Arctic-boreal lakes of interior Alaska dominated by contemporary carbon

Northern high-latitude lakes are critical sites for carbon processing and serve as potential conduits for the emission of permafrost-derived carbon and greenhouse gases. However, the fate and emission pathways of permafrost carbon in these systems remain uncertain. Here, we used the natural abundance of radiocarbon to identify and trace the predominant sources of methane, carbon dioxide, dissolved inorganic and organic carbon in nine lakes within the Yukon Flats National Wildlife Refuge in interior Alaska, a discontinuous permafrost region with high landscape heterogeneity and susceptibility to climate, permafrost, and hydrological changes. We find that although Yukon Flats lakes primarily process young carbon (modern to 1290 ± 60 years before present), permafrost-derived carbon is present in some of the sampled lakes and contributes, at most, 30 ± 10% of the dissolved carbon in lake surface waters. Apportionment of young carbon and legacy carbon (carbon with radiocarbon age ⩾5000 years before present) is decoupled among the dissolved inorganic and organic carbon species, with methane showing a stronger legacy signature. Our observations suggest that permafrost-thaw-related transport of carbon through Yukon Flats lacustrine ecosystems and into the atmosphere is small, and likely regulated by surficial sediments, permafrost distribution, wildfire occurrence, or masked by contemporary carbon processes. The heterogeneity of lakes across our study area and northern landscapes more broadly cautions against using any one region (e.g. Yedoma permafrost lakes) to upscale their contribution across the pan-Arctic.


Introduction
Northern circumpolar regions store about 60% of the world's soil organic carbon in permafrost (Hugelius et al 2014, Schuur et al 2015) and are extremely sensitive to climatic changes (Serreze and Francis 2006, Serreze and Barry 2011).Amplified warming in northern latitudes is causing permafrost thaw, mobilizing legacy carbon (defined here as Holoceneand/or Pleistocene-aged carbon with radiocarbon age ⩾5000 years before present (BP)) from long-term stores and exposing it to hydrological and microbial processes (Abbott et al 2015, Drake et al 2015, Vonk et al 2015, 2019, Tank et al 2020, Pedron et al 2022).The remineralization and release of previously sequestered soil organic carbon represents an important source of carbon dioxide (CO 2 ) and methane (CH 4 ) to the atmosphere and a net input of carbon into the contemporary global carbon cycle (Walter Anthony et al 2018, Turetsky et al 2020, Miner et al 2022).Yet, the magnitude of this feedback as mediated through northern high-latitude lakes remains highly uncertain due, in part, to the limited data and heterogeneity of lakes across northern landscapes (Wik et al 2016, Bogard et al 2019, Johnson et al 2022).
Several studies have used the natural abundance of radiocarbon ( 14 C) to estimate the contribution of permafrost carbon to lake carbon cycling (Zimov et al 1997, Nakagawa et al 2002, Walter et al 2006, Elder et al 2018, Gonzalez Moguel et al 2021).However, most studies focused on thermokarst systems overlying degrading Yedoma deposits, Pleistocene-aged permafrost soils with high ice and organic carbon content, in northern Alaska and Siberia (Matveev et al 2016, Walter Anthony et al 2018, Elder et al 2019, Dean et al 2020, Estop-Aragonés et al 2020, Kuhn et al 2023).Since Yedoma deposits constitute only about 13% (480 000 km 2 ; Strauss et al (2021)) of northern thermokarst landscapes (3600 000 km 2 ; Olefeldt et al (2016)), upscaling results from such studies likely overestimate permafrost carbon release from northern lakes.
Here, we focus on lakes within the Yukon Flats National Wildlife Refuge (NWR) in interior Alaska, U.S.A. Located in the boreal forest near the Arctic Circle approximately 160 km north of Fairbanks, the Yukon Flats NWR encompasses an area of 35 483 km 2 and occupies low-lying (30-303 m a.s.l) active and abandoned floodplains formed by fluvial, eolian, and thermokarst processes.It has a wide diversity of lake sizes and shapes, and is nested between the foothills of the Brooks Range to the north and the White Mountains to the south (Williams 1962; figure 1).Permafrost distribution maps indicate that the Yukon Flats NWR spans the discontinuous to continuous permafrost transitional boundary (Pastick et al 2013).Yedoma loess deposits are present in the foothills of the Yukon Flats NWR and alluvial sediments in the low-lying floodplains (Williams 1962, Strauss et al 2021).This area is undergoing hydrologic and environmental shifts that may be intensifying with permafrost degradation (Rover et al 2012, Walvoord et al 2012, Roach et al 2013).Moreover, these arid, topographically flat, low-elevation regions encompass ∼26% of the total northern circumpolar lake area but are sparsely represented in investigations of aquatic carbon cycling in Arctic-boreal lakes (Bogard et al 2019, Estop-Aragonés et al 2020, Kuhn et al 2021, Arsenault et al 2022).
Our study aims to identify the contribution of permafrost-derived carbon to the dissolved inorganic and organic carbon pools within lake systems in a discontinuous permafrost region in interior Alaska.We measured the radiocarbon and stable-isotope signatures of spatiotemporally integrated dissolved CH 4 and CO 2 , dissolved inorganic carbon (DIC), and dissolved organic carbon (DOC) to characterize carbon dynamics on whole-lake scales.This approach facilitates lake-to-lake comparisons and correlations with landscape-scale variables such as general surficial geology, permafrost distribution, and hydrology, all of which vary across the circumpolar north (Elder et al 2018(Elder et al , 2019)).

