Description: Thermokarst lakes formed across vast regions of Siberia and Alaska during the last deglaciation and are thought to be a net source of atmospheric methane and carbon dioxide during the Holocene epoch1, 2, 3, 4. However, the same thermokarst lakes can also sequester carbon5, and it remains uncertain whether carbon uptake by thermokarst lakes can offset their greenhouse gas emissions. Here we use field observations of Siberian permafrost exposures, radiocarbon dating and spatial analyses to quantify Holocene carbon stocks and fluxes in lake sediments overlying thawed Pleistocene-aged permafrost. We find that carbon accumulation in deep thermokarst-lake sediments since the last deglaciation is about 1.6 times larger than the mass of Pleistocene-aged permafrost carbon released as greenhouse gases when the lakes first formed. Although methane and carbon dioxide emissions following thaw lead to immediate radiative warming, carbon uptake in peat-rich sediments occurs over millennial timescales. We assess thermokarst-lake carbon feedbacks to climate with an atmospheric perturbation model and find that thermokarst basins switched from a net radiative warming to a net cooling climate effect about 5,000 years ago. High rates of Holocene carbon accumulation in 20 lake sediments (47 ± 10 grams of carbon per square metre per year; mean ± standard error) were driven by thermokarst erosion and deposition of terrestrial organic matter, by nutrient release from thawing permafrost that stimulated lake productivity and by slow decomposition in cold, anoxic lake bottoms. When lakes eventually drained, permafrost formation rapidly sequestered sediment carbon. Our estimate of about 160 petagrams of Holocene organic carbon in deep lake basins of Siberia and Alaska increases the circumpolar peat carbon pool estimate for permafrost regions by over 50 per cent (ref. 6). The carbon in perennially frozen drained lake sediments may become vulnerable to mineralization as permafrost disappears7, 8, 9, potentially negating the climate stabilization provided by thermokarst lakes during the late Holocene.

Description: The permafrost zone is expected to be a substantial carbon source to the atmosphere, yet large-scale models currently only simulate gradual changes in seasonally thawed soil. Abrupt thaw will probably occur in 〈20% of the permafrost zone but could affect half of permafrost carbon through collapsing ground, rapid erosion and landslides. Here, we synthesize the best available information and develop inventory models to simulate abrupt thaw impacts on permafrost carbon balance. Emissions across 2.5 million km2 of abrupt thaw could provide a similar climate feedback as gradual thaw emissions from the entire 18 million km2 permafrost region under the warming projection of Representative Concentration Pathway 8.5. While models forecast that gradual thaw may lead to net ecosystem carbon uptake under projections of Representative Concentration Pathway 4.5, abrupt thaw emissions are likely to offset this potential carbon sink. Active hillslope erosional features will occupy 3% of abrupt thaw terrain by 2300 but emit one-third of abrupt thaw carbon losses. Thaw lakes and wetlands are methane hot spots but their carbon release is partially offset by slowly regrowing vegetation. After considering abrupt thaw stabilization, lake drainage and soil carbon uptake by vegetation regrowth, we conclude that models considering only gradual permafrost thaw are substantially underestimating carbon emissions from thawing permafrost.

