BP). With younger age, sitosterol δ¹³C values show a continuous decrease toward the top of the core (−30.6‰), being consistently ¹³C-depleted relative to TOC by 2‰ to 3‰ until ~1400 cal yr BP. In most recent times, however, this difference increases up to ~4‰.

MAT and MST reconstructions

Our reconstruction of MAT variation is based on the analysis of brGDGTs in Lake Qalluuraq sediments and proxy determinations thereof (CBT/MBT; Weijers et al., 2007). CBT/MBT-derived MATs range from −3.2 to −8.8°C (Figure 2a), with most of the highest temperatures found in peat layer I that was partly deposited during the HTM (around −4.4°C, mean value of samples ranging from ~11,700 to ~10,000 cal yr BP) and the lowest occurring in samples from the terminal end of peat layer II deposited during the mid-Holocene (~3000 cal yr BP).

For MST reconstruction, we used fractional abundance of specific GDGTs according to Foster et al. (2016). Reconstructed MSTs range from 2.6 to 6.6°C (Figure 2b), with highest temperatures found between ~11,700 and ~11,000 cal yr BP and lowest occurring in samples between ~4000 and ~2400 cal yr BP. Generally, downcore MST shows a close resemblance to MAT, supporting the overall temperature evolution at Lake Qalluuraq.

Discussion

Our biomarker records from Lake Qalluuraq demonstrate strong links between lake evolution, climate, and methane dynamics on the Arctic Coastal Plain of Alaska during the Holocene (Figure 2). Documented lake expansion starting at ~10,000 cal yr BP (Wooller et al., 2012) triggered enhanced moss abundance and higher-plant-specific biomarkers from the surrounding watershed. After the onset of neoglaciation, intensified organic matter deposition occurred and is associated with relatively cold temperatures between ~7000 and ~3000 cal yr BP when effective moisture was decreased (Clegg and Hu, 2010), distinct pulses of glacial advances from the central Brooks range appeared (Ellis and Calkin, 1984), and lake level declined (Wooller et al., 2012). In contrast, our record implies a different situation before ~10,000 cal yr BP, a time period largely overlapping with the HTM that has been characterized by higher summer insolation (Berger and Loutre, 1991), increased moisture (Mann et al., 2002), warmer temperatures than today (Kaufman et al., 2004; Mann et al., 2010), and release of ancient permafrost carbon and sediments through enhanced thermokarsting (Gaglioti et al., 2014; Mann et al., 2010). At that time, the system evolved into an emergent fen with intense methane turnover based on higher abundance and stronger ¹³C-depletion of hop-17(21)-ene – a hopanoid lipid biomarker for bacteria including aerobic methanotrophs (e.g. Birgel and Peckmann, 2008) – while evidence for submerged mosses and terrestrial input markers was much reduced.

Thermal and moisture stimulation of methane oxidation during the early Holocene

Our MAT and MST estimates based on calibrations of brGDGT membrane lipids (Foster et al., 2016; Weijers et al., 2007) for the ~12,300–10,000 cal yr BP age ‘peat layer I’ are ~3°C warmer than the rest of the Holocene (Figure 2a and b) and closely coincide with the timing of the HTM in northern Alaska (Kaufman et al., 2004). This estimate is consistent with 2–3°C warmer than present summer temperatures occurring during this time period as indicated by the expansion of Populus sp. trees on the North Slope of Alaska between 11,500 and 9500 cal yr BP (Mann et al., 2010) and fossil midge stratigraphies at Trout Lake, northern Yukon Territory, between 10,800 and 9800 cal yr BP (Irvine et al., 2012). Additional evidence of elevated temperatures in the Arctic region during this time period is based on alkenone sea surface temperature reconstructions from sediments in the Gulf of Alaska where a similar difference of 2–3°C warmer temperatures, peaking at ~11,000 cal yr BP, than the rest of the Holocene has been identified (Praetorius et al., 2015).

