Vegetation shifts in circum-Arctic permafrost peatlands drive feedbacks with important consequences for peatland carbon budgets and the extent of permafrost thaw under changing climate. Recent shrub expansion across Arctic tundra environments has led to an increase in above-ground biomass, but the long-term spatiotemporal dynamics of shrub and tree growth in circum-Arctic peatlands remain unquantified. We investigate changes in peatland vegetation composition during the Holocene using previously-published plant macrofossil records from 76 sites across the circum-Arctic permafrost zone. In particular, we assess evidence for peatland shrubification at the continental scale. We identify increasing abundance of woody vegetation in circum-Arctic peatlands from ∼8000 years BP to present, coinciding with declining herbaceous vegetation and widespread Sphagnum expansion. Ecosystem shifts varied between regions and present-day permafrost zones, with late-Holocene shrubification most pronounced where permafrost coverage is presently discontinuous and sporadic. After ∼600 years BP, we find a proliferation of non-Sphagnum mosses in Fennoscandia and across the present-day continuous permafrost zone; and rapid expansion of Sphagnum in regions of discontinuous and isolated permafrost as expected following widespread fen-bog succession, which coincided with declining woody vegetation in eastern and western Canada. Since ∼200 years BP, both shrub expansion and decline were identified at different sites across the pan-Arctic, highlighting the complex ecological responses of circum-Arctic peatlands to post-industrial climate warming and permafrost degradation. Our results suggest that shrubification of circum-Arctic peatlands has primarily occurred alongside surface drying, resulting from Holocene climate shifts, autogenic peat accumulation, and permafrost aggradation. Future shrubification of circum-Arctic peatlands under 21st century climate change will likely be spatially heterogeneous, and be most prevalent where dry microforms persist.

Human-induced climate warming by 2100 is expected to thaw large expanses of northern permafrost peatlands. However, the spatio-temporal dynamics of permafrost peatland thaw remain uncertain due to complex permafrost–climate interactions, the insulating properties of peat soils and variation in model projections of future climate. Here we show that permafrost peatlands in Europe and Western Siberia will soon surpass a climatic tipping point under scenarios of moderate-to-high warming (Shared Socioeconomic Pathway (SSP) 2-4.5, SSP3-7.0 and SSP5-8.5). The total peatland area affected under these scenarios contains 37.0–39.5 Gt carbon (equivalent to twice the amount of carbon stored in European forests). Our bioclimatic models indicate that all of Fennoscandia will become climatically unsuitable for peatland permafrost by 2040. Strong action to reduce emissions (SSP1-2.6) by the 2090s could retain suitable climates for permafrost peatlands storing 13.9 Gt carbon in northernmost Western Siberia, indicating that socio-economic policies will determine the rate and extent of permafrost peatland thaw.

Palsas and peat plateaus are climatically sensitive landforms in permafrost peatlands. Climate envelope models have previously related palsa/peat plateau distributions in Europe to modern climate, but similar bioclimatic modelling has not been attempted for North America. Recent climate change has rendered many palsas/peat plateaus in this region, and their valuable carbon stores, vulnerable. We fitted a binary logistic regression model to predict palsa/peat plateau presence for North America by relating the distribution of 352 extant landforms to gridded modern climate data. Our model accurately classified 85.3% of grid cells that contain observed palsas/peat plateaus and 77.1% of grid cells without observed palsas/peat plateaus. The model indicates that modern North American palsas/peat plateaus are supported by cold, dry climates with large seasonal temperature ranges and mild growing seasons. We used palaeoclimate simulations from a general circulation model to simulate Holocene distributions of palsas/peat plateaus at 500-year intervals. We constrained these outputs with timings of peat initiation, deglaciation, and postglacial drainage across the continent. Our palaeoclimate simulations indicate that this climate envelope remained stationary in western North America throughout the Holocene, but further east it migrated northwards during 11.5–6.0 ka BP. However, palsa extents in eastern North America were restricted from following this moving climate envelope by late deglaciation, drainage and peat initiation. We validated our Holocene simulations against available palaeoecological records and whilst they agree that permafrost peatlands aggraded earliest in western North America, our simulations contest previous suggestions that late permafrost aggradation in central Canada was climatically-driven.