Description: Thermokarst (thaw) lakes emit methane (CH4) to the atmosphere, with the carbon (C) originating from terrestrial sources such as the Holocene soils of the lakes’ watersheds, thaw of Holocene- and Pleistoceneaged permafrost soil beneath and surrounding the lakes, and decomposition of contemporary organic matter (OM) in the lakes. However, the relative magnitude of CH4 production in surface lake sediments versus deeper thawed permafrost horizons is not well understood. We assessed anaerobic CH4 production potentials from 22 depths along a 590 cm long lake sediment core from the center of an interior Alaska thermokarst lake, Vault Lake, that captured the entire package of surface lake sediments, the talik (thaw bulb), and the top 40 cm of thawing permafrost beneath the talik. We also studied the adjacent Vault Creek permafrost tunnel that extends through icerich yedoma permafrost soils surrounding the lake and into underlying fluvial gravel. Our results show, in the center of a first generation thermokarst-lake, whole-column CH4 production is dominated by methanogenesis in the organic-rich surface lake sediments [151 cm thick; mean ± SD 5.95 ± 1.67 μg C-CH4 per g dry weight sediment per day (g dw−1 d−1); 125.9 ± 36.2 μg C-CH4 per g organic carbon per day (g Corg−1 d−1)]. The organic-rich surface sediments contribute the most (67%) to whole-column CH4 production despite occupying a lesser fraction (26%) of sediment column thickness. High CH4 production potentials were also observed in recently-thawed permafrost (1.18 ± 0.61 μg C-CH4 g dw−1 d−1; 59.60 ± 51.5 μg CCH4 g Corg−1 d−1) at the bottom of the talik, but the narrow thicknesses (43 cm) of this horizon limited its overall contribution to total sediment column CH4 production in the core. Lower rates of CH4 production were observed in sediment horizons representing permafrost that has been thawed in the talik for longer periods of time. The thickest sequence in the Vault Lake core, which consisted of combined Lacustrine silt and Taberite facies (60% of total core thickness), had low CH4 production potentials, contributing only 21% of whole sediment column CH4 production potential. No CH4 production was observed in samples obtained from the permafrost tunnel, whose sediments represent a non-lake environment. Our findings imply that CH4 production is highly variable in thermokarstlake systems and that both modern OM supplied to surface sediments and ancient OM supplied to both surface and deep lake sediments by in situ thaw, as well as shore erosion of yedoma permafrost, are important to lake CH4 production. Knowing where CH4 originates and what proportion of produced CH4 is emitted will aid in estimations of how C release and processing in a thermokarst-lake environment differs from other thawing permafrost and non-permafrost environments. References: Heslop, J.K.; Walter Anthony, K.M.; Sepulveda-Jauregui, A.; Martinez-Cruz, K.; Bondurant, A.; Grosse, G. and Jones, M.C. [2015]: Thermokarst lake methanogenesis along a complete talik profile. Biogeosciences, 12:4317–4331, doi:10.5194/bg-12-4317-2015.

Description: Methane emissions from northern high latitude wetlands are one of the largest natural sources of atmospheric methane, contributing an estimated 20% of the natural terrestrial methane emissions to the atmosphere. Methane fluxes vary among wetland types and are generally higher in peatlands, wetlands with 〉 40 cm of organic soil, than in wetlands with mineral soils. However, permafrost aggradation in peatlands reduces methane fluxes through the drying of the peat surface, which can decrease both methane production and increase methane oxidation within the peat. We reconstruct methane emissions from peatlands during the Holocene using a synthesis of peatland environmental classes determined from plant macrofossil records in peat cores from 〉 250 sites across the pan-arctic. We find methane emissions from peatlands decreased by 20% during the Little Ice Age due to the aggradation of permafrost within peatlands during this period. These bottom-up estimates of methane emissions for the present day are in agreement with other regional estimates and are significantly lower than the peak in peatland methane emissions 1300 years before present. Our results indicate that methane emissions from high latitude wetlands have been an important contributor to atmospheric methane concentrations during the Holocene and will likely change in the future with permafrost thaw.

Description: Permafrost soil organic carbon (C) in the Yedoma region comprises a large fraction of the total circumpolar permafrost C pool, yet estimates based on different approaches during the past decade have led to disagreement in the size and composition of the Yedoma region permafrost C pool. This research aims to reconcile different approaches and show that after accounting for thermokarst and fluvial erosion processes of this interglacial period, the Yedoma region C pool (456 ± 45 Pg C) is the sum of 172 ± 19 Pg Holocene-aged C and 284 ± 40 Pg Pleistocene-aged C. The size of the present-day Pleistocene-aged Yedoma C pool was originally estimated to be 450 Pg based on a mean deposit thickness of 25 m, 1×106 km2 areal extent, 2.6% total organic C content, 1.65times103 kg m−3dry bulk density, and 50% volumetric ice wedge content (Zimov et al. 2006). This estimate assumed that 17% of the Last Glacial Maximum yedoma C stock was lost to greenhouse