To constrain the regional hydroclimatic conditions, we measured the δD values of n-alkanes from the Lake Qalluuraq sediment core (Table 4). During the formation of peat layer I, δD values of sedge-derived n-C₂₇, n-C₂₉, and n-C₃₁ are up to 20‰ more negative relative to more recent times very likely resulting from high effective moisture during the early Holocene (Figure 2b). Increased effective moisture before ~10,000 cal yr BP has also been documented by other studies (Mann et al., 2002; Wooller et al., 2012) and is corroborated by more negative δ¹³C values of the aforementioned n-alkanes indicative of stronger ¹³C discrimination during photosynthetic CO₂ fixation when water-use efficiency is higher (Table 3; Lloyd and Farquhar, 1994; Stewart et al., 1995).

During the early Holocene, hop-17(21)-ene was most abundant and ¹³C-depleted as low as −56‰ (Figure 2c), suggesting a greater carbon contribution from methane, both for the biomarker specifically and the lake carbon cycle in general. The high concentration and ¹³C-depletion of hop-17(21)-ene during this time interval coincide with a record of elevated atmospheric methane concentration (Figure 2a, Brook et al., 2000). This increase in atmospheric methane has been associated with the rapid expansion of high-latitude fens and wetlands (Yu et al., 2013). In such habitats, aerobic methanotrophic bacteria consume a fraction of the methane but the remainder passes to the atmosphere. If much of the region was composed of shallow wetlands and fens during this initial Holocene warming episode, then the landscape-climate conditions would have acted in synergy to stimulate methane emissions from the region.

Methane oxidation in aerobic parts of the fen environment would also produce ¹³C-depleted lake water CO₂. Evidence that this occurred is recorded in the sediment record by sitosterol that is ¹³C-depleted relative to the TOC by up to ~6‰ (Figure 2f). Sitosterol is a widespread lipid biomarker that occurs in higher plants (Volkman, 1986), including submerged mosses (Liebner et al., 2011) as well as aquatic algae such as freshwater eustigmatophytes that obtain their carbon from dissolved CO₂ (Volkman, 2003; Volkman et al., 1999). Signals of submerged mosses are lower during the early Holocene as evidenced by higher ACL values (Figure 2d); however, sitosterol production by eustigmatophytes may have been stimulated by melting permafrost, higher precipitation, and enhanced nutrient runoff. This was likely accompanied by thermokarsting (Gaglioti et al., 2014), longer ice-free seasons (Arp et al., 2011), and warmer lake-bed temperatures as is the case in comparable shallow ponds today (Arp et al., 2012). The intensification of lake primary productivity is evidenced by the presence of a C_25:2 HBI, indicative of marine and freshwater diatom species (Belt et al., 2001; Volkman et al., 1994), present in higher abundance in sediments of early stage fen development between ~12,300 and ~11,000 cal yr BP (Figure 1b, Table 2). Lower ratios of TOC to total nitrogen (C:N) during this time period (12.3 ± 1.7) compared with the rest of the record (14.3 ± 0.7; Wooller et al., 2012) are also supportive of the presence of aquatic algae during this time period.

Thermokarst lake environment and methane oxidation potential during the mid- and late-Holocene

After ~10,000 cal yr BP, an increase in n-C₂₃ alkane abundance and more negative δD values relative to n-C₂₇ suggests a greater contribution from submerged mosses (Ficken et al., 2000; Mügler et al., 2008). A greater sedimentary contribution from submerged mosses is supported by lower ACL values following the termination of the HTM (Figure 2d). This environmental change is corroborated by a D-depletion for n-C₂₃ that was probably associated with a deepening of the lake basin because of thermokarsting that, in turn, created floating ice conditions, shorter ice-free seasons, and less evaporation from the lake (Arp et al., 2011). A general increase in δD-values of up to ~20‰ of sedge-derived n-C₂₇,n-C₂₉, and n-C₃₁ after ~10,000 cal yr BP, on the other hand, indicates a period of decreased effective moisture and/or greater evapotranspiration by land plants than during times before (Table 4). The period of increase in δD values is followed by an overall constant δD profile of n-C₂₇ of around −230‰ for the past 8000 years (Figure 2b) that is consistent with production of n-C₂₇ under a cool and dry climate (Sachse et al., 2006; Shanahan et al., 2013b). An extensive cold period is specifically in agreement with our coldest reconstructed MAT and MST estimates of −8.8°C and 2.6°C, respectively, between ~4000 and ~2400 cal yr BP (Figure 2a and b) and glacial advances from the central Brooks Range around 4400, 3500, and 2900 cal yr BP (Ellis and Calkin, 1984). Moreover, negative chlorophyll-inferred reconstructed summer temperature anomalies between 5700 and 3000 cal yr BP derived from Kurupa Lake at the northern front of the Brooks Range (Boldt et al., 2015) and positive deviation in δ¹⁸O values of Chara-stem encrustations at Takahula Lake in the central Brooks Range between 5000 and 3500 cal yr BP (Clegg and Hu, 2010) corroborate this period being cooler, drier, and thus decreased in effective moisture.