Study area and sample collection
Water and dissolved gas samples were collected from surface waters of nine lakes in the Yukon Flats NWR (figure 1).These lakes were selected to capture the range of geologic settings, hydrology, and permafrost coverage of the Yukon Flats NWR (figure 1, table 1).Lakes were categorized as upland and lowland (elevation >150 or <150 m a.s.l., respectively).Upland lakes are concentrated in the south near the foothills of the White Mountains, are generally deeper (4-16.1 m), and overlie Yedoma loess deposits.In contrast, lowland lakes cluster around the Yukon River, are typically shallower (0.9-4.5 m) and located within alluvial-fan sediments.Past, winter under-ice water sampling performed within the Yukon Flats indicate a seasonal floating ice regime and suggest that lakes within the Yukon Flats do not freeze to the lake bottom (O'Dwyer et al 2020).

Water sampling and laboratory analysis
Study lakes were accessed via float plane during the early fall season (29 August-2 September, 2021) to coincide with the potential release of soil carbon from deeper active soil layers (Koch et al 2022).Given the rapid equilibration rates within the inorganic carbon pool, it is generally assumed that DIC and dissolved CO 2 have similar 14 C signatures.However, historical data of lakes in the Yukon Flats NWR show that pH can range between 6.71 and 10.44 units (Halm andGuldager 2013, Halm andGriffith 2014), indicating that for some of the studied lakes the predominant DIC species are bicarbonate and carbonate, and dissolved CO 2 may be near negligible.To better capture the CO 2 signature associated with legacy carbon, we collected water samples for both DIC and dissolved CO 2 , in addition to DOC and dissolved CH 4 .Discrete water samples were collected from the upper 0.1 m where dissolved carbon gases exchange with the atmosphere and near the middle of each lake.Water temperature and pH were measured near the lake surface with a YSI EXO2 Sonde (YSI Incorporated, Yellow Springs, Ohio).Sensors were calibrated before and after sampling campaigns.Dissolved CH 4 and CO 2 concentrations were filtered in the field through a pre-rinsed 0.45 µm capsule filters from Geotech Environmental Equipment into in 30 ml serum vials.Analyses were conducted at the U.S. Geological Survey laboratories in Boulder, Colorado using a headspace equilibration method and a Li-Cor 6252 infrared CO 2 analyzer (LI-COR Biosciences, Lincoln, Nebraska).Dissolved CH 4 concentration was determined using a HP 5890 gas chromatograph with a flame ionization detector (Agilent Technologies, Santa Clara, California) to yield an uncertainty better than ±2.5%.Samples for isotopic analysis of oxygen ( 16 O-H 2 O) and hydrogen ( 2 H-H 2 O) in lake waters were collected in 10 ml borosilicate vials filled to capacity and analyzed at the University of Washington IsoLab laboratory using a Picarro Inc. L2130i liquid water cavity ring-down spectroscopy instrument (Picarro Inc. Santa Clara, California;Schauer et al 2016).
Radiocarbon ( 14 C) and stable isotope (δ 13 C) samples of dissolved CH 4 and CO 2 were collected from open waters using a novel gas extraction technique with a Liqui-Cel membrane contactor (3M Company, Charlotte, North Carolina) as described in Elder et al (2018) and Elder et al (2019).More detailed information can be found in the supplemental material.