gas production and emission when 50% of yedoma thawed beneath lakes during the Holocene. However, the regional scale yedoma C pool estimate of Zimov et al. (2006) did not include any Holocene C and assumed that all of the 450 Pg C was Pleistocene-aged. In subsequent global permafrost C syntheses, soil organic C content (SOCC, kg C m−2) data from the Northern Circumpolar Soil C Database (NCSCD) and Zimov et al. (2006) were used to estimate the soil organic C pool for the Yedoma region (450 Pg), assuming only Pleistocene-aged yedoma C from 3 to 25 m (407 Pg), and a mixture of C ages in the 0 to 3 m interval (43 Pg). A more recent synthesis of Yedoma-region C stocks based on extensive sampling by Strauss et al. (2013) took into account lower C bulk density values of yedoma, higher organic C concentrations of yedoma, a larger landscape fraction of thermokarst (70% of Yedoma region area), the larger C concentration of thermokarst, and remote-sensing based quantification of ice-wedge volumes. This synthesis produced lower mean- and median-based estimates of Yedoma-region C, 348+73 Pg and 211 +160/-153 Pg respectively. However, Strauss et al. (2013) focused on the remaining undisturbed yedoma and refrozen surface thermokarst deposits and thus did not include taberite deposits, which are the re-frozen remains of Yedoma previously thawed beneath thermokarst lakes and still present in large quantities on the landscape. In our study (Walter Anthony et al. 2014), we measured the dry bulk density directly on 89 yedoma and 311 thermokarst-basin samples, including taberites, collected in four yedoma subregions of the North Siberian Kolyma Lowlands. Multiplying the organic matter content of an individual sample by the same sample’s measured bulk density yielded an organic C bulk density data set for yedoma samples that was normally distributed. Combining our subregion-specific organic C bulk density results with those of Strauss et al. (2013) for other yedoma subregions extending to the far western extent of Siberian yedoma, we determined a mean organic C bulk density of yedoma for the total Yedoma region (26 ± 1.5 kg C m-3), which is similar to that previously suggested by Strauss et al. (2013) (27 kg C m-3 mean based approach; 16 kg C m-3 median based approach). Our estimate of the organic C pool size of undisturbed yedoma permafrost (129 ± 30 Pg Pleistocene C) in the 396,600 ± 39,700 km2 area that has not been degraded by thermokarst since the Last Glacial Maximum is based on this regional-mean C bulk density value. Our calculation also assumes an average yedoma deposit thickness of 25 m and 50% volumetric massive ice wedge content, as in previous estimates (Zimov et al. 2006, NCSCD; Table 1). Similar results found in the recent study of the Yedoma-region C inventory by Strauss et al.(2013) corroborate our estimate of the undisturbed yedoma C inventory. The size of the remaining yedoma C pool was estimated by Strauss et al. (2013) to be 112 Pg (vs. 129 Pg, this study) based on mean parameter values: organic C bulk density 27 kg C cm-3 (vs. 26.2 kg C cm-3 in this study), yedoma deposit thickness 19.4 m (vs. 25 m in this study), Yedoma volumetric ice wedge content 48% (vs. 50% in this study), and thermokarst extent (70% in both studies). The two studies took different approaches for estimating yedoma deposit thicknesses: Strauss et al. (2013) used 22 field sites from Siberia and Alaska with a mean thickness of 19.4 m; our calculations used a thickness value determined from Russian literature (25 ± 5m, references in Walter Anthony et al. 2014). The mean derived from our limited (n=17) field sites was 38 m in the Kolyma region. The two studies further differ slightly in calculating Yedoma-region area: Strauss et al. (2013), which focused on still frozen deposits vulnerable to future thaw, did not include thawed deposits in present-day lakes, but did include deposits in known smaller yedoma occurrences outside the core Yedoma region such as valleys of NW Canada, Chukotka, and the Taymyr Peninsula. Our study focused on the extent of core-yedoma deposits as well as organic-C stored in present-day Yedoma lake deposits. While differences in yedoma thickness and area values can impact upscaling calculations, efforts are underway by the Yedoma region synthesis IPA Action Group (Strauss and colleagues) to analyze more comprehensive data sets and better constrain the values. Based on our approach that includes a differentiation of thermokarst-lake facies, we estimate that 155 Pg Pleistocene-aged organic C is stored in thermokarst-lake basins and thermoerosional gullies in the Yedoma region of Beringia [155 Pg is the sum of 114 Pg in taberite deposits and 41 Pg in various lacustrine facies]. This 155 Pg Pleistocene-aged C represents the remains of yedoma that thawed and partially decomposed beneath and in thermokarst lakes and streams. Altogether we estimate a total Pleistocene-C pool size of 284 ± 40 Pg for the Beringian Yedoma region in the present day as the sum of Pleistocene C in undisturbed yedoma (129 Pg) and in thermokarst basins (155 Pg). Separately, Holocene-aged organic C assimilated and sequestered in deglacial thermokarst basins in the Yedoma-region is 159 ± 24 Pg. Our upscaling is based on the mean C stocks of individual permafrost exposures (Fig. 2e in Walter Anthony et al. 2014), which were normally distributed. To our knowledge, this is the first study to combine a