A cooler, drier climate at Lake Qalluuraq between ~7000 and ~3000 cal yr BP may have been accompanied by longer ice-cover seasons and a lower lake level (Wooller et al., 2012), thus attenuating decomposition of organic matter and leading to its accumulation in peat layer II (Figure 2e). Lake deposits rich in organic carbon have been reported elsewhere (Jones et al., 2011), indicating that thermokarst lakes shifted from greenhouse gas effective sources toward carbon sinks at ~5000 cal yr BP (Walter Anthony et al., 2014). The contrast in organic matter composition between peat layers I and II has been postulated previously by Wooller et al. (2012) based on higher carbon loads of detrital moss material in peat layer II. Support is provided here by very low ACL values at ~5000 cal yr BP and corroborated by increased amounts of Sphagnum moss–derived n-C₂₃ (Ficken et al., 2000; Mügler et al., 2008) and flowering (i.e. Ericaceae) higher-plant-derived taraxer-14-ene and friedel-3-ene (Table 2; Ageta and Arai, 1983; Pancost et al., 2002). Peat layer II also shows a distinct increase in sitosterol abundance (Figure 2e), which is dominantly derived from a high portion of moss-derived material as evidenced by decreased ACL values (Figure 2d). Since indications of methanotrophy based on hop-17(21)-ene abundance and its ¹³C-depletion seem less pronounced in peat layer II than in peat layer I, we suggest the time period between ~7000 and ~3000 cal yr BP to be characterized by slowed organic matter degradation and reduced methane production.

The transition from peat layer II to gyttja at Lake Qalluuraq in the late-Holocene time period is accompanied by a shift to 2°C warmer reconstructed MAT and 1°C warmer reconstructed MST during the last two millennia, continuously increased accumulation of n-alkanes, TOC, sitosterol, and intensification of methane turnover relative to the mid-Holocene, with the latter evidenced by both higher abundance and stronger ¹³C-depletion of hop-17(21)-ene (Figure 2c). This finding points toward increased aquatic primary production stimulated by nutrient input and terrestrial runoff from the surrounding thermokarst lake environment, which was probably caused by refreezing and heaving of the lake-bed after drainage or secondary thermokarsting (Jorgenson and Shur, 2007), overall leading to enhanced carbon deposition in the lake sediment. Anoxic conditions were probably widespread fostering intense methanogenesis, which, in turn, stimulated enhanced methanotrophic activity.

In comparison to the fen environment that developed at the Lake Qalluuraq site during the early Holocene and near the HTM, a much higher abundance of submerged mosses is found in the later Holocene (Wooller et al., 2012), which is documented in ¹³C-depleted moss markers n-C₂₁ and n-C₂₃ relative to sedge-derived ¹³C-enriched n-C₂₇ (Table 3). Submerged mosses often exist in symbiosis with aerobic methanotrophic bacteria that provide methane-derived carbon via carbon dioxide for photosynthetic plant growth (Kip et al., 2010; Raghoebarsing et al., 2005), supporting the methane oxidation capacity of the lake system. Biomarkers of this symbiosis include hop-17(21)-ene for methanotrophs (Van Winden et al., 2010) and phytosterols such as sitosterol for phototrophs (Liebner et al., 2011; Raghoebarsing et al., 2005). This routing of methane-derived carbon into biomass has also been found for aquatic invertebrate fossils, such as chironomids and cladocerans (Wooller et al., 2012), with variances in δ¹³C-values reflecting methane carbon assimilation changes. A stronger divergence of sitosterol δ¹³C values relative to TOC by up to ~4‰ during the last 1000 years (Figure 2f) would thus be consistent with enhanced methane carbon turnover and fixation of ¹³C-depleted dissolved inorganic carbon.