Water samples for DIC concentration ([DIC]), radiocarbon ( 14 C-DIC), and stable isotopes of DIC (δ 13 C-DIC) were collected in acid-washed and combusted 250 ml borosilicate bottles, preserved with 50 µl of a 55 µM mercuric chloride (HgCl 2 ) solution, sealed, and stored in the dark at room temperature until isotope analysis at the Keck Carbon Cycle Laboratory UC Irvine.Analytical precision was better than ±2‰ in [DIC], and 14 C-DIC aliquots >0.4 mg C (Gao et al 2014, Pack et al 2015), and ±0.2‰ for δ 13 C-DIC based on replicate measurements of standards and samples.
DOC concentration and radiocarbon ( 14 C-DOC) water samples were filtered in the field through a pre-rinsed 0.45 µm capsule filters from Geotech Environmental Equipment, or in some cases precombusted (450 • C, >4 h) Whatman GF/F filters (0.7 µm) and collected into acid-rinsed highdensity polyethylene bottles.Filtered samples were stored in a dark, cool (4 • C) place during transport and then frozen (−20 • C) until analysis.DOC concentrations were measured at Florida State University in Tallahassee, Florida via high temperature catalytic oxidation on a Shimadzu TOC-L CHP (Shimadzu Scientific Instruments, Columbia, Maryland).Analytical precision using this method was better than 1% and standard deviation ± 0.1 based on replicate injections of standards and samples. 14C-DOC were analyzed at the National Ocean Science Accelerator Mass Spectrometer facility at the Woods Hole Oceanographic Institution.
Radiocarbon results are presented using fraction modern (FM) and years before present (years BP).Modern ages are defined here as carbon fixed from the atmosphere after 1950 and include 14 C-enrichment from nuclear weapons testing (FM > 1).

Source partitioning using isotope mixing models 2.3.1. Source apportionment of carbon in lake water
We use a radiocarbon mixing model similar to that described in Parnell et al (2010) to distinguish between the different inputs of dissolved carbon incorporated into the Yukon Flats NWR lakes and to quantitatively determine the fractional contribution of each source to the mixture: where (2) Here, 14 C i represents the natural radiocarbon content of different potential end members as well as the observed values ( 14 C obs ) at each lake, and f i represents the fraction that each endmember contributes to the total observed carbon pool.
Equations ( 1) and (2) are a system of two equations and four or more unknown variables so there is no unique solution.However, we can take advantage of the known lower and upper bounds (0 ⩽ f 1 , f 2 , f 3, f 4 ⩽ 1) and use the following numerical procedure to compute a solution: we select the values of two of the unknowns randomly from the interval [0, 1] and, using equations ( 1) and ( 2), calculate the values of the other two unknowns.If all calculated fractions fall into the interval [0, 1], the solution is considered valid; otherwise, the solution is discarded.This procedure was repeated until 10 000 valid solutions were obtained and the average and the standard deviation for all fractions are reported.We experimentally verified that 10 000 valid solutions are sufficient for the convergence of the reported values (see supplemental material and figure S1).