geomorphologic classification of alas facies with C content, including the deeper lacustrine deposits, for the purpose of systematically upscaling to a regional alas C inventory. We did not measure the C content of Holocene terrestrial soils overlying undisturbed yedoma permafrost; however, applying values from the NCSCD in Siberia for Histels (44.3 kg C m-2, 9% of Yedoma region area), Orthels (26.0 kg C m-2, 17% of Yedoma region area) and Turbels (38.4 kg C m-2, 63% of Yedoma region area) to the extent of 1-m surface deposits overlying the area of undisturbed Yedoma permafrost (396,000 ± 39,600 km2), results in 12.9 ± 1.3 Pg of Holocene C. This calculation assumes that the 70/30 ratio of thermokarst to undisturbed yedoma applies across the Histel, Orthel and Turbel cover classes. Altogether, we estimate the Holocene and Pleistocene organic C pool size in the Yedoma region of Beringia as 456 ± 45 Pg (38% Holocene, 62% Pleistocene). Despite the differences in approaches and locations of study sites, similarities in the meanbased estimates of the Yedoma-region organic C pool size between Strauss et al. (2013) and this study corroborate our findings. Not accounting for diagenetically altered organic C from yedoma thawed in situ beneath lakes (taberites), Strauss et al. (2013) estimated 348 Pg C for the regional pool size. Without taberite C, our estimate would be similar (342 Pg C). For our study, focusing on the C balance shifts from the Pleistocene to the end of the Holocene, we show that taberite deposits are an important component and need to be included in the budget as these deposits are a large C pool that represents diagenetically-altered organic C from yedoma thawed in situ beneath lakes (Table 1b). Our estimate of yedoma-derived taberite deposits underlying thermokarst basins (114 Pg C), would bring the Yedoma-region C pool estimate by Strauss et al. (2013) up to 462 Pg C, which is similar to our estimate of 456 Pg C. In summary, the Yedoma-region organic C value (456 ± 45 Pg C, consisting of Pleistocene and Holocene C) determined by this study is similar to that calculated originally by Zimov et al. (2006) to represent only the Pleistocene yedoma C pool (450 Pg). Subsequently, the Pleistocene-aged yedoma C was considered to be 450 Pg C. Pleistocene-aged Yedoma carbon was considered to be 〉90% of the regional pool by the subsequent NCSCD syntheses for quantification of circumpolar permafrost carbon. The primary difference between the Yedoma-region C pool estimate presented here versus Zimov et al. (2006), which entered the NCSCD syntheses, is that in this study net C gains associated with a widespread thermokarst process are taken into account. The component of Pleistocene yedoma C was reduced in this study by 38% and a new Holocene-thermokarst C pool (159 Pg) was introduced. We lowered the Pleistocene-aged yedoma C pool based on larger, more recent data sets on yedoma’s dry bulk density by this study and Strauss et al. (2013) and based on our more recent map-based analysis showing a 20% larger areal extent of deep thermokarst activity in the Yedoma region. The major implications of this study pertain to the nature and fate of greenhouse gas emissions associated with permafrost thaw in the Yedoma region. Differentiation of the C pool in the Yedoma region (yedoma vs. thermokarst basins) is critical to understanding past and future C dynamics and climate feedbacks. Since a larger fraction of the yedoma landscape has already been degraded by thermokarst during the Holocene (70% instead of 50%), the size of the anaerobically-vulnerable yedoma C pool for the production of methane is 40% lower than that previously calculated. Second, there is concern that permafrost thaw will mobilize and release ’ancient’ organic C to the atmosphere. Assuming average radiocarbon ages of Pleistocene-yedoma and Holocene deposits of 30 kya and 6.5 kya respectively, accounting for the new Holocene-aged thermokarst C pool (159 Pg C) lowers the average age of the current Yedoma-region C pool by about one third. This result is important to global C-cycle modeling since C isotope signatures provide valuable constraints in models. Finally, given differences in permafrost soil organic matter origins for the Pleistocene-aged steppe-tundra yedoma C pool [accumulated under aerobic conditions; froze within decades to centuries after burial; and remained frozen for tens of thousands of years] and the lacustrine Holocene-aged C pool [accumulated predominately under anaerobic conditions and remained thawed for centuries to millennia prior to freezing after lakes drained], it is likely that organic matter degradability differs substantially between these two pools. This has implications for differences in their vulnerability to decomposition and greenhouse gas production under scenarios of permafrost thaw in the future. References: Strauss J., Schirrmeister L, Grosse G, Wetterich W, Ulrich M, Herzschuh U, Hubberten H-W. 2013. The deep permafrost carbon pool of the Yedoma region in Siberia and Alaska. Geophys. Res. Lett. 40, 6165–6170. Walter Anthony K M, Zimov SA, Grosse G, Jones MC, Anthony P, Chapin III FS, Finlay JC, Mack mC, Davydov S, Frenzel P, Frolking S. 2014. A shift of thermokarst lakes from carbon sources to sinks during the Holocene epoch. Nature, 511, 452-456, DOI:10.1038/nature13560. Zimov,SA, Schuur EAG, Chapin FS. 2006. Permafrost and the global carbon budget. Science 312: 1612–1613.