Carbon deposition and methane emission potential of a thermokarst lake

To assess the impact of feedbacks between climate change and carbon cycling at the lake level, we investigated relationships between (a) ACL and δ¹³C-TOC as a general measure of moss abundance and its capacity to sequester methanotrophy-derived and other forms of organic carbon and (b) the specific ratio of dominantly moss-derived n-C₂₃ and dominantly sedge-derivedn-C₂₇ versus reconstructed MAT, highlighting the dependence of submerged moss abundance relative to changes in the temperature record (Figure 3).

Figure 3.

Cross plots of (a) ACL with δ¹³C-TOC and (b) then-C₂₃/n-C₂₇ ratio with MAT in Lake Qalluuraq sediments, highlighting the sequestration of methane-derived carbon via moss biomass and its relation to temperature, respectively.

Lower ACL values, indicative of a greater contribution from submerged mosses, correlate with more negative δ¹³C-TOC values (R² = 0.88, Figure 3a), thus illustrating deposition of a higher portion of ¹³C-depleted organic carbon during periods of higher moss abundance. Considering the continuous trend of more negative δ¹³C-TOC and δ¹³C-sitosterol values after ~10,000 cal yr BP (Figure 2f), this relationship indicates intensification of the aquatic carbon cycle in Lake Qalluuraq over time. A significant negative correlation between n-C₂₃/n-C₂₇ ratios and MAT (R² = 0.81, Figure 3b) suggests submerged moss was more prevalent during colder periods. This correlation indicates that temperature plays an important role on submerged moss deposition in thermokarst lakes as has been similarly reported for south Alaskan peatland (Nichols et al., 2014). Hence, warmer and wetter conditions observed today could have the opposite effect of reducing carbon burial and storage in such environments.

Based on these correlations, the early Holocene and its associated sharp rise in atmospheric methane concentrations could be interpreted as a harbinger of the climate of the future, specifically considering that warming has been shown to occur recently in the Arctic following a long-term cooling trend (Kaufman et al., 2009; Marcott et al., 2013; Miller et al., 2010; PAGES 2k Consortium, 2013). With respect to the trajectory of the Anthropocene and its great acceleration in greenhouse gas emissions (Steffen et al., 2015), the warming of wetlands and thermokarst lakes such as Lake Qalluuraq could have a strong impact on the global methane production–consumption balance. Warmer temperatures will cause enhanced primary productivity followed by organic matter sedimentation and turnover, leading to a high methanogenic potential in the lake sediments that is not compensated by the accumulation of moss-derived and other methane-impacted organic matter. From a temperature standpoint, northern lakes and wetlands could thus turn into strong greenhouse gas emitters in response to climate warming as has been identified during the early Holocene (Yu et al., 2013).

Conclusion

Our study gives insight into regulative controls of methane dynamics and carbon deposition from a high-latitude lake environment and highlights critical time periods such as the HTM occurring during the early Holocene in Arctic Alaska. This period was characterized by higher atmospheric methane concentrations, higher summer insolation, increased moisture and warmer temperatures than today and is a climate condition supported by our biomarker-based temperature and vegetation reconstruction. Increased primary productivity in shallow fens at that time created an intense cycle of methane production and consumption, with the latter leading to biomarker evidence of methane oxidation in the lake sediment record. During the Holocene, a stable but colder climate evolved and thermokarsting started at Lake Qalluuraq, which is in line with our n-alkane and triterpenoid biomarker records of submerged moss abundance and flowering higher plants from the surrounding tundra. Organic matter rich submerged peat moss accumulated during the coldest period of the mid-Holocene as a result of attenuated organic matter decomposition associated with lower lake level. A lack of evidence for strong methane assimilation into biomass during the mid-Holocene implies that thermokarst lakes were greenhouse gas effective carbon sinks at that time. However, in the late-Holocene as temperatures increased, intensification of the carbon and methane cycles indicates that thermokarst lakes most likely evolve to become net carbon sources. Although today’s lake environment is in a different stage of the wetland-thaw cycle as during the early Holocene and the HTM, this study points to more rapid organic matter decomposition and enhanced methane emission potentials with possibly amplified climate feedbacks in the future.

Footnotes

Acknowledgements

The authors thank the community in Atqasuk for kindly hosting us. Notably, they would like to thank the Mayor of Atqasuk and members of the city council. Susanne Alfken and Helena Dannert performed sample extractions for MAT reconstructions and deuterium isotope measurements of n-alkanes, respectively. Finally, the authors are thankful for comments and suggestions by William Longo and two anonymous reviewers.

