Permafrost peatlands store one-third of the total carbon (C) in the atmosphere and are increasingly vulnerable to thaw as high-latitude temperatures warm. Large uncertainties remain about C dynamics following permafrost thaw in boreal peatlands. We used a chronosequence approach to measure C stocks in forested permafrost plateaus (forest) and thawed permafrost bogs, ranging in thaw age from young (<10 years) to old (>100 years) from two interior Alaska chronosequences. Permafrost originally aggraded simultaneously with peat accumulation (syngenetic permafrost) at both sites. We found that upon thaw, C loss of the forest peat C is equivalent to ~30% of the initial forest C stock and is directly proportional to the prethaw C stocks. Our model results indicate that permafrost thaw turned these peatlands into net C sources to the atmosphere for a decade following thaw, after which post-thaw bog peat accumulation returned sites to net C sinks. It can take multiple centuries to millennia for a site to recover its prethaw C stocks; the amount of time needed for them to regain their prethaw C stocks is governed by the amount of C that accumulated prior to thaw. Consequently, these findings show that older peatlands will take longer to recover prethaw C stocks, whereas younger peatlands will exceed prethaw stocks in a matter of centuries. We conclude that the loss of sporadic and discontinuous permafrost by 2100 could result in a loss of up to 24 Pg of deep C from permafrost peatlands.

Introduction

Peatlands account for nearly 30% of the soil organic carbon (C) within the northern permafrost region, storing 277–302 Pg of organic C (Tarnocai et al., 2009; Hugelius et al., 2014), which is equivalent to over one-third of the C currently in the atmosphere (Houghton, 2007). The presence of permafrost and waterlogged soils in northern peatlands has facilitated the long-term accumulation of C by slowing rates of microbial decomposition. However, recent warming at high latitudes has accelerated permafrost thaw in northern peatlands (Payette et al., 2004; Camill, 2005), which can enhance decomposition and release C from formerly frozen peat deposits (e.g., O'Donnell et al., 2012a) and can increase emissions of methane (CH₄) as the peat surface becomes inundated (Turetsky et al., 2002; Wickland et al., 2006; Johnston et al., 2014; Klapstein et al., 2014).

Permafrost thaw may result in a positive feedback to atmospheric warming depending on whether C losses from thawing soils outweigh C gains in soils and vegetation (Koven et al., 2011; Schuur et al., 2015). Most studies predict net C loss and positive feedback to warming (Koven et al., 2011; Schuur et al., 2015), with large amounts of permafrost degradation by the end of the century. As much as 40% to 90% of continuous and discontinuous permafrost area may thaw by the end of the century (Lawrence and Slater, 2005). For peatlands, studies estimate that 20% of permafrost peat by volume may be lost, the equivalent of 33 Pg of C (Wisser et al., 2011). However, complex and heterogeneous geomorphology and landscape history of the boreal and arctic landscape increase the difficulty in making accurate large-scale projections of carbon cycle feedbacks to permafrost thaw (Jorgenson et al., 2001).

The presence of permafrost influences local hydrology and vegetation composition in peatlands (Camill, 1999; Jorgenson et al., 2013; Connon et al., 2015), which impacts the amount, chemical composition, and decomposability of C stored in these settings (Treat et al., 2014). In the boreal discontinuous permafrost zone of western North America, peatland permafrost plateaus contain a relatively well-drained active layer that can support trees such as black spruce (Picea mariana), ericaceous shrubs, feathermoss (Pleurozium schreberi, Hylocomium splendens), and lichen (Cladonia spp), and Sphagnum fuscum in wetter locations (Camill et al., 2009). Permafrost can form syngenetically with peat accumulation (Kanevskiy et al., 2014), but permafrost can also form epigenetically, or after peat or sediment has already been deposited (e.g., Zoltai, 1993). As C at the base of the active layer becomes incorporated into the permafrost, it suspends decomposition (Strauss et al., 2015), a process that stabilizes C, until permafrost thaw occurs. Permafrost thaw in ice-rich peat plateaus typically results in thermokarst development, characterized by subsidence of the ground surface, resulting in inundation that causes tree mortality and transforms the dominant vegetation to wetter species such as Sphagnum riparium and Carex aquatilis. These thawed bogs are called collapse-scar bogs, as the landscape physically collapses with the conversion of ice to water in these relatively ice-rich peat and underlying mineral deposits (Kanevskiy et al., 2014).

Several studies have recognized changes in C inputs and losses of thawing landscapes, but debate remains about whether C stocks increase or decrease following permafrost thaw (i.e., Robinson & Moore, 1999, 2000; Camill et al., 2001; O'Donnell et al., 2012a; Jones et al., 2013). Long-term peat C accumulation is impacted by net primary production (NPP) and decomposition. Studies have shown that NPP increases following permafrost thaw (Camill, 1999; Camill et al., 2001), but decomposition rates also change. Following thaw, inundation of the peat surface decreases aerobic decomposition that occurred in the active layer prior to thaw, and decomposition changes to much slower anaerobic decomposition, resulting in rapid accumulation of post-thaw peat (Camill, 1999; Robinson & Moore, 1999, 2000; Camill et al., 2001; Jones et al., 2013). However, a multisite comparison of peat accumulation rates among permafrost peat plateaus and collapse-scar bogs found no significant difference in long-term accumulation rates for the past millennium (Treat et al., 2016). Despite high accumulation rates, collapse-scar bogs have higher CH₄ emissions than permafrost plateaus (Turetsky et al., 2002; Christensen et al., 2004; Wickland et al., 2006; Johnston et al., 2014). Most studies have shown that CH₄ emissions (Klapstein et al., 2014) and pore water DOC (Chanton et al., 1995, 2008) in peatlands are the product of growing season productivity. Based on the dominance of growing season processes in C exchange and the switch to an anaerobic environment, it has been assumed that the loss of deep peat C is not a significant contributor to post-thaw C loss. However, a comparison of multiple cores from a collapse-scar bog chronosequence (i.e., space-for-time since permafrost thaw) in west-central Alaska suggests a significant portion of the formerly frozen plateau peat decomposes in the years to decades following thaw (O'Donnell et al., 2012a).

Because of these conflicting findings, considerable debate remains about the fate of former permafrost C following thaw and whether collapse-scar bogs turn from a C sink into a C source or whether they enhance their ability to store C (Camill 1999; O'Donnell et al., 2012a; Jones et al., 2013; Turetsky et al., 2000; Turetsky et al., 2007). Single-core analyses bias our understanding toward only the C that remains in the core, without providing a context for what may have been lost, while the chronosequence approach enables us to observe C stocks from initial to late stages of thaw but only within decades to centuries. O'Donnell et al. (2012a) employed the chronosequence approach and concluded that the timing of thaw has large impacts on the overall C storage capacity of a peat plateau complex. Similarly, Johnston et al. (2014) found differences in growing season CH₄ efflux increased over decades following thaw, but smaller C stocks were apparent in thaw features found to be one to two decades old. These two studies assume that the primary difference among the points along the chronosequence was difference in timing of permafrost thaw. However, in addition to timing of permafrost thaw, the boreal landscape also exhibits large spatial heterogeneity with respect to peat initiation. Consequently, both peatland initiation and thaw timing must be explored in order to more accurately understand C gains and losses with time.

In this study, we evaluate the assumptions of the chronosequence and mass balance modeling approach used by O'Donnell et al. (2012a), including that each site along the chronosequence would have formed at the same time in the same way, in order to address the C dynamics following permafrost thaw. We re-examine multiple cores from permafrost plateaus and collapse-scar bogs at the Koyukuk NWR sites of O'Donnell et al. (2012a) and sites within the Innoko NWR (Jorgenson et al., 2013; Johnston et al., 2014; Kanevskiy et al., 2014). The landscape development processes between the two sites are similar, but the Koyukuk NWR peats are nearly twice as old as the Innoko NWR peats, making them ideal candidates for studying the impact of peat initiation age on thaw C dynamics. We critically assess some of the assumptions made in O'Donnell et al. (2012a,b) and modify how C stocks within the permafrost plateau are calculated. We also build on the mass balance approach used by O'Donnell et al. (2012a) to track the changes in and controls on peatland C balance from decades to millennia. Our goal is to incorporate C input and decomposition parameters into a simple and generalizable mass balance model (Harden et al., 2000; O'Donnell et al., 2011, 2012a) based on a modified chronosequence approach to simulate C accumulation and loss from peat initiation through permafrost thaw during the Holocene.

Materials and methods

Regional setting

The Innoko National Wildlife Refuge (NWR) (63.57°N, 157.72°W), a part of the Yukon River basin, is located in western Alaska and consists of 12 540 km² of low-lying land surrounding the Innoko River (Fig. 1). Wetlands comprise more than half of the Innoko NWR (Jorgenson et al., 2013), and sediments are composed of Pleistocene and Holocene flood plain deposits, including alluvial, colluvial, glacial, and aeolian deposits (Patton et al., 2009). The Koyukuk NWR (65.19°N, 155.36°W), located to the northeast of Innoko NWR, spans from the southern foothills of the Brooks Range to the Yukon River and covers 14 160 km² adjacent to the Koyukuk River (Fig. 1). Most of this area experienced aeolian silt deposition during the late Pleistocene (Muhs et al., 2003; Muhs & Budahn, 2006). This ice-rich, silty soil, referred to as yedoma, contains massive ice wedges that can be 10 m wide and 50 m high (Kanevskiy et al., 2014).

Details are in the caption following the image — **Figure 1**
Open in figure viewer PowerPoint

Innoko study area shown on the map of Alaska (black star). Black dot is the Koyukuk study area (O'Donnell *et al*., 2012a). Coring locations are shown on the areal image of the Innoko study area. Dark black/blue areas represent open water, orange represents *Sphagnum-*dominated mats over open water, dark green and white areas are the permafrost plateaus with black spruce and lichen, and the bright green are recently formed collapse-scar bogs.