Description: Permafrost presence is determined by a complex interaction of climatic, topographic, and ecological conditions operating over long time scales. In particular, vegetation and organic layer characteristics may act to protect permafrost in regions with a mean annual air temperature (MAAT) above 0°C. In this study, we document the presence of residual permafrost plateaus in the western Kenai Peninsula lowlands of south-central Alaska, a region with a MAAT of 1.5+/-1 °C (1981–2010). Continuous ground temperature measurements between 16 September 2012 and 15 September 2015, using calibrated thermistor strings, documented the presence of warm permafrost (-0.04 to -0.08 °C). Field measurements (probing) on several plateau features during the fall of 2015 showed that the depth to the permafrost table averaged 1.48m but at some locations was as shallow as 0.53 m. Late winter surveys (augering, coring, and GPR) in 2016 showed that the average seasonally frozen ground thickness was 0.45 m, overlying a talik above the permafrost table. Measured permafrost thickness ranged from 0.33 to 〉6.90 m. Manual interpretation of historic aerial photography acquired in 1950 indicates that residual permafrost plateaus covered 920 ha as mapped across portions of four wetland complexes encompassing 4810 ha. However, between 1950 and ca. 2010, permafrost plateau extent decreased by 60.0 %, with lateral feature degradation accounting for 85.0% of the reduction in area. Permafrost loss on the Kenai Peninsula is likely associated with a warming climate, wildfires that remove the protective forest and organic layer cover, groundwater flow at depth, and lateral heat transfer from wetland surface waters in the summer. Better understanding the resilience and vulnerability of ecosystem-protected permafrost is critical for mapping and predicting future permafrost extent and degradation across all permafrost regions that are currently warming. Further work should focus on reconstructing permafrost history in south-central Alaska as well as additional contemporary observations of these ecosystem-protected permafrost sites south of the regions with relatively stable permafrost.

Description: Permafrost influences roughly 80% of the Alaskan landscape (Jorgenson et al. 2008). Permafrost presence is determined by a complex interaction of climatic, topographic, and ecological conditions operating over long time scales such that it may persist in regions with a mean annual air temperature (MAAT) that is currently above 0 °C (Jorgenson et al. 2010). Ecosystem-protected permafrost may be found in these regions with present day climatic conditions that are no longer conducive to its formation (Shur and Jorgenson, 2007). The perennial frozen deposits typically occur as isolated patches that are highly susceptible to degradation. Press disturbances associated with climate change and pulse disturbances, such as fire or human activities, can lead to immediate and irrevocable permafrost thaw and ecosystem modification in these regions. In this study, we document the presence of residual permafrost plateaus on the western Kenai Peninsula lowlands of southcentral Alaska (Figure 1a), a region with a MAAT of 1.5±1 °C (1981 to 2010). In September 2012, field studies conducted at a number of black spruce plateaus located within herbaceous wetland complexes documented frozen ground extending from 1.4 to 6.1 m below the ground surface, with thaw depth measurements ranging from 0.49 to 〉1.00 m. Ground penetrating radar surveys conducted in the summer and the winter provided additional information on the geometry of the frozen ground below the forested plateaus. Continuous ground temperature measurements between September 2012 and September 2015, using thermistor strings calibrated at 0 °C in an ice bath before deployment, documented the presence of permafrost. The permafrost (1 m depth) on the Kenai Peninsula is extremely warm with mean annual ground temperatures that range from -0.05 to -0.11 °C. To better understand decadal-scale changes in the residual permafrost plateaus on the Kenai Peninsula, we analyzed historic aerial photography and highresolution satellite imagery from ca. 1950, ca. 1980, 1996, and ca. 2010. Forested permafrost plateaus were mapped manually in the image time series based on our field observations of characteristic landforms with sharply defined scalloped edges, marginal thermokarst moats, and collapse-scar depressions on their summits. Our preliminary analysis of the image time series indicates that in 1950, permafrost plateaus covered 20% of the wetland complexes analyzed in the four change detection study areas, but during the past six decades there has been a 50% reduction in permafrost plateau extent in the study area. The loss of permafrost has resulted in the transition of forested plateaus to herbaceous wetlands. The degradation of ecosystem-protected permafrost on the Kenai Peninsula likely results from a combination of press and pulse disturbances. MAAT has increased by 0.4 °C/decade since 1950, which could be causing top down permafrost thaw in the region. Tectonic activity associated with the Great Alaska Earthquake of 1964 caused the western Kenai Peninsula to lower in elevation by 0.7 to 2.3 m (Plafker 1969), potentially altering groundwater flow paths and influencing lateral as well as bottom up permafrost degradation. Wildfires have burned large portions of the Kenai Peninsula lowlands since 1940 and the rapid loss of permafrost at one site between 1996 and 2011 was in response to fires that occurred in 1996 and 2005. Better understanding the resilience and vulnerability of the Kenai Peninsula ecosystem-protected permafrost to degradation is of importance for mapping and predicting permafrost extent across colder permafrost regions that are currently warming.