Funding

Funding for ME, KWB, and K-UH was provided by the ‘Deutsche Forschungsgemeinschaft’ through the DFG-Research Center/Excellence Cluster MARUM ‘The Ocean in the Earth System’. This work was also supported by funding from the US Department of Energy National Energy Technology Laboratory (Grant DE-NT000565) and from an award from the US National Science Foundation awarded to MJW (Grant ARC-0909523). Additional support for this research was provided by Interagency Agreement DE-FE0002911 between the USGS Gas Hydrates Project and the US Department of Energy Methane Hydrates R&D Program. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government.

References

Ageta

Arai

(1983) Fern constituents: Pentacyclic triterpenoids isolated from Polypodium Miponicum and P. Formosanum. Phytochemistry 22: 1801–1808.

Arp

Jones

. et al. (2012) Shifting balance of thermokarst lake ice regimes across the Arctic Coastal Plain of northern Alaska. Geophysical Research Letters 39: L16503.

Arp

Jones

Urban

. et al. (2011) Hydrogeomorphic processes of thermokarst lakes with grounded-ice and floating-ice regimes on the Arctic coastal plain, Alaska. Hydrological Processes 25: 2422–2438.

Atahan

Grice

Dodson

(2007) Agriculture and environmental change at Qingpu, Yangtze delta region, China: A biomarker, stable isotope and palynological approach. The Holocene 17: 507–515.

Belt

Massé

Allard

. et al. (2001) Identification of a C₂₅ highly branched isoprenoid triene in the freshwater diatom Navicula sclesvicensis. Organic Geochemistry 32: 1169–1172.

Berger

Loutre

(1991) Insolation values for the climate of the last 10 million years. Quaternary Science Reviews 10: 297–317.

Birgel

Peckmann

(2008) Aerobic methanotrophy at ancient marine methane seeps: A synthesis. Organic Geochemistry 39: 1659–1667.

Boldt

Kaufman

McKay

. et al. (2015) Holocene summer temperature reconstruction from sedimentary chlorophyll content, with treatment of age uncertainties, Kurupa Lake, Arctic Alaska. The Holocene 25: 641–650.

Brook

Harder

Severinghaus

. et al. (2000) On the origin and timing of rapid changes in atmospheric methane during the last glacial period. Global Biogeochemical Cycles 14: 559–572.

10.

Brosius

Walter Anthony

Grosse

. et al. (2012) Using the deuterium isotope composition of permafrost meltwater to constrain thermokarst lake contributions to atmospheric CH₄ during the last deglaciation. Journal of Geophysical Research 117: G01022.

11.

Clegg

(2010) An oxygen-isotope record of Holocene climate change in the south-central Brooks Range, Alaska. Quaternary Science Reviews 29: 928–939.

12.

Eglinton

Hamilton

(1967) Leaf epicuticular waxes. Science 156: 1322–1334.

13.

Eller

Deines

Grey

. et al. (2005) Methane cycling in lake sediments and its influence on chironomid larval δ¹³C. FEMS Microbiology Ecology 54: 339–350.

14.

Ellis

Calkin

(1984) Chronology of Holocene glaciation, central Brooks Range, Alaska. Geological Society of America Bulletin 95: 897–912.

15.

Ficken

Swain

. et al. (2000) An n-alkane proxy for the sedimentary input of submerged/floating freshwater aquatic macrophytes. Organic Geochemistry 31: 745–749.

16.

Fischer

Behrens

Bock

. et al. (2008) Changing boreal methane sources and constant biomass burning during the last termination. Nature 452: 864–867.

17.

Flückiger

Monnin

Stauffer

. et al. (2002) High-resolution Holocene N₂O ice core record and its relationship with CH₄ and CO₂. Global Biogeochemical Cycles 16: 1010.

18.

Foster

Pearson

Juggins

. et al. (2016) Development of a regional glycerol dialkyl tetraether (GDGT)-temperature calibration for Antarctic and sub-Antarctic lakes. Earth and Planetary Science Letters 433: 370–379.

19.

French

(2007) The Periglacial Environment. Chichester: John Wiley & Sons.

20.