The locations of both the Innoko and Koyukuk NWR study sites comprise a heterogeneous landscape, consisting of permafrost plateaus interspersed with collapse-scar bogs and fens, that ultimately was born out of the thawing of ice wedges in Pleistocene frozen loess (yedoma). These ice-rich loess deposits were subject to rapid thaw during the warm early Holocene in Alaska, resulting in widespread thermokarst lake development (Kaufman et al., 2004; Walter et al., 2007; Kanevskiy et al., 2014; Anthony et al., 2014). Subsequent lake drainage, coupled with peat accumulation under a cooler mid-Holocene climate (Jones and Yu, 2010), resulted in syngenetic or quasi-syngenetic permafrost aggradation in drained lake basins (Kaufman et al., 2004; Jones et al., 2012; Kanevskiy et al., 2014) and the formation of permafrost plateaus (Kanevskiy et al., 2014). Discontinuous permafrost, ranging from 10 to 130 m thick (Kanevskiy et al., 2014), is found in this region today. Permafrost occurs mainly in uplifted permafrost plateaus, which are interspersed with collapse-scar bogs and fens that generally lack permafrost in the top 2–3 m of soil. The high ice content of permafrost plateaus, coupled with warming air temperatures in the region, has increased the vulnerability of these sites to develop thermokarst features, such as collapse-scar bogs and fens (O'Donnell et al., 2012a,b; Kanevskiy et al., 2014; Nossov et al., 2015).

The climate of the region is continental, with a mean annual air temperature (MAAT) at Galena of −3.8 °C from 1971 to present (Alaska Climate Research Center. http://climate.gi.alaska.edu). July is the warmest month (20.4 °C), January is the coldest (−26.9 °C), and annual precipitation averages 331 mm, with 40% falling from July to September. Precipitation typically falls as snow from October to April, and average snow thicknesses are typically 50 cm.

Vegetation is similar at Innoko and Koyukuk (Jorgenson et al., 2013). Vegetation on permafrost peat plateaus is dominated by black spruce (Picea mariana) with a Sphagnum fuscum-dominated understory (Jorgenson et al., 2013; Johnston et al., 2014). The understory of drier plateaus is dominated by feathermosses (Pleurozium schreberi and Hylocomium splendens) and lichen (Cladonia spp.). Other common understory species are Rhododendron groenlandicum, Vaccinium uligunosum, and Betula nana. Wetter collapse-scar bogs are dominated by Sphagnum riparium, S. jensenii, S. lindbergii, and Eriophorum scheuchzeri. Drier, later stage collapse-scar bogs are dominated by S. balticum, S. flexuosum, S. lenense, Andromeda polifolia, Oxycoccus microcarpus, and Rhododendron palustre.

We used a chronosequence approach to understand the trajectories of C storage following thaw by coring numerous locations from various stages of thaw at Innoko NWR (Fig. 1; see also Johnston et al., 2014). To do this, we sampled the centers of the permafrost plateau with a Sphagnum fuscum understory (mesic permafrost plateau) and along drying margins of the plateau that contained a feathermoss understory (drying margin [DM]). Based on information from prior studies (O'Donnell et al., 2012a; Johnston et al., 2014) about the age of the features, we sampled nearby collapse-scar bogs and categorized them based on physical features and vegetation composition (Jorgenson et al., 2013; Johnston et al., 2014), where in the field, small features (<10 m²) with earliest successional post-thaw vegetation (S. riparium, S. lindbergii) were assumed to be young (young bog), medium-sized features with early successional post-thaw vegetation (S. riparium, S. lindbergii, Eriophorum scheuchzerii) were assumed to have initially thawed in the last few decades (intermediate bog), and large features with late-successional vegetation were assumed to have thawed several centuries ago (old bog), assumptions that have previously been demonstrated in similar published chronosequence studies (O'Donnell et al., 2012a,b; Johnston et al., 2014). We confirmed the ages of these thaw features with radiometric dating and dendrochronology (O'Donnell et al., 2012a,b). At Innoko, samples collected in 2009 indicated that thaw occurred 14–21 years ago in young bogs, 30–60 years ago in intermediate bogs and 200–1400 years ago in old bogs (Johnston et al., 2014). At Koyukuk NWR (details provided in O'Donnell et al., 2012a,b), permafrost thaw ranged from 31 to 61 years ago in young bogs and 400–1200 years ago in old bogs (O'Donnell et al., 2012a). The basal peat ages of the features studied were early Holocene (~7500 to ~10 000 cal yr BP) in age (O'Donnell et al., 2012a), while younger basal ages were indicated at Innoko (~3500 cal yr BP; Kanevskiy et al., 2014).

Field and laboratory sampling

Field sampling of the Innoko chronosequence occurred in 2009 and 2012 in the Innoko Flats NWR near Horseshoe Lake. Permafrost cores from the peat plateaus were collected using the Snow, Ice, and Permafrost Establishment (SIPRE) permafrost corer (Rand & Mellor, 1985) and cores were collected from the collapse-scar bogs using a Russian peat auger. Cores were described and subdivided following the similar methodology to O'Donnell et al. (2012b), reflecting a modification of USDA-NRCS (Soil Survey Staff, 1998) and Canadian (Soil Classification Working Group, 1998) methods. The lithology and plant macrofossils were described to identify transitions from limnic sediments to frozen peat plateau and thaw (Table S1). Subsamples were oven-dried, ground, and analyzed for C and N using a Carlo Erba NA1500 elemental analyzer. C stocks were calculated for each core (Table 1). C stocks for Koyukuk (Table 2) are updated values for Koyukuk from O'Donnell et al. (2012a,b) and for Innoko 2009 samples from Johnston et al. (2014). Changes reflect differences in gap-filling methodologies for % C and bulk density (see Jorgenson et al., 2013; for more information).

Table 1. Summary of basal ages and transitions in Innoko cores. Dating was performed using both ¹⁴C and ²¹⁰Pb, using the constant rate of supply (CRS) method to attain ages. Ages are reported in calendar years before 1950 in order to be consistent with ¹⁴C age reporting

Laboratory ID	Core ID	Depth (cm)	Age type	CRS age (cal yr BP)	Uncalibrated age (yr BP)	Error	Calibrated 2–sigma age range (cal yr BP)	Best fit Median age from Bacon age model (cal yr BP)	Stratigraphic significance
WW8370	Old Bog 1	80	Radiocarbon		120	35	10–150, 185–273	411	Transition to thaw
WW7896	Old Bog 1	100	Radiocarbon		950	25	795–887	750	Basal peat above limnic sediments
WW9261	Old Bog 2	165	Radiocarbon		870	25	726–801	817	Transition to thaw
WW9262	Old Bog 2	176	Radiocarbon		1885	25	1767–1764	1273	Basal forest peat
WW7759	Old Bog 3	115	Radiocarbon		1470	20	1310–1392	1263	Transition to thaw
WW7760	Old Bog 3	119	Radiocarbon		1735	25	1568–1707	1525	Basal forest peat
WW9282	Old Bog 3	153	Radiocarbon		1495	25	1312–1414	1439	Transition to thaw
WW9309	Old Bog 3	164	Radiocarbon		1780	35	1609–1817	1615	Basal peat above limnic sediments
WW9309	Intermediate Bog 2	32	210-Pb, CRS	15.00				13.4	Transition to thaw
WW9280	Intermediate Bog 2	101	Radiocarbon		2105	30	1996–2149	2060	First transition to thaw; 87–101 is a water pocket
WW9259	Intermediate Bog 2	124	Radiocarbon		2675	25	2749–2799	2771	Basal peat above limnic sediments
WW8373	Intermediate Bog 3	60	210-Pb, CRS	−28.00				15.2	Transition to thaw
	Intermediate Bog 3	139	No date; age model derived					3926	Basal peat above limnic sediments
	Young Bog 1	19	210-Pb, CRS	−42.00		2.6			Transition to thaw
WW7761	Young Bog 1	65	Radiocarbon		225	20	126–166	148	Basal forest peat
WW7761	Young Bog 2	25	210-Pb, CRS	−44.40		0.9			Transition to thaw
WW7861	Young Bog 2	48	Radiocarbon	36.11	420	25
WW7762	Young Bog 2	53	Radiocarbon		1430	30	1293–1376
	Young Bog 2	58	No date; age model derived; high uncertainty					2387	Basal peat above limnic sediments
	Mesic PP 2	2	210-Pb, CRS	−57.86		0.4
	Mesic PP 2	4	210-Pb, CRS	−56.50		0.5
	Mesic PP 2	6	210-Pb, CRS	−53.47		0.7
	Mesic PP 2	8	210-Pb, CRS	−48.27		1.2
	Mesic PP 2	10	210-Pb, CRS	−42.29		1.6
	Mesic PP 2	15	210-Pb, CRS	−29.74		3.7
	Mesic PP 2	20	210-Pb, CRS	−25.87		3.9
	Mesic PP 2	25	210-Pb, CRS	−20.00		4.6
	Mesic PP 2	30	210-Pb, CRS	−0.34		9.3
	Mesic PP 2	33	210-Pb, CRS	19.00		18.5
	Mesic PP 2	38	210-Pb, CRS	40.54		39.4
WW7899	Mesic PP 2	60	Radiocarbon		600	30	577–653	539
WW7978	Mesic PP 2	90	Radiocarbon		1935	30	1820–1949	1685
WW8374	Permafrost Plateau DM 1	104	Radiocarbon		2160	35	1940–2315	2128	Basal peat above limnic sediments

PP, permafrost plateau; DM, drying margin.