Description: Permafrost presence is determined by a complex interaction of climatic, topographic, and ecological conditions operating over long time scales. In particular, vegetation and organic layer characteristics may act to protect permafrost in regions with a mean annual air temperature (MAAT) above 0 °C. In this study, we document the presence of residual permafrost plateaus on the western Kenai Peninsula lowlands of southcentral Alaska, a region with a MAAT of 1.5 ± 1 °C (1981 to 2010). Continuous ground temperature measurements between 16 September 2012 and 15 September 2015, using calibrated thermistor strings, documented the presence of warm permafrost (−0.04 to −0.08 °C). Field measurements (probing) on several plateau features during the fall of 2015 showed that the depth to the permafrost table averaged 1.48 m but was as shallow as 0.53 m. Late winter surveys (drilling, coring, and GPR) in 2016 showed that the average seasonally frozen ground thickness was 0.45 m, overlying a talik above the permafrost table. Measured permafrost thickness ranged from 0.33 to 〉 6.90 m. Manual interpretation of historic aerial photography acquired in 1950 indicates that residual permafrost plateaus covered 920 ha as mapped across portions of four wetland complexes encompassing 4810 ha. However, between 1950 and ca. 2010, permafrost plateau extent decreased by 60 %, with lateral feature degradation accounting for 85 % of the reduction in area. Permafrost loss on the Kenai Peninsula is likely associated with a warming climate, wildfires that remove the protective forest and organic layer cover, groundwater flow at depth, and lateral heat transfer from wetland surface waters in the summer. Better understanding the resilience and vulnerability of ecosystem-protected permafrost is critical for mapping and predicting future permafrost extent and degradation across all permafrost regions that are currently warming. Further work should focus on reconstructing permafrost history in southcentral Alaska as well as additional contemporary observations of these ecosystem-protected permafrost sites lying south of the regions with relatively stable permafrost.

Description: The sudden collapse of thawing soils in the Arctic might double the warming from greenhouse gases released from tundra, warn Merritt R. Turetsky and colleagues.

Description: The sources of atmospheric methane (CH4) during the Holocene remain widely debated, including the role of high latitude wetland and peatland expansion and fen-to-bog transitions. We reconstructed CH4 emissions from northern peatlands from 13,000 before present (BP) to present using an empirical model based on observations of peat initiation (〉3600 14C dates), peatland type (〉250 peat cores), and contemporary CH4 emissions in order to explore the effects of changes in wetland type and peatland expansion on CH4 emissions over the end of the late glacial and the Holocene. We find that fen area increased steadily before 8000 BP as fens formed in major wetland complexes. After 8000 BP, new fen formation continued but widespread peatland succession (to bogs) and permafrost aggradation occurred. Reconstructed CH4 emissions from peatlands increased rapidly between 10,600 BP and 6900 BP due to fen formation and expansion. Emissions stabilized after 5000 BP at 42 ± 25 Tg CH4 y-1 as high-emitting fens transitioned to lower-emitting bogs and permafrost peatlands. Widespread permafrost formation in northern peatlands after 1000 BP led to drier and colder soils which decreased CH4 emissions by 20% to 34 ± 21 Tg y-1 by the present day.