Northern peatlands have accumulated large carbon (C) stocks, acting as a long-term atmospheric C sink since the last deglaciation. How these C-rich ecosystems will respond to future climate change, however, is still poorly understood. Furthermore, many northern peatlands exist in regions underlain by permafrost, adding to the challenge of projecting C balance under changing climate and permafrost dynamics. In this study, we used a paleoecological approach to examine the effect of past climates and local disturbances on vegetation and C accumulation at a peatland complex on the southern Seward Peninsula, Alaska over the past ∼15 ka (1 ka = 1000 cal yr BP). We analyzed two cores about 30 m apart, NL10-1 (from a permafrost peat plateau) and NL10-2 (from an adjacent thermokarst collapse-scar bog), for peat organic matter (OM), C accumulation rates, macrofossil, pollen and grain size analysis.

A wet rich fen occurred during the initial stages of peatland development at the thermokarst site (NL10-2). The presence of tree pollen from Picea spp. and Larix laricinia at 13.5–12.1 ka indicates a warm regional climate, corresponding with the well-documented Bølling–Allerød warm period. A cold and dry climate interval at 12.1–11.1 ka is indicated by the disappearance of tree pollen and increase in Poaceae pollen and an increase in woody material, likely representing a local expression of the Younger Dryas (YD) event. Following the YD, the warm Holocene Thermal Maximum (HTM) is characterized by the presence of Populus pollen, while the presence of Sphagnum spp. and increased C accumulation rates suggest high peatland productivity under a warm climate. Toward the end of the HTM and throughout the mid-Holocene a wet climate-induced several major flooding disturbance events at 10 ka, 8.1 ka, 6 ka, 5.4 ka and 4.7 ka, as evidenced by decreases in OM, and increases in coarse sand abundance and aquatic fossils (algae Chara and water fleas Daphnia). The initial peatland at permafrost site (NL10-1) is characterized by rapid C accumulation (66 g C m⁻² yr⁻¹), high OM content and a peak in Sphagnum spp. at 5.8–4.6 ka, suggesting the lack of permafrost. A transition to extremely low C accumulation rates of 6.3 g C m⁻² yr⁻¹ after 4.5 ka at this site suggests the onset of permafrost aggradation, likely in response to Neoglacial climate cooling as documented across the circum-Arctic region. A similar decrease in C accumulation rates also occurred at non-permafrost site NL10-2. Time-weighted C accumulation rates are 21.8 g C m⁻² yr⁻¹ for core NL10-1 during the last ∼6.5 ka and 14.8 g C m⁻² yr⁻¹ for core NL10-2 during the last ∼15 ka. Evidence from peat-core analysis and historical aerial photographs shows an abrupt increase in Sphagnum spp. and decrease in area of thermokarst lakes over the last century, suggesting major changes in hydrology and ecosystem structure, likely due to recent climate warming.

Our results show that the thermokarst–permafrost complex was much more dynamic with high C accumulation rates under warmer climates in the past, while permafrost was stabilized and C accumulation slowed down following the Neoglacial cooling in the late Holocene. Furthermore, permafrost presence at local scales is controlled by both regional climate and site-specific factors, highlighting the challenge in projecting responses of permafrost peatlands and their C dynamics to future climate change.