Source apportionment of dissolved CO 2 in lake water
Sources of lake CO 2 include the uptake of atmospheric CO 2 , the remineralization of recently-fixed allochthonous and autochthonous organic carbon, autotrophic respiration, and the microbial oxidation of CH 4 (McCallister and del Giorgio 2008, Solomon et al 2013, Bogard and del Giorgio 2016, Elder et al 2019).Here, we use a three-endmember isotope mixing model to approximate the range of possible contributions to the observed dissolved 14 C-CO 2 : where 14 C Atm .accounts for the contribution of recently fixed carbon and is assumed to have the same signature as atmospheric 14 C-CO 2 .f CH4 and f DOC represent the fraction of dissolved CO2 that originated from CH 4 oxidization and remineralization of DOC, respectively. 14C CH4 and 14 C DOC represent the measured radiocarbon contents of CH 4 and DOC.
While chemical weathering of carbonates has been identified in the Yukon River Basin (Striegl et al 2007(Striegl et al , 2012) ) and could signify a possible source of depleted 14 C-DIC to some of the Yukon Flat lakes ( 14 C = −1000‰ or 14 C dead), the enriched 14 C signature in CO 2 and DIC suggest that this input may be negligible and was not explicitly considered in the mixing model.

Results
We selected the Yukon Flats NWR in interior Alaska to assess the inputs of legacy carbon of mid-Holocene to Pleistocene origin to lake surface waters because of its landscape heterogeneity and susceptibility to changes in climate, permafrost, and hydrological conditions.Despite our study lakes spanning a relatively small area (10% of the Yukon Flats NWR), we captured a diverse range of lakes (figure 2(b)).Concentrations of dissolved CH 4 ranged from 0.11 to 0.73 µM, CO 2 from 27 to 314 µM, DIC from 706 to 20 338 µM, and DOC from 1043 to 9825 µM on our sampling dates.Sampled lakes ranged from neutral (pH = 7.6-7.9) to basic (pH > 8.1; table S1).
Radiocarbon values varied widely, from modern to 1290 ± 20 years BP for dissolved CH 4 , from modern to 130 ± 15 years BP for dissolved CO 2 (figure 2(a)), and from modern to 1120 ± 15 years BP for DOC (figure 3(b)).Generally, DIC was younger than dissolved CO 2 and relatively modern (modern to 55 ± 20 years BP; figure 3 and table S1).Thus, most carbon pools were dominated by modern carbon or carbon with a mean age younger than 5000 years BP, in line with previous work analyzing dissolved CO 2 and CH 4 (Elder et al 2018, Bogard et al 2019).However, two out of the nine sampled lakes (Boot and Greenpepper) displayed a clear presence of legacy carbon in either the dissolved CH 4 and/or DOC pools.
The 16 O-H 2 O and 2 H-H 2 O isotope ratios of lake water ranged from −15‰ to −10‰ and −142‰ to −109‰ (figure S5).Lakes were generally enriched in the heavy isotope when compared to the Global Meteoric Water Line, and the river and groundwater values of the Yukon Flats indicating weak lake connectivity to above-and below-ground hydrologic flows (Anderson et al 2013).

Discussion
While Arctic-Boreal lakes have generally been perceived as conduits for legacy carbon to the atmosphere, this study shows that lake systems are highly heterogeneous and dominated by the cycling of young carbon.Age decoupling between the dissolved carbon species further highlight the complexity of factor regulating lacustrine carbon cycling in the Yukon Flats NWR.From surficial sediment characteristics to landscape disturbances to sinks and sources of lake carbon.