Gaglioti

Mann

Jones

. et al. (2014) Radiocarbon age-offsets in an arctic lake describe responses of permafrost carbon to climate changes. Journal of Geophysical Research: Biogeosciences 119: 1630–1651.

21.

Wooller

Pohlman

. et al. (2012) Diversity of active aerobic methanotrophs along depth profiles of arctic and subarctic lake water column and sediments. The ISME Journal 6: 1937–1948.

22.

Wooller

Pohlman

. et al. (2015) Methane-derived carbon flow through microbial communities in arctic lake sediments. Environmental Microbiology 17: 3233–3250.

23.

Hinkel

Eisner

Bockheim

. et al. (2003) Spatial extent, age, and carbon stocks in drained thaw lake basins on the Barrow Peninsula, Alaska. Arctic, Antarctic, and Alpine Research 35: 291–300.

24.

Hinrichs

Hmelo

Sylva

(2003) Molecular fossil record of elevated methane levels in late Pleistocene coastal waters. Science 299: 1214–1217.

25.

Irvine

Cwynar

Vermaire

. et al. (2012) Midge-inferred temperature reconstructions and vegetation change over the last ~15,000 years from Trout Lake, northern Yukon Territory, eastern Beringia. Journal of Paleolimnology 48: 133–146.

26.

Jones

(2010) Rapid deglacial and early Holocene expansion of peatlands in Alaska. Proceedings of the National Academy of Sciences of the United States of America 107: 7347–7352.

27.

Jones

Grosse

Jones

. et al. (2012) Peat accumulation in drained thermokarst lake basins in continuous, ice-rich permafrost, northern Seward Peninsula, Alaska. Journal of Geophysical Research 117: G00M07.

28.

Jones

Grey

(2011) Biogenic methane in freshwater food webs. Freshwater Biology 56: 213–229.

29.

Jones

Solovieva

Self

. et al. (2011) The influence of Holocene tree-line advance and retreat on an arctic lake ecosystem: a multi-proxy study from Kharinei Lake, North Eastern European Russia. Journal of Paleolimnology 46: 123–137.

30.

Jorgenson

Shur

(2007) Evolution of lakes and basins in northern Alaska and discussion of the thaw lake cycle. Journal of Geophysical Research 112: F02S17.

31.

Kaufman

Agerb

Anderson

. et al. (2004) Holocene thermal maximum in the western Arctic (0–180°W). Quaternary Science Reviews 23: 529–560.

32.

Kaufman

Schneider

McKay

. et al. (2009) Recent warming reverses long-term Arctic cooling. Science 325: 1236–1239.

33.

Kip

van Winden

Pan

. et al. (2010) Global prevalence of methane oxidation by symbiotic bacteria in peat-moss ecosystems. Nature Geoscience 3: 617–621.

34.

Kristen

Wilkes

Vieth

. et al. (2011) Biomarker and stable carbon isotope analyses of sedimentary organic matter from Lake Tswaing: Evidence for deglacial wetness and early Holocene drought from South Africa. Journal of Paleolimnology 44: 143–160.

35.

Lambers

Chapin

III Pons

(1998) Plant Physiological Ecology. New York: Springer.

36.

Le Mer

Roger

(2001) Production, oxidation, emission and consumption of methane by soils: A review. European Journal of Soil Biology 37: 25–50.

37.

Liebner

Zeyer

Wagner

. et al. (2011) Methane oxidation associated with submerged brown mosses reduces methane emissions from Siberian polygonal tundra. Journal of Ecology 99: 914–922.

38.

Lloyd

Farquhar

(1994) ¹³C discrimination during CO₂ assimilation by the terrestrial biosphere. Oecologia 99: 201–215.

39.

Loomis

Russell

Sinninghe Damsté

(2011) Distributions of branched GDGTs in soils and lake sediments from western Uganda: Implications for a lacustrine paleothermometer. Organic Geochemistry 42: 739–751.

40.

Loomis

Russell

Ladd

. et al. (2012) Calibration and application of the branched GDGT temperature proxy on East African lake sediments. Earth and Planetary Science Letters 357–358: 277–288.

41.

MacDonald

Beilman

Kremenetski

. et al. (2006) Rapid early development of circumarctic peatlands and atmospheric CH₄ and CO₂ variations. Science 314: 285–288.

42.