Table 2. C stocks and apparent C accumulation rates for bog and permafrost plateau peats for each core at Innoko and Koyukuk. C accumulation rates could only be calculated for cores with basal and transitional ages

	cm	cm	cal yr BP	kg m⁻²	kg m⁻²	kg m⁻²	g C m⁻² yr⁻¹	g C m⁻² yr⁻¹
	Thickness of Bog peat	Thickness of forest peat	Basal age	C in Forest peat	C in bog peat	Total C in peat	Full-profile apparent C accumulation rates	Post-thaw apparent C accumulation rates
Innoko
Permafrost Plateau (mesic) 1	0	95	1840	64.4	–	64.4	35.0	n.a.
Permafrost Plateau (mesic) 2	0	105	n.d.	45.7	–	45.7	–	n.a.
Permafrost Plateau (mesic) 3	0	113	n.d.	51.2	–	51.2	–	n.a.
Permafrost Plateau DM 1	0	104+	2129	>90.2	–	>90.2	42.4	n.a.
Permafrost Plateau DM 2	0	105+	n.d.	>61.9	–	>61.0	–	n.a.
Permafrost Plateau DM 3	0	82	n.d.	54.2	–	54.2	–	n.a.
Young Bog 1	19	46	290	16.3	2.5	18.8	56.2	507.5
Young Bog 2	25	33	n.d.	12.5	1.6	14.1	–	146.1
Young Bog 3	16	48	n.d.	39.6	5.1	44.7	–	–
Intermediate Bog 1	21	87+	n.d.	>34.5	1.4	35.9	–	–
Intermediate Bog 2	32	107	486	14.1	2.8	16.9	34.8	37.1
Intermediate Bog 3	60	79	3643	29.8	5.5	35.3	9.7	138.1
Old Bog 1	80	20	675	7.8	18.9	26.7	28.0	82.6
Old Bog 2a	165	10	1826	9.7	53.7	63.4	29.4	70.3
Old Bog 3	115	39	1713	39.1	53.0	92.1	31.0	39.2
Koyukuk
Permafrost Plateau 1	–	175	7830	77.9	–	77.9	9.9	–
Permafrost Plateau 2	–	405	10435	203.5	–	203.5	19.4	–
Permafrost Plateau 3	–	248	n.d.	128.1	–	128.1	–	–
Young Bog 1	12	188	–	>66.5	0.6	>67.1	–	11.5
Young Bog 2	10	132	–	63.1	0.8	63.9	–	13.5
Young Bog 3	19	21	–	>8.8	3.2	>12	–	103.2
Young Bog 4	29	244	–	95.1	4	99.1	–	74.1
Young Bog 5	20	40	–	>10	1.1	>11.1	–	19.1
Young Bog 6	16	286	–	135	1.7	136.7	–	51.4
Old Bog 1	290	180	–	53.6	35.3	–	–	29.5
Old Bog 2	250	40	–	20.3	35.3	–	–	–
Old Bog 3	87	100	–	39.5	20.3	–	–	48.8

n.d., not determined; + = basal peat not reached; ≥ minimum values because basal peat not reached.

Radiometric dating

We used two radiogenic isotopes, lead-210 (²¹⁰Pb) and radiocarbon (¹⁴C), to estimate age of ecosystem transitions, following the results for 2009 cores reported in Johnston et al. (2014) and updated with new core data. Both dating methods were used to constrain the ages of transition from limnic sediments to peat (indicating a lake drainage event) and from permafrost plateau to collapse-scar bog (indicating permafrost thaw, collapse, and inundation). We used both plant macrofossils and peat type as evidence for these transitions (Table 2; Table S1). ²¹⁰Pb and ²²⁶Ra was measured by gamma spectrometry using a Princeton Gamma HPGe germanium well detector. Unsupported ²¹⁰Pb was defined as the difference between measured total ²¹⁰Pb and ²²⁶Ra. Horizon subsamples from each soil profile were measured until unsupported ²¹⁰Pb was not detectable. We used the constant rate of supply (CRS) method to calculate ages of basal horizon depths (Appleby & Oldfield, 1978). To account for compaction, we modeled unsupported ²¹⁰Pb as a function of cumulative dry mass (g cm⁻²) instead of depth. Uncertainties in ²¹⁰Pb from the CRS method are based on error analysis described in Van Metre & Fuller (2009). Although mobility of ²¹⁰Pb has been raised as a potential problem when dating surface organic soil layers (Urban et al., 1990), our data suggest that ²¹⁰Pb mobility is not an issue at our sites, based on the similar surficial ²¹⁰Pb and ¹⁴C-based ages found for the Innoko site by Johnston et al. (2014; Fig. S1).

For deeper peat stratigraphy, plant macrofossils were identified and picked for ¹⁴C dating. If macrofossils were not available, we submitted bulk peat picked free of roots. Samples were graphitized at the USGS Radiocarbon Laboratories in Reston, VA, and ¹⁴C measurements were determined at the Center for Accelerator Mass Spectrometry (CAMS), Lawrence Livermore National Lab in Livermore, CA. Radiocarbon ages were calibrated Calib 7.1 (Reimer et al., 2013), and age models were generated on selected cores using the Bayesian age-depth modeling program BACON (Blaauw & Christen, 2011), which uses IntCal13 ¹⁴C age calibration curves (Reimer et al., 2013). Age models were generated using all available ²¹⁰Pb and ¹⁴C ages (Table 1, Table S1) to constrain the timing of transition between our dated profiles. Johnston et al. (2014) compared ²¹⁰Pb and ¹⁴C age results from Innoko to a pair of sites at Koyukuk Flats, Alaska, and found close agreement of age methods in shallow peat samples, validating the use of both ²¹⁰Pb and ¹⁴C to generate age models of the sites (Fig. S1). The age models helped constrain the timing of transitions between our dated depth intervals when no age was obtained at a transition (Table 1).

Constraining rates of soil carbon accumulation and loss

To quantify C stocks in each of our cores along a chronosequence, we first calculated cumulative C stocks (kg m⁻²) within each stratum of a given core. These strata are as follows (Fig. 2): permafrost plateau peat (thawed and currently frozen forest peat from within the permafrost plateau forest), forest peat (formerly frozen permafrost plateau peat, now a part of the collapse-scar bog), and bog peat (collapse-scar bog peat that has accumulated over the forest peat). We also calculated apparent C accumulation rates (g C m⁻² yr⁻¹), or the product of C densities and the peat accretion rate (cm yr⁻¹) across sites. We assumed that the net change in C stocks (kg C m⁻²) over time is controlled by annual C inputs (I; kg C m⁻² yr⁻¹) and the fractional first-order decomposition constant (k; yr⁻¹) and C stocks in a given year (C(t)) (Clymo, 1984; Trumbore and Harden, 1997; O'Donnell et al., 2012a). The C balance for any given year can then be determined from the equation

$urn:x-wiley:13541013:media:gcb13403:gcb13403-math-0001$ (1)

Assuming the initial OC concentration is zero, solving Eqn 1 yields

$urn:x-wiley:13541013:media:gcb13403:gcb13403-math-0002$ (2)

We estimated the C accumulation rates in the permafrost plateaus by fitting Eqn 2 to the C stocks of the entire organic soil profile vs. the radiocarbon age at the base of the peat profile and fitting the C input rate (I_permafrost) and the decomposition constant (k_permafrost) using the measured data (Trumbore and Harden, 1997; Clymo, 1984). We estimated post-thaw C accumulation rates in the bog peat strata by measuring C stored in only these strata across the collapse-scar bog chronosequence, using similar methods as O'Donnell et al. (2012a), fitting Eqn 2 to the relationship between the C stored bog peat (C(t)) and collapse-scar bog age (t), which was based on either ¹⁴C dating or ²¹⁰Pb. Using this relationship, we were able to estimate the C input rate (I_bog) and decomposition constant (k_bog) for bog peat.

For this study, we expand upon the work of O'Donnell et al. (2012a) by disentangling the effects of peatland initiation age (and subsequent permafrost aggradation) and the timing of permafrost thaw on post-thaw C loss and recovery. Our approach differs in several key respects. First, as basal ages suggest that each site initiated as a peatland at a different time, we first incorporated peatland initiation ages, which were not considered by O'Donnell et al. (2012a). Second, we calculated the absolute time that each thawed feature was a permafrost plateau (basal age minus thaw age), rather than assuming that the forest peat was the same age as the extant permafrost plateau, and therefore would have the same amount of C stocks, as the still-frozen permafrost plateau. Using these dates, we developed regression equations of C stocks vs. time for the permafrost plateau cores and forest peat C stocks, which allowed us to more accurately calculate the total C loss from the formerly frozen peat plateau.

O'Donnell et al. (2012a) used an exponential decay equation to determine the loss of organic carbon (OC) stocks from the forest peat along each stage of the chronosequence:

$urn:x-wiley:13541013:media:gcb13403:gcb13403-math-0003$ (3)

where y₀ is the y-intercept going through zero at the time of thaw, a is the asymptote, k is the decomposition constant for the thawed forest peat, and t is time since thaw. However, this approach assumes that the permafrost plateau peat should have comparable stocks to the forest peat in the thawed bog cores and any difference between the peat plateau C stocks and the post-thaw forest C stocks within the bog would represent loss. Because the C stocks within the currently frozen peat plateau had longer to accumulate the forest-based C stocks within the thawed bogs, this approach likely overestimates loss. To measure loss of forest-based peat in a way that does not assume that all sites initiated as peatlands at the same time, nor that they thawed at the same time, we took the following steps. First, we generated two regression slopes. The first equation modeled the C stocks of the still-frozen permafrost plateau peat C stocks vs. the amount of time each site has had permafrost (t_plateau; where t_plateau = t_basal - t_sampling; Figs 2 and S2b). The second regression equation modeled the post-thaw forest peat C stocks vs. the amount of time the site had permafrost (t_forest; where t_forest = t_basal - t_thaw; Figs 2, 4b and S2b). Next, we computed the predicted amount of permafrost plateau-based forest peat C (C₀) based on t_plateau, as determined by the first regression equation, and the amount of forest peat C (t_forest) remaining after thaw (C_t; Fig. 4b). This procedure was performed for each collapse-scar bog core for which we had both a basal and thaw date (and thus could calculate t_forest). Finally, we plotted the fraction of initial forest peat C remaining (C_t/C₀) on a natural log scale, to obtain the asymptote (a) and decomposition constant (k), fitting k_forest and a empirically at time t after permafrost thaw using the equation

$urn:x-wiley:13541013:media:gcb13403:gcb13403-math-0004$ (4)

Given the relatively small sample size and degree of scatter in the C stock vs. years as plateau peat for our Innoko and Koyukuk data, we increased the sample size by including sites with similar features from across the northern discontinuous permafrost zone (SI Fig. 2b; Treat et al., 2016). We used a linear regression to test the significance of the relationship between age and C stock, as well as a site-by-age interaction that would indicate that the relationship differed among sites. The interaction term was not significant (P = 0.664) and inclusion of site as an explanatory variable did not significantly improve the model fit (F_3,15= 0.4, P = 0.78).