<p>Human-induced climate change during the 21<sup>st</sup> century is expected to thaw large expanses of permafrost peatlands - one of Earth’s largest terrestrial carbon stores. Whilst frozen, peatland carbon fluxes are inhibited by cold temperatures, but emissions of carbon dioxide (CO<sub>2</sub>) and methane (CH<sub>4</sub>) are expected to substantially increase post-thaw. Peatland permafrost is often characterised by the presence of frost mounds, termed palsas/peat plateaus, or by ice-wedge polygons in more northerly regions. The spatio-temporal dynamics of future permafrost peatland thaw remain highly uncertain due to incomplete mapping of their modern distribution, the insulating properties of organic soils, and the variation in model projections of future climate.</p><p>Here, we present simulations of the modern and future climate envelopes of permafrost peatlands in Europe and Western Siberia. We collated > 2,000<strong> </strong>site observations from across the northern hemisphere to quantify the modern distributions of palsas/peat plateaus and polygon mires. We fitted novel climate envelope models by relating landform distributions to modern climate data. We forced our climate envelope models with decadal projections of future climate under four Shared Socioeconomic Pathway (SSP) scenarios from 2020–2090, taken from an ensemble of 12 general circulation models included in the Coupled Model Intercomparison Project 6 (CMIP6). We then combined our simulations with recent soil organic carbon maps to estimate the total peat carbon stocks that may be at risk from future losses of suitable climate space.</p><p>Our simulations indicate that permafrost peatlands in Europe and Western Siberia will soon surpass a climatic tipping point under scenarios of moderate-to-high warming (SSP2-4.5, SSP3-7.0, and SSP5-8.5). We show that permafrost peatlands in Fennoscandia currently exist under warmer, wetter climates than those in Western Siberia. Our projections suggest that Fennoscandia will no longer be climatically suitable for peatland permafrost by 2040. Projected climate space losses by 2100 under these scenarios would affect peatlands containing 37.0–39.5 Gt carbon in Europe and Western Siberia (equivalent to twice the amount of carbon stored in European forests). Under a scenario with strong climate change mitigation (SSP1-2.6), our analyses show that permafrost peatlands storing 13.9 Gt carbon in the northernmost parts of Western Siberia would remain climatically supported by the 2090s. These results indicate that the rate and extent of 21st century permafrost peatland thaw will be determined by near-future socioeconomic developments.</p>

<p>Northern permafrost peatlands represent one of Earth’s largest terrestrial carbon stores and are highly sensitive to climate change. Whilst frozen, peatland carbon fluxes are restricted by cold temperatures, but once permafrost thaws and saturated surficial conditions develop, emissions of carbon dioxide (CO<sub>2</sub>) and methane (CH<sub>4</sub>) substantially increase. This positive feedback mechanism threatens to accelerate future climate change globally. Whilst future permafrost distributions in mineral soils have been modelled extensively, the insulating properties of organic soils mean that peatland permafrost responses are highly uncertain. Peatland permafrost is commonly evidenced by frost mounds, termed palsas/peat plateaus, or by polygonal patterning in more northerly regions. Although the distribution of palsas in northern Fennoscandia is well-studied, the extent of palsas/peat plateaus and polygon mires elsewhere remains poorly constrained, which currently restricts predictions of their future persistence under climate change.  </p><p>Here, we present the first pan-Arctic analyses of the modern climate envelopes and future distributions of permafrost peatland landforms in North America, Fennoscandia, and Western Siberia. We relate a novel hemispheric-scale catalogue of palsas/peat plateaus and polygon mires (>2,100<strong> </strong>individual sites) to modern climate data using one-vs-all (OVA) binary logistic regression. We predict future distributions of permafrost peatland landforms across the northern hemisphere under four Shared Socioeconomic Pathway (SSP) scenarios, using future climate projections from an ensemble of 12 general circulation models included in the Coupled Model Intercomparison Project 6 (CMIP6). We then combine our simulations with recent soil organic carbon maps to estimate how northern peatland carbon stocks may be affected by future permafrost redistribution. These novel analyses will improve our understanding of future peatland trajectories across the northern hemisphere and assist predictions of climate feedbacks resulting from peatland permafrost thaw. </p>

Vegetation shifts in circum-Arctic permafrost peatlands drive feedbacks with important consequences for peatland carbon budgets and the extent of permafrost thaw under changing climate. Recent shrub expansion across Arctic tundra environments has led to an increase in above-ground biomass, but the long-term spatiotemporal dynamics of shrub and tree growth in circum-Arctic peatlands remain unquantified. We investigate changes in peatland vegetation composition during the Holocene using previously-published plant macrofossil records from 76 sites across the circum-Arctic permafrost zone. In particular, we assess evidence for peatland shrubification at the continental scale. We identify increasing abundance of woody vegetation in circum-Arctic peatlands from ∼8000 years BP to present, coinciding with declining herbaceous vegetation and widespread Sphagnum expansion. Ecosystem shifts varied between regions and present-day permafrost zones, with late-Holocene shrubification most pronounced where permafrost coverage is presently discontinuous and sporadic. After ∼600 years BP, we find a proliferation of non-Sphagnum mosses in Fennoscandia and across the present-day continuous permafrost zone; and rapid expansion of Sphagnum in regions of discontinuous and isolated permafrost as expected following widespread fen-bog succession, which coincided with declining woody vegetation in eastern and western Canada. Since ∼200 years BP, both shrub expansion and decline were identified at different sites across the pan-Arctic, highlighting the complex ecological responses of circum-Arctic peatlands to post-industrial climate warming and permafrost degradation. Our results suggest that shrubification of circum-Arctic peatlands has primarily occurred alongside surface drying, resulting from Holocene climate shifts, autogenic peat accumulation, and permafrost aggradation. Future shrubification of circum-Arctic peatlands under 21st century climate change will likely be spatially heterogeneous, and be most prevalent where dry microforms persist.