Mann

Groves

Reanier

. et al. (2010) Floodplains, permafrost, cottonwood trees, and peat: What happened the last time climate warmed suddenly in arctic Alaska? Quaternary Science Reviews 29: 3812–3830.

43.

Mann

Peteet

Reanier

. et al. (2002) Responses of an Arctic landscape to late glacial and early Holocene climatic changes: The importance of moisture. Quaternary Science Reviews 21: 997–1021.

44.

Marcott

Shakun

Clark

. et al. (2013) A reconstruction of regional and global temperature for the past 11,300 years. Science 339: 1198–1201.

45.

Miller

Brigham-Grette

Alley

. et al. (2010) Temperature and precipitation history of the Arctic. Quaternary Science Reviews 29: 1679–1715.

46.

Mügler

Sachse

Werner

. et al. (2008) Effect of lake evaporation on δD values of lacustrine n-alkanes: A comparison of Nam Co, Tibetan Plateau and Holzmaar, Germany. Organic Geochemistry 39: 711–729.

47.

Nichols

Peteet

Moy

. et al. (2014) Impacts of climate and vegetation change on carbon accumulation in a south-central Alaskan peatland assessed with novel organic geochemical techniques. The Holocene 24: 1146–1155.

48.

Nichols

Walcott

Bradley

. et al. (2009) Quantitative assessment of precipitation seasonality and summer surface wetness using ombrotrophic sediments from an Arctic Norwegian peatland. Quaternary Research 72: 443–451.

49.

PAGES 2k Consortium (2013) Continental-scale temperature variability during the past two millennia. Nature Geoscience 6: 339–346.

50.

Pancost

Baas

van Geel

. et al. (2002) Biomarkers as proxies for plant inputs to peats: An example from a sub-boreal ombrotrophic bog. Organic Geochemistry 33: 675–690.

51.

Pearson

Juggins

Talbot

. et al. (2011) A lacustrine GDGT-temperature calibration from the Scandinavian Arctic to Antarctic: Renewed potential for the application of GDGT-paleothermometry in lakes. Geochimica et Cosmochimica Acta 75: 6225–6238.

52.

Peterse

van der Meer

Schouten

. et al. (2012) Revised calibration of the MBT–CBT paleotemperature proxy based on branched tetraether membrane lipids in surface soils. Geochimica et Cosmochimica Acta 96: 215–229.

53.

Peterse

Vonk

Holmes

. et al. (2014) Branched glycerol dialkyl glycerol tetraethers in Arctic lake sediments: Sources and implications for paleothermometry at high latitudes. Journal of Geophysical Research: Biogeosciences 119: 1738–1754.

54.

Praetorius

Mix

Walczak

. et al. (2015) North Pacific deglacial hypoxic events linked to abrupt ocean warming. Nature 527: 362–366.

55.

Raghoebarsing

Smolders

AJP

Schmid

. et al. (2005) Methanotrophic symbionts provide carbon for photosynthesis in peat bogs. Nature 436: 1153–1156.

56.

Raynolds

Walker

Maier

(2006) Alaska Arctic Tundra Vegetation Map. Scale 1:4,000,000. Conservation of Arctic Flora and Fauna (CAFF) map no. 2. Anchorage, AK: US Fish and Wildlife Service.

57.

Sachse

Radke

Gleixner

(2006) δD values of individual n-alkanes from terrestrial plants along a climatic gradient – Implications for the sedimentary biomarker record. Organic Geochemistry 37: 469–483.

58.

Shanahan

Hughen

van Mooy

BAS

(2013a) Temperature sensitivity of branched and isoprenoid GDGTs in Arctic lakes. Organic Geochemistry 64: 119–128.

59.

Shanahan

Hughen

Ampel

. et al. (2013b) Environmental controls on the ²H/¹H values of terrestrial leaf waxes in the eastern Canadian Arctic. Geochimica et Cosmochimica Acta 119: 286–301.

60.

Sowers

(2010) Atmospheric methane isotope records covering the Holocene period. Quaternary Science Reviews 29: 213–221.

61.

Steffen

Broadgate

Deutsch

. et al. (2015) The trajectory of the Anthropocene: The Great Acceleration. The Anthropocene Review 2: 81–98.

62.