Mass balance model of OC accumulation and loss

In order to move beyond the limitations of understanding the evolution of peat C in response to permafrost aggradation and degradation, we also used a simple mass balance model, originally developed to evaluate the effects of wildfire on ecosystem C balance in the boreal region (Harden et al., 2000), and subsequently modified to examine the interactive effects of wildfire and permafrost on C balance (O'Donnell et al., 2011) and the effects of permafrost thaw on C dynamics in boreal peatlands (O'Donnell et al., 2012a). We modified it further to reconstruct peat C accumulation and loss using the data to determine the I, k, and a parameters from Eqns 2 and 4, above. Briefly, we used a ‘forward’ modeling approach (e.g., Zhuang et al., 2002) with peatland initiation timing and thermokarst events scheduled at specific times during the model run to reflect peatland development and collapse-scar bog formation dates observed across the chronosequence. The model operates on decadal time steps, tracking changes in C stocks calculated by Eqn 1. To initiate each model run, we simulated prethaw C accumulation using averaged I_permafrost and k_permafrost values derived from the Koyukuk and Innoko permafrost plateaus. Next, we prescribed a thermokarst event, which was designed to reflect a transition from permafrost plateau to collapse-scar bog. To model post-thaw C accumulation and loss, we separately tracked bog peat C using I_bog and k_bog parameters, and thawed forest peat C using the k_forest parameter, as observed across the chronosequences.

We conducted a full factorial model simulation using a spreadsheet model to examine the interactive effects of peatland initiation time (10, 7, 5, 3.5, and 2 ki years BP; ki = peatland initiation in thousands of years) and permafrost thaw timing (100, 500, 750, 1000, 1500 years BP). While O'Donnell et al. (2012a) used site-specific C-cycling parameters in their model, here we bootstrapped the I and k values for both permafrost plateau and post-thaw peat accumulation for the Innoko and Koyukuk data combined, as well as the k_forest value for the post-thaw forest peat C loss (Table 3; O'Donnell et al., 2012a). This process generated 1500 randomized samples, for which we took the mean of the values within the 95% confidence intervals for our model. Because the forest loss could only be constrained by initiation ages and thaw ages for Innoko due to a lack of adequate age control in Koyukuk cores, the decomposition of forest loss is based solely on Innoko data, although the goal was to develop a more generalizable model for peatlands with thawing permafrost in the boreal region.

Table 3. Summary of peat carbon accumulation parameters and statistics for Koyukuk and Innoko NWR study region

		I (g C m⁻² yr⁻¹)	k (yr⁻¹)	r ²	P
Koyukuk	Permafrost plateau
(O'Donnell et al., 2012a,b)	Observed data	23 ± 8	0.0001 ± 0.0001	0.93	<0.0001
	Model input	15	0.0001
	Shallow bog peat
	Observed data	61 ± 2	0.0014 ± 0.0004	0.98	<0.0001
	Model input	61	0.0018
	Thawed forest peat
	Observed data		0.0140 ± 0.0093	0.78	0.1
	Model input		0.0047
Innoko	Permafrost plateau
(This study)	Observed data	25 ± 2	0.0004 ± 0.00001	1.00	<0.0001
	Model input	25	0.0004
	Shallow bog peat
	Observed data	82.8 ± 2	0.002 ± 0.0004	0.99	<0.0001
	Thawed forest peat
	Observed data		0.0002	0.68
Combined Parameters, model input	Permafrost Plateau	24 (22–26)	0.00025 (0.0001–0.0004)
	Shallow bog peat	71.9 (59–85)	0.0017 (0.001–0.0024)
	Thawed forest peat		0.0001868
	Thawed forest peat for model	0.0002 (0.00002–0.0006)

Results

Permafrost plateaus and forest peat beneath intermediate and old bogs at Innoko began accumulating peat following drainage of a thermokarst lake between 2000 and 3500 cal yr BP, as inferred from the presence of lake sediments beneath the peat (SI Table 1; Kanevskiy et al., 2014). Forest peats beneath the young collapse bogs at Innoko are significantly younger (~250–1500 cal yr BP) than the forest peats under the intermediate and old bogs (~2000–3500 cal yr BP; Table 2), based on radiocarbon measurements of peat. The forest peat underlying the young bogs, therefore, formed later and also thawed most recently. In contrast, the Koyukuk forest peat initiated ~7500 to 10 000 cal yr BP (Table 1; O'Donnell et al., 2012a). At both sites, thermokarst drainage was mostly rapid, reflected by a sharp transition between the lake sediments and the accumulation of peat indicative of a black spruce forest peat plateau (e.g., spruce needles, ericaceous shrubs, feather moss) typically associated with permafrost plateaus. In few cases at Innoko, wet fen vegetation (mostly sedges and Sphagnum riparium) formed initially, followed by quick succession to a forested bog (Sphagnum spp., ericaceous shrubs, and black spruce trees), most likely indicative of permafrost aggradation. The cryostratigraphy in the frozen peat cores collected from the permafrost plateau indicates that freezing occurred syngenetically in the case of rapid lake drainage, or quasi-syngenetically in the case of slower or incomplete lake drainage (Kanevskiy et al., 2014). Plant macrofossils in one of the plateau cores from Innoko suggest initial thawing and eventual refreezing, indicating that permafrost aggradation occurred epigenetically for at least a portion of the record (Kanevskiy et al., 2014; Table S1).

Core-based data from the Innoko chronosequence

Total peat C stocks (forest peat) were higher in the permafrost plateaus than the total peat C stocks (forest peat + collapse-scar bog peat) in the young and intermediate bogs, but about the same as the old bog (Tables 1 and 2, Fig. 3a). The similar basal ages of the intermediate and old bogs, and the difference in their overall stocks, suggest that forest peat was lost as a result of thawing (Table 1; Fig. 3a). The young bogs, however, had relatively low overall total C stocks and highly variable basal ages, leading to substantial uncertainty in assessing their peat loss. Apparent long-term C accumulation rates for the new peat formed in collapse-scar bog (bog peat) are highest in the youngest bogs (327 ± 255 g C m⁻² yr⁻¹) and decrease with time since collapse (113 ± 100 g C m⁻² yr⁻¹ for intermediate and 64.0 ± 22 g C m⁻² yr⁻¹ for old) (Table 2; Fig. 3c). The permafrost plateau sites have much lower long-term apparent C accumulation rates compared with collapse-scar bog peat (35 g C m⁻² yr⁻¹ for the mesic plateau and 42.4 g C m⁻² yr⁻¹ for the drying margin Fig. 3c). The median C/N ratios of frozen forest peat in permafrost plateaus, not including the uppermost fresh peat litter, and bog peat in young, intermediate, and old bogs were similar, while the median C/N ratio in thawed forest peat underlying bog peat was half that of peat in permafrost plateaus (Fig. 3d).

Regression-based estimates of C loss

The rates of C accumulation in new bog peat and rates of C loss from thawed old forest peat, as estimated through regression modeling, indicate rapid rates of loss of permafrost peat C and slow gains of post-thaw bog peat. C stock changes at both Innoko and Koyukuk showed more rapid accumulation early post-thaw and slower rates thereafter (Fig. 3a). The regression model (Fig. S2b; Fig. 3b) of C stocks with time for sites that remain a permafrost plateau vs. the forest peat in sites that have thawed (Fig. 3b; Fig. S2b) suggests that ~30 ± 20% of the forest peat C had been lost from each of the thermokarst peats (Fig. S2b, Fig. 3b). Using the regression equation for the predicted permafrost plateau C stocks and comparing it to the actual forest C stocks for the forest regression equation, we fitted an exponential decay equation to the data (Fig. 4c; r²= 0.68) that allowed for different starting C stocks. Although the k value derived for long-term forest peat loss (k = 0.0002) is lower than or on par with both the k constants for the permafrost plateau (k = 0.00025) and bog peat (k = 0.0017) (Table 3), losses of forest peat were large (up to 55% after 1500 years) over the long term (Fig. 4b). Bootstrapping the potential k values from the regression curve (Fig. 4c, Table 3) resulted in an overall loss of forest C ranging from 40% to up to 75%, with the largest loss occurring in the first decade.

Permafrost thaw model

Peatland initiation age has a large effect on the total loss of C stocks following permafrost thaw (Figs 5 and 6). The sites that initiated as peatlands the longest time ago but that thawed most recently lose the most forest C overall, while the most recently initiated sites that thawed the longest time ago lose the least amount of carbon relative to the prethaw C stocks (Fig. 5a). Thus, the loss is proportional to the starting C stock and therefore older peats lose more C to thawing. C losses ranged from 35 to 45 kg C m⁻² in peatlands initiating at 10 ki to ~15 to 5 kg C m⁻² in the youngest initiation category (2 ki) (Fig. 5a).

Peatland initiation age also has an effect on the time it takes for a thermokarst peat to recover the C lost to thaw. The youngest sites recovered C losses after centuries, while the oldest peatlands do not recover their C stocks after 1500 years (Fig. 6). When comparing the C stocks to plateaus that never thawed, the 10ki and 7.5ki categories have lower C stocks in all thaw categories by −7 to −28 kg C m⁻² (Fig 5b; negative indicates loss). The two youngest initiation ages (2, 3.5 ki) only have lower C stocks than sites that never thawed in the 100-year thaw category, and end up with more C than permafrost plateau in <500 years (Fig. 5b). The 5 ki peatland category achieves C stocks higher than the permafrost plateau ~1000 years after thaw. All initiation ages that thawed 100 years ago had lower C stocks (−5 to −28 kg C m⁻²) than prethaw stocks (Fig. 5c). However, overall net C stocks gains relative to prethaw (positive values) were highest in the most recently initiated site that thawed the longest time ago (up to ~35 kg C m⁻²) (Figs 5c and 6), and only the oldest peatland (10 ki) never gains all of the carbon back it lost following permafrost thaw. Because the asymptote derived from the data is relatively low (a = 4.1184), it continues to lose a significant amount of peat C, totaling 55% of the total stock in the 1500-yr BP thaw category.