Human-induced climate warming by 2100 is expected to thaw large expanses of northern permafrost peatlands. However, the spatio-temporal dynamics of permafrost peatland thaw remain uncertain due to complex permafrost–climate interactions, the insulating properties of peat soils and variation in model projections of future climate. Here we show that permafrost peatlands in Europe and Western Siberia will soon surpass a climatic tipping point under scenarios of moderate-to-high warming (Shared Socioeconomic Pathway (SSP) 2-4.5, SSP3-7.0 and SSP5-8.5). The total peatland area affected under these scenarios contains 37.0–39.5 Gt carbon (equivalent to twice the amount of carbon stored in European forests). Our bioclimatic models indicate that all of Fennoscandia will become climatically unsuitable for peatland permafrost by 2040. Strong action to reduce emissions (SSP1-2.6) by the 2090s could retain suitable climates for permafrost peatlands storing 13.9 Gt carbon in northernmost Western Siberia, indicating that socio-economic policies will determine the rate and extent of permafrost peatland thaw.

In northern peatlands, near-saturated surface conditions promote valuable ecosystem services such as carbon storage and drinking water provision. Peat saturated hydraulic conductivity (Ksat) plays an important role in maintaining wet surface conditions by moderating drainage and evapotranspiration. Peat Ksat can exhibit intense spatial variability in three dimensions and can change rapidly in response to disturbance. The development of skillful predictive equations for peat Ksat and other hydraulic properties, akin to mineral soil pedotransfer functions, remains a subject of ongoing research. We report a meta-analysis of 2,507 northern peat samples, from which we developed linear models that predict peat Ksat from other variables, including depth, dry bulk density, von Post score (degree of humification), and categorical information such as surface microform type and peatland trophic type (e.g. bog and fen). Peat Ksat decreases strongly with increasing depth, dry bulk density, and humification; and increases along the trophic gradient from bog to fen peat. Dry bulk density and humification are particularly important predictors and increase model skill greatly; our best model, which includes these variables, has a cross-validated r2 of 0.75 and little bias. A second model that includes humification but omits dry bulk density, intended for rapid field estimations of Ksat, also performs well (cross-validated r2 = 0.64). Two additional models that omit several predictors perform less well (cross-validated r2 ∼ 0.5), and exhibit greater bias, but allow Ksat to be estimated from less comprehensive data. Our models allow improved estimation of peat Ksat from simpler, cheaper measurements.

Palsas and peat plateaus are climatically sensitive landforms in permafrost peatlands. Climate envelope models have previously related palsa/peat plateau distributions in Europe to modern climate, but similar bioclimatic modelling has not been attempted for North America. Recent climate change has rendered many palsas/peat plateaus in this region, and their valuable carbon stores, vulnerable. We fitted a binary logistic regression model to predict palsa/peat plateau presence for North America by relating the distribution of 352 extant landforms to gridded modern climate data. Our model accurately classified 85.3% of grid cells that contain observed palsas/peat plateaus and 77.1% of grid cells without observed palsas/peat plateaus. The model indicates that modern North American palsas/peat plateaus are supported by cold, dry climates with large seasonal temperature ranges and mild growing seasons. We used palaeoclimate simulations from a general circulation model to simulate Holocene distributions of palsas/peat plateaus at 500-year intervals. We constrained these outputs with timings of peat initiation, deglaciation, and postglacial drainage across the continent. Our palaeoclimate simulations indicate that this climate envelope remained stationary in western North America throughout the Holocene, but further east it migrated northwards during 11.5–6.0 ka BP. However, palsa extents in eastern North America were restricted from following this moving climate envelope by late deglaciation, drainage and peat initiation. We validated our Holocene simulations against available palaeoecological records and whilst they agree that permafrost peatlands aggraded earliest in western North America, our simulations contest previous suggestions that late permafrost aggradation in central Canada was climatically-driven.