Landscape characteristics and lake thermokarst potential
Local factors impacting soil thermal characteristics, such as surficial sedimentary deposits, hydrology, elevation, and wildfires, have a substantial influence on permafrost distribution and thermokarst potential (Smith 1975, Duguay et al 2005, Jorgenson and Shur 2007, Koch et al 2022).Near-surface permafrost probability maps for the Yukon Flats NWR reveal lower permafrost probability in lowland areas near the Yukon River (0%-80%) compared to uplands (31%-100%; figure 2(b)).This is consistent with surficial soil characteristics, which show that lowland lakes underlain by alluvial sediments are characterized by the absence of near-surface permafrost, while upland lakes underlain by Yedoma loess deposits are associated with higher distributions of near-surface permafrost (figures 1 and 2).Yedoma loess deposits in the upland have distinct thermal characteristics, soil moisture, and organic matter composition, which provide ideal conditions for substantial accumulation of ground ice to be present (Williams 1962).As a result, these deposits are more vulnerable to lake thermokarst processes (Edwards et al 2016, Anderson et al 2019).For example, although Greenpepper is recognized to have originated from thermokarst processes (Anderson et al 2019), lake formation in the Yukon Flats NWR uplands have been shown to be heterogeneous and not all lakes have thermokarst origins (Edwards et al 2016).

Drivers and sources of lake methane
Consistent with surficial soil characteristics and finding of other studies, dissolved CH 4 in lowland lakes was dominated by recently produced carbon, whereas mixtures of modern and legacy CH 4 are more apparent in upland lakes such as Boot and Greenpepper, which are found in areas with higher permafrost abundance (figures 1 and 2).As estimated by the radiocarbon mass balance model (equations ( 1) and ( 2)), dissolved CH 4 in Boot and Greenpepper may have up to 30 ± 10% legacy carbon from mid-Holocene to Pleistocene origin (figure 4(a)).These lakes also exhibited the highest fractions of carbon derived from active layer soils (∼585 years BP and up to 35 ± 24%; figure 4(a)).An exception to this pattern was West Crazy Lake, which despite being in the uplands, had modern 14 C-CH 4 .This may in part be explained by the steeper terrain surrounding West Crazy when compared to the other upland lakes and which may result in lower probability of nearsurface permafrost (figures 2 and S2; Johnson et al (2013), Pastick et al (2013)).Alternatively, this pattern may be reflective of the difference in lake water depth.West Crazy is relatively shallow (∼4 m deep) and may undergo different lakebed dynamics than Boot or Greenpepper (16.1 and 14.0 m, respectively).Values greater than 1 (gray shaded area) represent modern carbon fixed from the atmosphere after 1950.Note that due to uncertainties associated with small sample yields, some of the δ 13 C values were excluded from our analysis (see supplemental material for details).
Another explanation may be the difference in wildfire disturbances.Fire history shows a more recent fire occurrence surrounding Boot and Greenpepper (Big Creek fire in 2009) than in West Crazy (Glacier Creek fire in 2005; figure S3).Furthermore, investigation by Anderson et al (2019)

Drivers and sources of lake inorganic carbon
Unlike the 14 C-CH 4 patterns observed between upland and lowland lakes, dissolved 14 C-CO 2 and 14 C-DIC were predominantly modern across all lake types (f modern > 77 ± 17%).We estimate a minor contribution of legacy carbon to CO 2 and DIC, 6 ± 2% and 5 ± 2% respectively (figures 4(b) and (c)).
Despite Boot displaying the highest fraction of legacy carbon in the form of CH 4 (30 ± 10%), this signature was not reflected in the DIC pool.Dissolved CO 2 concentrations were two orders of magnitude higher than dissolved CH 4 (table 1), and modern.This suggests that processes other than the oxidation of aged CH 4 are regulating inorganic carbon at Boot, this is in contrast to what was reported for expanding Yedoma-type thermokarst lakes near Fairbanks (Elder et   16 O-H 2 O and −136‰ for 2 H-H 2 O) suggests that Boot is weakly connected to above-and below-ground hydrologic flows, omitting allochthonous carbon as a likely source (table S1).This is further confirmed by the measured δ 13 C-CO 2 and δ 13 C-DIC values (figure 3(a)), which closely match atmospheric δ 13 C-CO 2 (−8.8‰August 2020 average at Point Barrow or Nuvuk).These findings are supported by metabolic measurements conducted in the Yukon Flats NWR, which revealed that allochthonous organic carbon plays a relatively minor role in carbon cycling of shallow lakes in interior Alaska (Bogard et al 2019, Johnston et al 2020).Instead, inorganic carbon at Boot may be driven primarily by internally produced carbon fixed from atmospheric CO 2 .