For each model scenario, the bog was a net source of C to the atmosphere in the first decade following thaw (Fig. 5d), even after accounting for C gains from bog peat accumulation in the first decade. Losses modeled for the first decade ranged from −3.5 kg C m⁻²yr⁻¹ in the 10 ki category and <−0.5 kg C m⁻² yr⁻¹ in the 2 ki category. After the first decade, each site returns to a net annual carbon sink (greater bog C accumulation than forest C loss), although C lost to thaw may not be recovered for decades to millennia.

Discussion

Generalized model for C dynamics of thaw

The approach presented here allows for a more generalized model for C dynamics following permafrost thaw, improving upon the chronosequence approach of O'Donnell et al. (2012a,b) and Johnston et al. (2014) by taking into account differences in timing of peat initiation as well as thaw. The results presented here reduce the apparent C loss based on the chronosequence approach used by O'Donnell et al. (2012a) at Koyukuk Flats and Johnston et al. (2014) at Innoko by almost half. Such a generalized model has potential for more regional assessments where landscapes vary in peatland age and timing of thaw. Accounting for the differences in the amount of time that each site existed as a permafrost plateau (basal age – thaw age) allowed for more accurate portrayal of C loss. Because peat accumulates according to Eqn 1, sites that existed as peat plateaus longer would have higher C stocks upon thaw (Figs S3b and 4b).

Post-thaw dynamics of deep peat

The decline in post-thaw C stocks from forest peat, regardless of bog age, suggests that rapid (years to decades) decomposition of forest peat C occurs following permafrost thaw. However, the small number of samples that met the criteria to perform the revised regression analysis results in high uncertainty of the decomposition constant, and using Eqn 4 based on these sparse data points resulted in a range of 40–75% C loss of forest peat after 1500 years in the mass balance model, rather than our regression-based loss of ~30% (Figs 4b and S2b). The long-term decomposition constants of the thawing forest peat are not significantly different from that of the permafrost plateau and the shallow bog peat, which again points toward a rapid, initial loss upon thaw. Initial, rapid loss rates are consistent with those for fresh plant litter (Moore et al., 2007). Therefore, while the regression-based C loss approach is theoretically sound, more data are needed to evaluate the degree and rate of decomposition of forest peat following permafrost thaw. The degree of scatter in our data (Fig. S2b) highlights the variability associated with site-specific conditions, such as differences in hydrology, nutrients, microsite temperature variations, and vegetation (Loisel et al., 2014). The scatter is also likely influenced by the difficulty associated with accurately measuring the timing of transitions (Fig. S3; Jones et al., 2013), due to the influence of thaw slumping and heaving, which can create age reversals and cluster ages around narrow depth horizons. Nonetheless, the ~30% loss calculated from our regression approach (Fig. 4b, c) is in agreement with a 28% anaerobic C loss calculated using a mass balance approach from thawing permafrost in yedoma regions (Walter Anthony et al., 2014), lending support to our results that substantial anaerobic C loss is not unprecedented.

Because a decrease in C/N can indicate decomposition, particularly when peat composition is similar (Kuhry & Vitt, 1996), a decrease in the C/N of the forest peat relative to both the permafrost plateau, not including the uppermost undecomposed peat layers, and the collapse-scar bog peat (Fig. 3d) indicates that forest peat is indeed more decomposed than plateau peat. The transition from frozen to unfrozen state likely accelerates decomposition of labile organic matter in deep forest peat layers, leading to old C loss (O'Donnell et al., 2012a; Olefeldt et al., 2014; Treat et al., 2014), while also simply exposing previously frozen carbon to microbial processes (Schuur et al., 2015).

Collapse-scar bog formation and recovery to prethaw C stocks

C loss from forest peat is slowly recovered with the accumulation of collapse-scar bog peat C. In each thaw category, it takes each peatland several hundred years to a millennium (or more) to recover to its prethaw C stocks, depending on both the peatland initiation age and timing of thaw (Fig. 6). The importance of the initiation age (and therefore prethaw C stocks) can be observed in Figs 5 and 6, where the oldest two initiation categories (10ki, 7ki) cannot recover their prethaw C stocks after 1500 years, similar to what O'Donnell et al. (2012a) found for Koyukuk. This is in part because bog peat accumulation slows with time, while our decomposition constants allow for continued decomposition of the deep peat, albeit slowly. The likelihood of this continued forest peat loss for millennia is low, as it would eventually move to deeper, colder depths where decomposition is limited by oxygen, carbon, and nutrient availability. The source-to-sink timing, as well as the recovery of prethaw stocks, also depends on the rate of bog peat dynamics (input, loss, net accumulation). Differences in post-thaw bog peat accumulation, which likely depend on local site conditions and climate, can result in faster recovery of the C stocks (Fig. 4a).

Environmental, biogeochemical, and physical controls on net C balance

The chronosequence, or space-for-time approach, to understand gains and losses of C in a permafrost peatland following thaw, is useful in that it allows us to evaluate the possible magnitude of C loss from the system. Studies that rely on single-core analyses (Robinson & Moore, 1999, 2000; Myers-Smith et al., 2008; Jones et al., 2013) can only record what remains and therefore conclude that C accumulates faster post-thaw than in permafrost plateaus. Recent studies, however, have found that permafrost peat plateaus, particularly those with syngenetic permafrost, and nonpermafrost peatlands may accumulate at similar rates over long periods (Olefeldt et al., 2012; Treat et al., 2016), suggesting that the inputs may be larger in nonpermafrost peatlands but that they also decompose faster.

The utility of the chronosequence approach in thawing permafrost depends on careful examination of assumptions. First, it assumes that conditions for peat accumulation are the same along each stage of the chronosequence. Factors that could alter the rate of C accumulation include site-level differences in nutrients, hydrology, vegetation, active layer depth, and ice content, which may explain some of the variability in Fig. S2b. Second, it assumes that each location along the chronosequence is equally susceptible to thaw, when in reality, the susceptibility may vary depending on peat thickness, insulative properties of surface vegetation, and ice content (Camill & Clark, 1998). Third, it assumes that each site has only thawed once and it does not consider the impact of fire on C stocks in these sites, which need to be explored to more accurately assess C loss in these fire-prone ecosystems.

In order to evaluate the short-term C dynamics, we can compare our decadal model of net flux rates to field measurements of gaseous flux. Our model results show high rates of C fluxes in the first decade following thaw (Fig 5d), up to 3500 g C m⁻²yr⁻¹, which are unreasonably high compared with observed fluxes (e.g., Johnston et al., 2014). The fluxes measured at Innoko during the growing season were lower by an order of magnitude and counteracted by a net uptake of CO₂ (Johnston et al., 2014) or near-neutral C losses (e.g., Wickland et al., 2006; Myers-Smith et al., 2008; Euskirchen et al., 2014; Johnston et al., 2014). Given the uncertainty in our data, especially in the youngest thaw categories, we acknowledge that an exponential loss likely overestimates the rate of loss initially and that the loss is more likely to occur over a longer period than years to a few decades.

Growing season efflux likely does not capture total C loss and is dominated by growing season processes (e.g., Prater et al., 2007; Klapstein et al., 2014). However, studies that have quantified the net annual C balance of permafrost thaw are extremely limited. Old C loss has been measured from thawing permafrost using radiocarbon (e.g., Schuur et al., 2009; Nowinski et al., 2010), and nongrowing season C losses amount to 40–50% of total annual flux (Whalen & Reeburgh, 1988; Zona et al., 2016). We note that methane ebullition occurred under snow at Innoko during well installation in April 2011 (Jorgenson MT and Koch J, personal communication), indicating that Johnston et al. (2014) observed only a portion of the annual fluxes. Thus, summer biological activity likely masks deep peat respiration that occurs year-round, which is more likely to be captured during the shoulder seasons (Mastepanov et al., 2008; Song et al., 2012; Pirk et al., 2015).

Physical drivers of peat decomposition

Physical conditions such as temperature, particularly in the case of talik (thaw bulb) formation, also likely factor into the decomposition of the forest peat. At Innoko, the warmest temperatures were observed in the intermediate and old sites (Johnston et al., 2014) and the temperatures at the young bog at Koyukuk were higher than the active layer of the permafrost plateau (O'Donnell et al., 2012a). The high ice content of lowland permafrost plateaus often results in talik formation following permafrost thaw, which remains unfrozen year-round due to the thermal inertia of the water (Jorgenson et al., 2010; O'Donnell et al., 2012a). Indeed, at the young bog site at Koyukuk, the temperature never dropped below 0 °C in the winter (O'Donnell et al., 2012a). These above-freezing temperatures year-round extend active decomposition beyond the growing season. Incubation data, in conjunction with overwinter temperature profiles, have shown high CO₂ and CH₄ production from a talik in an interior Alaskan collapse-scar bog when compared to the C mass of profiles from adjacent permafrost environments (Waldrop et al., 2013).

High levels of water saturation, leading to oxygen limitation, is commonly given as a reason for low decomposition in peatlands, as oxygen diffusion in water is slow and leads to the energy-inefficient use of alternate electron acceptors for respiration (Schuur et al., 2008). However, the apparent loss of deep C despite high water table following permafrost thaw in the collapse-scar bog suggests other mechanisms are at work in these systems, including changes in nutrient availability (Keuper et al., 2012), microbial communities (Mackelprang et al., 2011; McCalley et al., 2014; Hultman et al., 2015), and biogeochemistry of carbon compounds in frozen soils that result in rapid mineralization upon thaw (Vonk et al., 2013; Drake et al., 2015; Ewing et al., 2015).

Regional landscape implications

The permafrost-affected landscapes at Innoko and Koyukuk are a highly patchy mosaic of permafrost plateaus, collapse-scar bogs and fens, and thermokarst lakes of varying ages (Kanevskiy et al., 2014). Furthermore, the soil stratigraphy reveals that some permafrost plateaus and bogs have undergone previous episodes of thawing and refreezing, which would undoubtedly impact C dynamics. While our model examines the role of different landscape ages to better understand the mechanisms of C gain and loss, this study does not address the fate of C under multiple cycles of permafrost aggradation and degradation. Several studies have demonstrated repeated cycles of freezing and thawing in the discontinuous permafrost zone (Zoltai, 1993; Zoltai, 1995), but the amount of time it takes a site to aggrade permafrost post-thaw is difficult to estimate, due to local and regional differences in ground and air temperatures, hydrology, and vegetation. Furthermore, macrofossil evidence for permafrost aggradation is often indistinguishable from late-successional bog stages (Treat et al., 2016), making it difficult to identify the exact timing of permafrost aggradation in a system.