Stewart

Turnbull

Schmidt

. et al. (1995) ¹³C natural abundance in plant communities along a rainfall gradient – A biological integrator of water availability. Australian Journal of Plant Physiology 22: 51–55.

63.

Tierney

Russell

(2009) Distributions of branched GDGTs in a tropical lake system: Implications for lacustrine application of the MBT/CBT paleoproxy. Organic Geochemistry 40: 1032–1036.

64.

Tierney

Russell

Eggermont

. et al. (2010) Environmental controls on branched tetraether lipid distributions in tropical East African lake sediments. Geochimica et Cosmochimica Acta 74: 4902–4918.

65.

Van Hardenbroek

Heiri

Grey

. et al. (2010) Fossil chironomid delta ¹³C as a proxy for past methanogenic contribution to benthic food webs in lakes? Journal of Paleolimnology 43: 235–245.

66.

Van Winden

Kip

Reichart

G-J

. et al. (2010) Lipids of symbiotic methane-oxidizing bacteria in peat moss studied using stable carbon isotopic labelling. Organic Geochemistry 41: 1040–1044.

67.

Volkman

(1986) A review of sterol markers for marine and terrigenous organic matter. Organic Geochemistry 9: 83–99.

68.

Volkman

(2003) Sterols in microorganisms. Applied Microbiology and Biotechnology 60: 495–506.

69.

Volkman

Barrett

Blackburn

(1999) Eustigmatophyte microalgae are potential sources of C₂₉ sterols, C₂₂-C₂₈ n-alcohols and C₂₈-C₃₂ n-alkyl diols in freshwater environments. Organic Geochemistry 30: 307–318.

70.

Volkman

Barrett

Dunstan

(1994) C₂₅ and C₃₀ highly branched isoprenoid alkenes in laboratory cultures of two marine diatoms. Organic Geochemistry 21: 407–414.

71.

Walter Anthony

Chanton

Chapin

III . et al. (2008) Methane production and bubble emissions from arctic lakes: Isotopic implications for source pathways and ages. Journal of Geophysical Research 113: G00A08.

72.

Walter Anthony

Edwards

Grosse

. et al. (2007) Thermokarst lakes as a source of atmospheric CH₄ during the last deglaciation. Science 318: 633–636.

73.

Walter Anthony

Zimov

Chanton

. et al. (2006) Methane bubbling from Siberian thaw lakes as a positive feedback to climate warming. Nature 443: 71–75.

74.

Walter Anthony

Zimov

Grosse

. et al. (2014) A shift of thermokarst lakes from carbon sources to sinks during the Holocene epoch. Nature 511: 452–456.

75.

Weijers

JWH

Bernhardt

Peterse

. et al. (2011) Absence of seasonal patterns in MBT-CBT indices in mid-latitude soils. Geochimica et Cosmochimica Acta 75: 3179–3190.

76.

Weijers

JWH

Schouten

van den Donker

. et al. (2007) Environmental controls on bacterial tetraether membrane lipid distribution in soils. Geochimica et Cosmochimica Acta 71: 703–713.

77.

Whalen

(2005) Biogeochemistry of methane exchange between natural wetlands and the atmosphere. Environmental Engineering Science 22: 73–94.

78.

Wooller

Pohlman

Gaglioti

. et al. (2012) Reconstruction of past methane availability in an Arctic Alaska wetland indicates climate influenced methane release during the past ~12,000 years. Journal of Paleolimnology 48: 27–42.

79.

Wooller

Ruppel

Pohlman

. et al. (2009) Permafrost gas hydrates and climate change: Lake-based seep studies on the Alaska north slope. Fire in the Ice 4: 6–9.

80.

Loisel

Turetsky

. et al. (2013) Evidence for elevated emissions from high-latitude wetlands contributing to high atmospheric CH₄ concentration in the early Holocene. Global Biogeochemical Cycles 27: 131–140.

81.

Zhang

Osterkamp

Stamnes

(1996) Some characteristics of the climate in northern Alaska, USA. Arctic and Alpine Research 28: 509–518.

82.

Zheng

Singarayer

Cheng

. et al. (2014) Holocene variations in peatland methane cycling associated with the Asian summer monsoon system. Nature Communications 5: 4631.

83.

Zimov

Schuur

EAG

Chapin

III (2006) Permafrost and the global carbon budget. Science 312: 1612–1613.