Combining constituents to understand lake carbon sources and dynamics
When considering lake carbon as a whole, Greenpepper was the only lake out of all the investigated lakes that consistently displayed legacy carbon across all carbon species (figure 4).It also exhibited the largest fraction of legacy-derived DOC (19 ± 6%) and DIC (5 ± 2), and second largest fractions of aged CH 4 (25 ± 9%) and CO 2 (6 ± 2%).Altogether, this suggests that permafrost-derived carbon older than 5000 years BP is being processed and may be playing an important role in carbon cycling and greenhouse gas dynamics at Greenpepper.
Because the observed dissolved CO 2 at Greenpepper may represent a mixture of aged-carbon derived from CH 4 oxidation and the DOC remineralization, we use a three-endmember isotope mixing model (equations (3) and ( 4)) to approximate the fraction of dissolved CO 2 that originated from recently fixed carbon and oxidized CH 4 and DOC.Results from this isotope mixing model indicate that CO 2 in surface water at Greenpepper is a mixture of 65 ± 5% recently fixed carbon, 19 ± 11% CH 4derived carbon, and 17 ± 10% DOC-derived CO 2 .This first-order estimate suggests that even in relatively deep lakes such as Greenpepper (14 m depth) oxidation of aged dissolved CH 4 may contribute to aged dissolved CO 2 , similar to observation in lakes near Fairbanks (Elder et al 2019).
The variability in carbon sources of dissolved CH 4 and CO 2 across lakes reflects the heterogeneity of these landscapes and the variability in controls on greenhouse gas cycling.While some dissolved carbon characteristics are similar amongst lakes with shared landscape variables, the specific characteristics are not always consistent.For example, even though Boot and Greenpepper originated within the same sort of surficial geology, experienced similar fire disturbances, are relatively deep lakes, and exhibited similar dissolved CO 2 concentrations, dissolved CH 4 concentrations at Boot were twice as high as at Greenpepper's (table 1).The difference in lake CH 4 dynamics may be reflective of differences in methanogenic pathways and the microbial oxidation of CH 4 to CO 2 .Methanogenesis results in CH 4 that is depleted in 13 C-CH 4 , whereas oxidation results in enrich δ 13 C-CH 4 as bacteria and archaea preferentially consume 12 C-CH 4 (Whiticar 1999).The enriched δ 13 C-CH 4 signature at Greenpepper (−53 ± 0.1‰) is reflective of CH 4 oxidation, which helps to explain the age of dissolved CO 2 .The depleted δ 13 C-CH 4 signature at Boot (−61 ± 0.1‰) indicates a dominance of methanogenic processes and a weaker influence of oxidation on the dissolved CH 4 .This further suggest a greater control of autochthonous processes on the inorganic carbon pool (figure S4).While we recognize the importance of ebullitive CH 4 and CO 2 fluxes as a pathway of legacy carbon emissions, these were beyond the scope of this work and are not discussed here.Overall, results from our isotope mixing model, along with CH 4 concentration and its δ 13 C content, point to microbial oxidation of aged carbon as a key driver of CH 4 cycling at Greenpepper.

Conclusions
Our dataset represents one of the very few to date that simultaneously examined the incorporation of legacy carbon into both the inorganic and organic carbon pools within lake ecosystems within discontinuous permafrost regions (Gonzalez Moguel et al 2021).Our results indicate that aged carbon in lakes is only identifiable in patches and dictated by landscapescale properties such as permafrost distribution, landscape position, and wildfire activity within discontinuous permafrost in the Yukon Flats NWR.We hypothesize that as permafrost thaws and wildfires becomes more frequent, short pulse-like disturbances may mobilize legacy carbon and periodically represent a substantial proportion of active carbon pools across boreal-arctic lake ecosystems.It is also likely that most of that mobilized legacy carbon is remineralized within the changing soil environment, and rapidly respired before reaching lake surface waters (Drake et al 2015).These findings highlight the high heterogeneity of lakes systems and caution against using one region or lake type to extrapolate estimates of permafrost-derived carbon across the pan-Arctic.To obtain accurate upscaling estimates of the contribution of legacy carbon to greenhouse emissions, more prolonged and continuous efforts that consider the complexity of these systems are required to reliably characterize permafrost feedback on climate and carbon cycling in Arctic-boreal lakes.