This mosaic of lakes, bogs, and permafrost plateaus then provides a challenge for integrating the C dynamics of ecosystems with widely varying C accumulation and loss rates across the broader landscape. In the boreal region of central Alaska with discontinuous permafrost, Jorgenson et al. (2008) estimated that 5% of the region had thermokarst terrain and 62% had permafrost and that lowland areas had nearly half (35% of total area) of the permafrost and nearly all of the thermokarst (4%). For comparison, Pastick et al. (2015) estimated that the similar Intermontane Boreal ecoregion had near-surface permafrost (upper 1 m) over 45% of the area. These relatively fine-scale processes represent an important challenge for earth system modelers conducting simulations at the regional and circumboreal scale, particularly given the disproportionate impact of thermokarst on C-cycle feedbacks to the atmosphere.

Boreal lowland landscapes hold a large proportion of the overall soil C (Tarnocai et al., 2009; Hugelius et al., 2014), so the history of permafrost degradation has large implications for carbon emissions. Enhanced atmospheric warming, particularly in high-latitude regions, will increasingly destabilize permafrost, especially in the discontinuous permafrost zone, where mean annual air temperatures can go above 0 °C in any given year. The projected atmospheric warming of 1.8 to >6 °C by 2100 (Vaughan et al., 2013), along with increased frequency and severity of wildfires (Turetsky et al., 2011), will warm ground temperatures and completely destabilize discontinuous permafrost by 2100 (Camill, 2005). Within our broader Innoko study area, we roughly estimate that 40% of the landscape has permafrost, 30% of the landscape has old bogs, 10% has young/intermediate bogs, and 10% has old thermokarst lakes, indicating that most of the permafrost has already thawed. Assuming that the remaining permafrost in this area will thaw during the next century (Jafarov et al., 2012), roughly half of the landscape will have young/intermediate bogs with net C loss and half will have previously thawed terrain, which will have net gains of C. The rapid and massive C losses at the young/intermediate bogs will outweigh the C gains of slowly accumulating old bogs in the next decades to several centuries, but eventually (several centuries to millennia), these peatlands will resume overall C burial and net accumulation.

If we expand this estimate to the entire sporadic and discontinuous permafrost zone, which contain 85.1 Pg of C (Tarnocai et al., 2009; Hugelius et al., 2014), we estimate that up to 24 Pg of deep C could be lost to the atmosphere following thaw in the coming years to centuries (Tarnocai et al., 2009; Hugelius et al., 2014), if all permafrost peatlands follow the trend of deep C loss as in our Alaskan study sites. However, following that initial loss, these landscapes will shift back to net C sinks as bogs develop and offset the initial losses from thaw after centuries or millennia, depending on hydrological settings. Moreover, differences in permafrost aggradation processes and lability of parent material could dampen the quantity C loss (i.e., syngenetic vs. epigenetic permafrost aggradation). Therefore, the need exists to study permafrost peat C loss following permafrost thaw from a range of peatland environments, including those that aggraded permafrost epigenetically and those from the southern limits of permafrost, where permafrost can persist with MAAT >1.5 °C but is also most vulnerable to thaw (Jones et al., 2016).

This study highlights significant losses from collapse-scar bog systems that were previously thought to not lose much old C (Robinson & Moore, 2000; Turetsky et al., 2000; Jones et al., 2013). It also highlights the utility of the chronosequence approach compared with single-core analyses to measure long-term C dynamics, as the chronosequence approach allows us to better account for C loss from the system. The results of the mass balance model suggest that losses will eventually (several centuries to millennia) recuperate prethaw C stocks through gains in bog peat accumulation, but the absence of the confining permafrost surrounding the collapse-scar bogs as permafrost thaw progresses could result in changes in drainage patterns that could lower accumulation as water and nutrients are redistributed on the landscape, reducing their overall C sink capacity (Jorgenson et al., 2013).

Because the overall amount of C loss from our field sites shows a proportional loss to the age of the site and therefore how much peat had accumulated prior to thaw, an understanding of landscape development history is important for predicting the C dynamics of these lowland landscapes to thaw. While we have determined from our two chronosequences that roughly 30% of forest peat C is lost following thaw, the amount of time it takes for it lose this amount of C remains poorly constrained, and more data are needed to better constrain the loss term of forest peat. In addition, questions remain about how permafrost aggradation processes impact rates of C loss following thaw, as well as about the mechanisms and modes of C loss. Future work should also include well-dated chronosequences in locations where permafrost aggraded epigenetically, as well as sites that have undergone multiple cycles of permafrost aggradation and degradation at locations across the northern permafrost zone to better understand how regional site and climate histories contribute to C dynamics in thawing permafrost peatlands. Because fire is a part of the natural ecology of the boreal forest zone, future models should also account for the impact of fire and peat burning in carbon dynamics as it relates to permafrost thaw.

Acknowledgements

The authors thank the Innoko NWR and Koyukuk NWR for allowing us to collect samples. The manuscript was significantly improved by two anonymous reviewers and Benjamin Jones. This research was funded by the U.S. Geological Survey Climate and Land Use Research and Development Program. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the USA Government.

Supporting Information

Filename	Description
gcb13403-sup-0001-FigS1.pdfPDF document, 35.8 KB	Figure S1. Relationship between ²¹⁰Pb age and ¹⁴C age for selected samples where both were measured.
gcb13403-sup-0002-FigS2.pdfPDF document, 146 KB	Figure S2. (a) Original model of forest peat loss following permafrost thaw, based on O'Donnell et al. (2012a). (b) Data for Permafrost plateau vs. years as plateau, including all discontinuous permafrost thaw records from Treat et al. (2016), including Innoko and Koyukuk (red circles), and forest peat C stocks from Innoko and Koyukuk (blue squares) in addition to two collapse-scar bog records from Meyers-Smith et al. (2008) and Jones et al. (2013) (slightly larger blue squares).
gcb13403-sup-0003-FigS3.pdfPDF document, 432.1 KB	Figure S3. Representative age model generated in BACON (Blaauw, 2010) for an old bog, showing slow accumulation during the ‘forest’ permafrost stage and rapid accumulation post-thaw (80 cm).
gcb13403-sup-0004-TableS1.docxWord document, 95.9 KB	Table S1. Core position and description of lithology.

Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

References

Anthony KW, Zimov SA, Grosse G et al. (2014) A shift of thermokarst lakes from carbon sources to sinks during the Holocene epoch. Nature, 511, 452–456.
10.1038/nature13560
CAS PubMed Web of Science® Google Scholar
Appleby PG, Oldfield F (1978) The calculation of lead-210 dates assuming a constant rate of supply of unsupported 210 Pb to the sediment. Catena, 5, 1–8.
10.1016/S0341-8162(78)80002-2
CAS Google Scholar
Blaauw M, Christen JA (2011) Flexible paleoclimate age-depth models using an autoregressive gamma process. Bayesian Analysis, 6, 457–474.
10.1214/ba/1339616472
Web of Science® Google Scholar
Camill P (1999) Peat accumulation and succession following permafrost thaw in the boreal peatlands of Manitoba, Canada. Ecoscience, 6, 592–602.
10.1080/11956860.1999.11682561
Web of Science® Google Scholar
Camill P (2005) Permafrost thaw accelerates in boreal peatlands during late-20th century climate warming. Climatic Change, 68, 135–152.
10.1007/s10584-005-4785-y
CAS Web of Science® Google Scholar
Camill P, Clark JS (1998) Climate change disequilibrium of boreal permafrost peatlands caused by local processes. American Naturalist, 151, 207–222.
10.1086/286112
CAS PubMed Web of Science® Google Scholar
Camill P, Lynch JA, Clark JS et al. (2001) Changes in biomass, aboveground net primary production, and peat accumulation following permafrost thaw in the boreal peatlands of Manitoba, Canada. Ecosystems, 4, 461–478.
10.1007/s10021-001-0022-3
Web of Science® Google Scholar
Camill P, Barry A, Williams E, Andreassi C, Limmer J, Solick D (2009) Climate-vegetation-fire interactions and their impact on long-term carbon dynamics in a boreal peatland landscape in northern Manitoba, Canada. Journal of Geophysical Research, 114, G04017.
10.1029/2009JG001071
CAS Web of Science® Google Scholar
Chanton JP, Bauer JE, Glaser PA et al. (1995) Radiocarbon evidence for the substrates supporting methane formation within northern Minnesota peatlands. Geochimica et Cosmochimica Acta, 59, 3363–3668.
10.1016/0016-7037(95)00240-Z
Web of Science® Google Scholar
Chanton JP, Glaser PH, Chasar LS et al. (2008) Radiocarbon evidence for the importance of surface vegetation on fermentation and methanogenesis in contrasting types of boreal peatlands. Global Biogeochemical Cycles, 22. doi:10.1029/2008GB003274.
10.1029/2008GB003274
CAS Web of Science® Google Scholar
Christensen TR, Johansson T, Åkerman HJ et al. (2004) Thawing sub-arctic permafrost: effects on vegetation and methane emissions. Geophysical Research Letters, 31. doi:10.1029/2003GL018680.
10.1029/2003GL018680
CAS Web of Science® Google Scholar
Clymo RS (1984) The limits to peat bog growth. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 303, 605–654.
10.1098/rstb.1984.0002
Web of Science® Google Scholar
Connon RF, Quinton WL, Craig JR et al. (2015) The hydrology of interconnected bog complexes in discontinuous permafrost terrains. Hydrological Processes, 29, 3831–3847.
10.1002/hyp.10604
Web of Science® Google Scholar
Drake TW, Wickland KP, Spencer RG et al. (2015) Ancient low–molecular-weight organic acids in permafrost fuel rapid carbon dioxide production upon thaw. Proceedings of the National Academy of Sciences, 112, 13946–13951.
10.1073/pnas.1511705112
CAS PubMed Web of Science® Google Scholar
Euskirchen ES, Edgar CW, Turetsky MR et al. (2014) Differential response of carbon fluxes to climate in three peatland ecosystems that vary in the presence and stability of permafrost. Journal of Geophysical Research: Biogeosciences, 119, 1576–1595.
10.1002/2014JG002683
CAS Web of Science® Google Scholar
Ewing SA, O'Donnell JA, Aiken GR et al. (2015) Long-term anoxia and release of ancient, labile carbon upon thaw of Pleistocene permafrost. Geophysical Research Letters, 42, 10730–10738.
10.1002/2015GL066296
CAS Web of Science® Google Scholar
Harden JW, Trumbore SE, Stocks BJ et al. (2000) The role of fire in the boreal carbon budget. Global Change Biology, 6, 174–184.
10.1046/j.1365-2486.2000.06019.x
PubMed Web of Science® Google Scholar
Houghton RA (2007) Balancing the global carbon budget. Annual Review of Earth and Planetary Sciences, 35, 313–347.
10.1146/annurev.earth.35.031306.140057
CAS Web of Science® Google Scholar
Hugelius G, Strauss J, Zubrzycki S et al. (2014) Estimated stocks of circumpolar permafrost carbon with quantified uncertainty ranges and identified data gaps. Biogeosciences, 11, 6573–6593.
10.5194/bg-11-6573-2014
Web of Science® Google Scholar
Hultman J, Waldrop MP, Mackelprang R et al. (2015) Multi-omics of permafrost, active layer and thermokarst bog soil microbiomes. Nature, 521, 208–212.
10.1038/nature14238
CAS PubMed Web of Science® Google Scholar
Jafarov EE, Marchenko SS, Romanovsky VE (2012) Numerical modeling of permafrost dynamics in Alaska using a high spatial resolution dataset. The Cryosphere, 6, 613–624.
10.5194/tc-6-613-2012
Web of Science® Google Scholar
Johnston CE, Ewing SA, Harden JW et al. (2014) Effect of permafrost thaw on CO2 and CH4 exchange in a western Alaska peatland chronosequence. Environmental Research Letters, 9, 085004.
10.1088/1748-9326/9/8/085004
CAS Web of Science® Google Scholar
Jones MC, Yu Z (2010) Rapid deglacial and early Holocene expansion of peatlands in Alaska. Proceedings of the National Academy of Sciences, 107, 7347–7352.
10.1073/pnas.0911387107
CAS PubMed Web of Science® Google Scholar
Jones MC, Grosse G, Jones BM, Walter AK (2012) Peat accumulation in drained thermokarst lake basins in continuous ice-rich permafrost, northern Seward Peninsula, Alaska. Journal of Geophysical Research-Biogeosciences, 117, G00M07.
10.1029/2011JG001766
CAS Web of Science® Google Scholar
Jones MC, Booth RK, Yu Z, Ferry P (2013) A 2200-year record of permafrost dynamics and carbon cycling in a collapse-scar bog, interior Alaska. Ecosystems, 16, 1–19.
10.1007/s10021-012-9592-5
Web of Science® Google Scholar
Jones BM, Baughman CA, Romanovsky VE et al. (2016) Presence of rapidly degrading permafrost plateaus in southcentral Alaska. The Cryosphere Discussions. doi:10.5194/tc-2016-91.
10.5194/tc‐2016‐91
Web of Science® Google Scholar
Jorgenson MT, Racine CH, Walters JC, Osterkamp TE (2001) Permafrost degradation and ecological changes associated with a warming climate in central Alaska. Climatic Change, 48, 551–579.
10.1023/A:1005667424292
CAS Web of Science® Google Scholar
Jorgenson MT, Yoshikawa K, Kanevskiy M et al. (2008) Permafrost Characteristics of Alaska. Institute of Northern Engineering, University of Alaska, Fairbanks.
Google Scholar
Jorgenson MT, Romanovsky V, Harden JH et al. (2010) Resilience and vulnerability of permafrost to climate change. Canadian Journal of Forest Research, 40, 1219–1236.
10.1139/X10-060
Web of Science® Google Scholar
Jorgenson MT, Harden J, Kanevskiy M et al. (2013) Reorganization of vegetation, hydrology, and soil carbon after permafrost degradation across heterogeneous boreal landscapes. Environmental Research Letters, 8, 035017.
10.1088/1748-9326/8/3/035017
Web of Science® Google Scholar
Kanevskiy M, Jorgenson T, Shur Y et al. (2014) Cryostratigraphy and permafrost evolution in the Lacustrine lowlands of West Central Alaska. Permafrost and Periglacial, 25. Processes doi:10.1002/ppp.1800.
10.1002/ppp.1800
Web of Science® Google Scholar
Kaufman DS, Ager TA, Anderson NJ et al. (2004) Holocene thermal maximum in the western Arctic (0–180 W). Quaternary Science Reviews, 23, 529–560.
10.1016/j.quascirev.2003.09.007
Web of Science® Google Scholar
Keuper F, Bodegom PM, Dorrepaal E et al. (2012) A frozen feast: thawing permafrost increases plant-available nitrogen in subarctic peatlands. Global Change Biology, 18, 1998–2007.
10.1111/j.1365-2486.2012.02663.x
Web of Science® Google Scholar
Klapstein SJ, Turetsky MR, McGuire AD et al. (2014) Controls on methane released through ebullition in peatlands affected by permafrost degradation. Journal of Geophysical Research: Biogeosciences, 119, 418–431.
10.1002/2013JG002441
CAS Web of Science® Google Scholar
Koven CD, Ringeval B, Friedlingstein P et al. (2011) Permafrost carbon-climate feedbacks accelerate global warming. Proceedings of the National Academy of Sciences, 108, 14769–14774.
10.1073/pnas.1103910108
CAS PubMed Web of Science® Google Scholar
Kuhry P, Vitt DH (1996) Fossil carbon/nitrogen ratios as a measure of peat decomposition. Ecology, 77, 271–275.
10.2307/2265676
Web of Science® Google Scholar
Lawrence DM, Slater AG (2005) A projection of severe near-surface permafrost degradation during the 21st century. Geophysical Research Letters, 32, doi:10.1029/2005GL025080.
10.1029/2005GL025080
CAS Web of Science® Google Scholar
Loisel J, Yu Z, Beilman DW et al. (2014) A database and synthesis of northern peatland soil properties and Holocene carbon and nitrogen accumulation. The Holocene, 24, 1028–1042.
10.1177/0959683614538073
Web of Science® Google Scholar
Mackelprang R, Waldrop MP, DeAngelis KM et al. (2011) Metagenomic analysis of a permafrost microbial community reveals a rapid response to thaw. Nature, 480, 368–371.
10.1038/nature10576
CAS PubMed Web of Science® Google Scholar
Mastepanov M, Sigsgaard C, Dlugokencky EJ et al. (2008) Large tundra methane burst during onset of freezing. Nature, 456, 628–630.
10.1038/nature07464
CAS PubMed Web of Science® Google Scholar
McCalley CK, Woodcroft BJ, Hodgkins SB et al. (2014) Methane dynamics regulated by microbial community response to permafrost thaw. Nature, 514, 478–481.
10.1038/nature13798
CAS PubMed Web of Science® Google Scholar
Moore TR, Bubier JL, Bledzki L (2007) Litter decomposition in temperate peatland ecosystems: the effect of substrate and site. Ecosystems, 10, 949–963.
10.1007/s10021-007-9064-5
Web of Science® Google Scholar
Muhs DR, Budahn JR (2006) Geochemical evidence for the origin of late Quaternary loess in central Alaska. Canadian Journal of Earth Sciences, 43, 323–337.
10.1139/e05-115
Web of Science® Google Scholar
Muhs DR, Ager TA, Bettis EA et al. (2003) Stratigraphy and palaeoclimatic significance of Late Quaternary loess–palaeosol sequences of the Last Interglacial-Glacial cycle in central Alaska. Quaternary Science Reviews, 22, 1947–1986.
10.1016/S0277-3791(03)00167-7
Web of Science® Google Scholar
Myers-Smith IH, McGuire AD, Harden JW, Chapin FS (2008) Influence of disturbance on carbon exchange in a permafrost collapse and adjacent burned forest. Journal of Geophysical Research: Biogeosciences, 112, G04017. doi:10.1029/2007JG000423.