Figure 1 .
Figure 1.Map of the surficial geology of the Yukon Flats National Wildlife Refuge (YFNWR) from Williams (1962) with lake names and locations for this study.Note that general descriptions of loess deposits in central Alaska indicate that West Crazy Lake (located outside the mapped extent of Williams 1962) overlies loess deposits (Muhs et al 2003; L. Anderson personal communication).© Williams (1962).U.S. Geoglogical Survey.

Figure 2 .
Figure 2. Lake and landscape properties in the Yukon Flats NWR.(a) Radiocarbon content of CH4 and CO2 dissolved in lake surface water expressed as fraction modern (FM) and 14 C years before present (years BP).Values greater than 1 (gray shaded area) represent modern carbon fixed from the atmosphere after 1950.Take note of the overlapping radiocarbon signatures for Shack and Thumb Lake.(b) Permafrost probability obtained from Pastick et al (2013).Pastick et al (2013).John Wiley & Sons.

Figure 3 .
Figure3.Radiocarbon content expressed as fraction modern (FM) and δ 13 C content of (a) dissolved CO2 (blue markers) and DIC (gray markers), and (b) DOC for the sampled lakes (symbols).Values greater than 1 (gray shaded area) represent modern carbon fixed from the atmosphere after 1950.Note that due to uncertainties associated with small sample yields, some of the δ 13 C values were excluded from our analysis (see supplemental material for details).
at Greenpepper revealed that the lake margins were undergoing permafrost thaw following the 2009 forest fire.This supports findings from prior investigations which indicate that fires in boreal landscapes may accelerate permafrost thaw and the mobilization of legacy carbon (Pastick et al 2013, Minsley et al 2016, Anderson et al 2019, Koch et al 2022) and represents a potential topic for future research.Our findings are consistent with previous studies which highlighted (1) the relationship between surficial geology and dissolved CH 4 sources in lake waters (Elder et al 2018, Gonzalez Moguel et al 2021), and (2) the connection between wildfire occurrence and permafrost thaw (O'Donnel et al 2011, Minsley et al 2016, Holloway et al 2020, Ludwig et al 2022).

Figure 4 .
Figure 4. Source apportionment of lake water for (a) dissolved CH4, (b) dissolved CO2, (c) DIC and (d) DOC ordered by lakes with highest (top) to lowest (bottom) atmospheric contributions.Where Atm.C is recently fixed carbon, active layer soil organic carbon (SOC) is 585 years BP permafrost carbon, Holocene carbon is 5000 years BP permafrost carbon, and Yedoma carbon is 35 800 years BP permafrost carbon.The boxes show the extent of feasible solutions from the first to the third quartile, with a white line at the median.The whiskers extend from the box to 1.5 times the inter-quartile range.Please note that CO2 source apportionment for Canvasback, Thumb, and YF18 were excluded due to inflated uncertainties for small sample yields.
al 2019).The signature of the oxidation of aged CH 4 may be overwhelmed by additional sources of contemporary carbon into Boot.Potential sources of young carbon to lake waters include inputs through groundwater transport (Kling et al 1992, Macpherson 2009, Paytan et al 2015), the direct inputs of autotrophic respiration, uptake of atmospheric CO 2 , and the remineralization of recently-fixed allochthonous and autochthonous organic carbon (McCallister and del Giorgio 2008, Solomon et al 2013, Bogard and del Giorgio 2016).Enrichment of lake water isotopes (−14‰ for

Table 1 .
Physical and biogeochemical characteristics of lakes within the Yukon Flats NWR.Discrete samples and measurements were taken from the upper 0.1 m of open waters at each lake.Alluvial indicates alluvial-fan and terrace deposits.