10.1029/2007JG000423
Google Scholar
Nossov D, Jorgenson MT, Kanevskiy M et al. (2015) Role of ground ice dynamics and ecological feedbacks in recent ice wedge degradation and stabilization. Journal of Geophysical Research: Earth Surface, 120, 2280–2297.
10.1002/2015JF003602
Google Scholar
Nowinski NS, Taneva L, Trumbore SE, Welker JM (2010) Decomposition of old organic matter as a result of deeper active layers in a snow depth manipulation experiment. Oecologia, 163, 785–792.
10.1007/s00442-009-1556-x
PubMed Web of Science® Google Scholar
O'Donnell JA, Harden JW, McGuire AD et al. (2011) The effect of fire and permafrost interactions on soil carbon accumulation in an upland black spruce ecosystem of interior Alaska: implications for post-thaw carbon loss. Global Change Biology, 17, 1461–1474.
10.1111/j.1365-2486.2010.02358.x
Web of Science® Google Scholar
O'Donnell JA, Jorgenson MT, Harden JW et al. (2012a) The effects of permafrost thaw on soil hydrologic, thermal, and carbon dynamics in an Alaskan peatland. Ecosystems, 15, 213–229.
10.1007/s10021-011-9504-0
CAS Web of Science® Google Scholar
O'Donnell JA, Harden JW, Manies KL, Jorgenson MT (2012b) Soil data for a collapse-scar bog chronosequence in Koyukuk Flats National Wildlife Refuge, Alaska, 2008. U.S. Geological Survey Open-File Report 2012-1230.
Google Scholar
Olefeldt D, Roulet NT, Bergeron O (2012) Net carbon accumulation of a high-latitude permafrost palsa mire similar to permafrost-free peatlands. Geophysical Research Letters, 39, L03501. doi:10.1029/2011GL050355.
10.1029/2011GL050355
CAS Web of Science® Google Scholar
Olefeldt D, Pelletier N, Talbot J, Blodau C, Turetsky M (2014) Peat carbon stocks and potential microbial lability of boreal peatlands with varying permafrost histories. AGU Fall Meeting Abstract: B31G-0108.
Google Scholar
Pastick NJ, Jorgenson MT, Wylie BK et al. (2015) Distribution of near-surface permafrost in Alaska: estimates of present and future conditions. Remote Sensing of Environment, 168, 301–315.
10.1016/j.rse.2015.07.019
Web of Science® Google Scholar
Patton WW Jr, Wilson FH, Labay KA, Shew N (2009) Geologic map of the Yukon-Koyuku Basin, Alaska. US Geological Survey Scientific Investigations Map 2909, scale 1:500,000, 2 sheets.
Google Scholar
Payette S, Delwaide A, Caccianiga M, Beauchemin M (2004) Accelerated thawing of subarctic peatland permafrost over the last 50 years. Geophysical Research Letters, 31, doi:10.1029/2004GL020358.
10.1029/2004GL020358
Web of Science® Google Scholar
Pirk N, Santos T, Gustafson C et al. (2015) Methane emission bursts from permafrost environments during autumn freeze-in: new insights from ground-penetrating radar. Geophysical Research Letters, 42, 6732–6738.
10.1002/2015GL065034
CAS Web of Science® Google Scholar
Prater JL, Chanton JP, Whiting GJ (2007) Variation in methane production pathways associated with permafrost decomposition in collapse scar bogs of Alberta, Canada. Global Biogeochemical Cycles, 21, GB4004.
10.1029/2006GB002866
CAS Web of Science® Google Scholar
Rand J, Mellor M (1985) Ice-coring augers for shallow depth sampling. In: U.S. Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire. 27 pp.
Google Scholar
Reimer PJ, Bard E, Bayliss A et al. (2013) IntCal13 and Marine13 radiocarbon age calibration curves 0–50 000 years cal BP. Radiocarbon, 55, 1869–1887.
10.2458/azu_js_rc.55.16947
CAS Web of Science® Google Scholar
Robinson SD, Moore TR (1999) Carbon and peat accumulation over the past 1200 years in a landscape with discontinuous permafrost, northwestern Canada. Global Biogeochemical Cycles, 13, 591–601.
10.1029/1999GB900008
CAS Web of Science® Google Scholar
Robinson SD, Moore TR (2000) The influence of permafrost and fire upon carbon accumulation in high boreal peatlands, Northwest Territories, Canada. Arctic, Antarctic, and Alpine Research, 32, 155–166.
10.2307/1552447
Web of Science® Google Scholar
Schuur EAG, Bockheim J, Canadell JG et al. (2008) Vulnerability of permafrost carbon to climate change: Implications for the global carbon cycle. BioScience, 58, 701–714.
10.1641/B580807
Web of Science® Google Scholar
Schuur EAG, Vogel JG, Crummer KG et al. (2009) The effect of permafrost thaw on old carbon release and net carbon exchange from tundra. Nature, 459, 556–559.
10.1038/nature08031
CAS PubMed Web of Science® Google Scholar
Schuur EA, McGuire AD, Schädel C et al. (2015) Climate change and the permafrost carbon feedback. Nature, 520, 171–179.
10.1038/nature14338
CAS PubMed Web of Science® Google Scholar
Soil Classification Working Group (1998) The Canadian System of Soil Classification ( 3rd edn). Agriculture and Agri-Food Canada Publication, Ottowa, ON.
Google Scholar
Soil Survey Staff (1998) Keys to Soil Taxonomy, pp. 599. Pocahontas Press, Inc, Blacksburg (VA).
Google Scholar
Song C, Xu X, Sun X et al. (2012) Large methane emission upon spring thaw from natural wetlands in the northern permafrost region. Environmental Research Letters, 7, 034009.
10.1088/1748-9326/7/3/034009
Web of Science® Google Scholar
Strauss J, Schirrmeister L, Mangelsdorf K et al. (2015) Organic-matter quality of deep permafrost carbon – a study from Arctic Siberia. Biogeosciences, 12, 2227–2245.
10.5194/bg-12-2227-2015
CAS Web of Science® Google Scholar
Tarnocai C, Canadell JG, Schuur EAG et al. (2009) Soil organic carbon pools in the northern circumpolar permafrost region. Global Biogeochemical Cycles, 23, GB2023.
10.1029/2008GB003327
CAS Web of Science® Google Scholar
Treat CC, Wollheim WM, Varner RK et al. (2014) Temperature and peat type control CO2 and CH4 production in Alaskan permafrost peats. Global change biology, 20, 2674–2686.
10.1111/gcb.12572
CAS PubMed Web of Science® Google Scholar
Treat CC, Jones MC, Camill P et al. (2016) Effects of permafrost aggradation on peat properties as determined from a pan-arctic synthesis of plant macrofossils. Journal of Geophysical Research-Biogeosciences, 121, 78–94.
10.1002/2015JG003061
CAS Web of Science® Google Scholar
Trumbore SE, Harden JW (1997) Accumulation and turnover of carbon in organic and mineral soils of the BOREAS northern study area. Journal of Geophysical Research: Atmospheres, 102, 28817–28830.
10.1029/97JD02231
CAS Web of Science® Google Scholar
Turetsky MR, Wieder RK, Williams CJ, Vitt DH (2000) Organic matter accumulation, peat chemistry, and permafrost melting in peatlands of boreal Alberta. Ecoscience, 7, 379–392.
10.1080/11956860.2000.11682608
Web of Science® Google Scholar
Turetsky MR, Wieder K, Halsey L, Vitt D (2002) Current disturbance and the diminishing peatland carbon sink. Geophysical Research Letters, 29, 1526. doi:10.1029/2001GL014000.
10.1029/2001GL014000
Web of Science® Google Scholar
Turetsky MR, Wieder RK, Vitt DH et al. (2007) The disappearance of relict permafrost in boreal North America: effects on peatland carbon storage and fluxes. Global Change Biology, 13, 1922–1934.
10.1111/j.1365-2486.2007.01381.x
Web of Science® Google Scholar
Turetsky MR, Kane ES, Harden JW et al. (2011) More severe burning and accelerated carbon losses due to a changing boreal fire regime. Nature Geoscience, 4, 27–31.
10.1038/ngeo1027
CAS Web of Science® Google Scholar
Urban NR, Eiseinriech SJ, Grigal DF et al. (1990) Mobility and diagenesis of Pb and 210-Pb in peat. Geochimica et Cosmochimica Acta, 54, 332–3346.
10.1016/0016-7037(90)90288-V
Web of Science® Google Scholar
Van Metre PC, Fuller CC (2009) Dual-core mass-balance approach for evaluating Mercury and 210Pb Atmospheric fallout and focusing to lakes. Environmental Science and Technology Letters, 43, 26–32.
10.1021/es801490c
CAS PubMed Web of Science® Google Scholar
Vaughan DG, Comiso JC, Allison I et al. (2013) Observations: cryosphere. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds TF Stocker, D Qin, G-K Plattner, M Tignor, SK Allen, J Boschung, A Nauels, Y Xia, V Bex, PM Midgley). Cambridge University Press, Cambridge, UK and New York, NY, USA.
Google Scholar
Vonk JE, Mann PJ, Davydov S et al. (2013) High biolability of ancient permafrost carbon upon thaw. Geophysical Research Letters, 40, 2689–2693.
10.1002/grl.50348
CAS Web of Science® Google Scholar
Waldrop MP, Blazewicz S, Jones M et al. (2013) Microbial communities of the deep unfrozen: Do microbes in taliks increase permafrost carbon vulnerability? American Geophysical Union, Fall Meeting 2013, abstract #B32C-03
Google Scholar
Walter KM, Edwards ME, Grosse G (2007) Thermokarst lakes as a source of atmospheric CH4 during the last deglaciation. Science, 318, 633–636.
10.1126/science.1142924
CAS PubMed Web of Science® Google Scholar
Whalen SC, Reeburgh WS (1988) A methane flux time series for tundra environments. Global Biogeochemical Cycles, 2, 399–409.
10.1029/GB002i004p00399
CAS Web of Science® Google Scholar
Wickland KP, Striegl RG, Neff JC, Sachs T (2006) Effects of permafrost melting on CO2 and CH4 exchange of a poorly drained black spruce lowland. Journal of Geophysical Research: Biogeosciences, 111, G02011.
10.1029/2005JG000099
CAS Web of Science® Google Scholar
Wisser D, Marchenko S, Talbot J, Treat C, Frolking S (2011) Soil temperature response to 21st century global warming: the role of and some implications for peat carbon in thawing permafrost soils in North America. Earth System Dynamics, 2, 1–8.
10.5194/esd-2-121-2011
Web of Science® Google Scholar
Zhuang Q, McGuire AD, O'Neill KP et al. (2002) Modeling soil thermal and carbon dynamics of a fire chronosequence in interior Alaska. Journal of Geophysical Research, Atmospheres, 107, No. D1, 8147. doi:10/1029/2001JD001244.
Google Scholar
Zoltai SC (1993) Cyclic development of permafrost in the peatlands of Northwestern Alberta, Canada. Arctic and Alpine Research, 25, 240–246.
10.2307/1551820
Web of Science® Google Scholar
Zoltai SC (1995) Permafrost distribution in peatlands of west-central Canada during the Holocene warm period 6000 years BP. Géographie physique et Quaternaire, 49, 45–54.
10.7202/033029ar
Web of Science® Google Scholar
Zona D, Gioli B, Commane R et al. (2016) Cold season emissions dominate the Arctic tundra methane budget. Proceedings of the National Academy of Sciences, 113, 40–45.
10.1073/pnas.1516017113
CAS PubMed Web of Science® Google Scholar

Citing Literature

Volume23, Issue3

March 2017

Pages 1109-1127

Rapid carbon loss and slow recovery following permafrost thaw in boreal peatlands

Abstract

Introduction