Content uploaded by David Olefeldt
Author content
All content in this area was uploaded by David Olefeldt on Apr 08, 2015
Content may be subject to copyright.
REVIEW doi:10.1038/nature14338
Climate change and the permafrost
carbon feedback
E. A. G. Schuur
1,2
, A. D. McGuire
3
, C. Scha
¨del
1,2
, G. Grosse
4
, J. W. Harden
5
, D. J. Hayes
6
, G. Hugelius
7
, C. D. Koven
8
, P. Kuhry
7
,
D. M. Lawrence
9
, S. M. Natali
10
, D. Olefeldt
11,12
, V. E. Romanovsky
13,14
, K. Schaefer
15
, M. R. Turetsky
11
, C. C. Treat
16
& J. E. Vonk
17
Large quantitiesof organic carbon arestored in frozen soils (permafrost) within Arctic and sub-Arctic regions. A warming
climate can induce environmental changes that accelerate the microbial breakdownof organic carbon and the release of
the greenhouse gases carbon dioxide and methane. This feedback can accelerate climate change, but the magnitude and
timing of greenhouse gas emission from these regions and their impact on climate change remain uncertain. Here we find
that current evidence suggests a gradual and prolonged release of greenhouse gas emissions in a warming climate and
present a research strategy with which to target poorly understood aspects of permafrost carbon dynamics.
In high-latitude regions of Earth, temperatures have risen 0.6 uC per
decade over the last 30 years, twice as fast as the global average
1
.This
is causing normally frozen ground to thaw
2–4
, exposing substantial
quantities of organic carbon to decomposition by soil microbes. This
permafrost carbon is the remnant of plants and animals accumulated in
perennially frozen soil over thousands of years, and the permafrost region
contains twice as much carbon as there is currently in the atmosphere
5,6
.
Conversion of just a fraction of this frozen carbon pool into the green-
house gases carbon dioxide (CO
2
) and methane (CH
4
) and their release
into the atmosphere could increase the rate of future climate change
7
.
Climate warming as a result of human activities causes northern regions
to emit additional greenhouse gases to the atmosphere, representing a
feedback that will probably make climate change happen faster than is
currently projected by Earth System models. The critical question centres
on how fast this process will occur, andrecent publications differ in their
outlook on this issue. Abrupt releases of CH
4
forecast to cause trillions of
dollars of economic damageto global society
8
contrast with predictionsof
slower, sustained greenhouse gas release that, although substantial,would
give society more time to adapt
1,9
. This range of viewpoints is due in part
to the wide uncertainty surrounding processes that are only now being
quantified in these remote regions.
Here we provide an overview of newinsights froma multi-yearsynthesis
of data with the aim of constraining our current understanding of the
permafrost carbon feedback to climate, and providing a framework for
developing research initiatives in the permafrost region
10,11
. We begin by
reviewingnew research, much of it published since the Intergovernmental
Panel on Climate Change (IPCC)’s Fifth Assessment Report (AR5)
1
,on
the size of the carbon pool stored in the permafrost region. Synthesis research
has enlarged the number of observations in the permafrost region soil carbon
pool database tenfold
12
, and confirms that tremendous quantities of carbon
accumulated deep in permafrost soils are widespread
5,6
. We then discuss
new long-term laboratoryincubations of these permafrost soils that reveal
that a substantial fraction of this material can be mineralized by microbes
and converted to CO
2
and CH
4
on timescales of years to decades, which
would contribute to near-term climatewarming. Initialestimates of green-
house gas release point towards the potential for substantial emissions of
carbon from permafrostin a warmer world, but these could still be under-
estimates. Field observations reveal that abrupt thawprocessesare common
in northern landscapes,but our review shows that mechanisms that speed
thawing of frozen ground and release of permafrost carbon are entirely
absent from the large-scale models used to predict the rate of climate change.
Bringing together this wealth of new observations, wepropose that green-
house gas emissions from warmingpermafrost are likely to occur at a mag-
nitude similar to other historically important biospheric carbon sources
(such as land-use change) but that will be only a fraction of current fossil-fuel
emissions. At the proposed rates,the observed and projected emissions of
CH
4
and CO
2
from thawing permafrost are unlikely to cause abrupt climate
change overa period of a few years to a decade. Instead,permafrost carbon
emissions are likely to be felt over decades to centuries as northern regions
warm, making climate change happen faster than we would expect on the
basis of projected emissions from human activities alone. This improved
knowledge of the magnitude and timing of permafrost carbon emissions
based on the synthesis of existing data needs to be integrated into policy
decisions about the management of carbon in a warming world, but at the
same time may help temper the worst fears about the impact of carbon
emissions from warming northern high-latitude regions.
Permafrost carbon pool
The first studies that brought widespread attention to permafrost carbon
estimated that almost 1,700 billion tons of organic carbon were stored in
terrestrial soils in the northern permafrost zone
6,7,13
. The recognition of
this vast pool stored in Arctic and sub-Arctic regions was in part due to sub-
stantial carbon stored at depth (.1 m) in permafrost, below thetraditional
zone of soil carbon accounting
14
. Deeper carbon measurements were initially
rare, and it was not even possible to quantify the uncertainty for the permafrost
carbon pool size estimate.However, important new syntheses continue to
report large quantities of deep carbon preserved in permafrost at many
previouslyunsampled locations, and thata substantialfraction of this deep
1
Center for Ecosystem Science and Society and Department of Biological Sciences, Northern Arizona University, Flagstaff, Arizona 86011, USA.
2
Department of Biology, University of Florida, Gainesville,
Florida 32611, USA.
3
US Geological Survey, Alaska Cooperative Fish and Wildlife Research Unit, University of Alaska Fairbanks, Alaska 99775, USA.
4
Alfred Wegener Institute Helmholtz Centre for Polar and
Marine Research, 14473 Potsdam, Germany.
5
US Geological Survey, Menlo Park, California 94025, USA.
6
Climate Change Science Institute and Environmental Sciences Division, Oak Ridge National
Laboratory, Oak Ridge, Tennessee 37831, USA.
7
Department of Physical Geography, Stockholm University, 10691 Stockholm, Sweden.
8
Earth Sciences Division, Lawrence Berkeley National Laboratory,
Berkeley, California 94720, USA.
9
National Center for Atmospheric Research, Boulder, Colorado 80305, USA.
10
Woods Hole Research Center, Falmouth, Massachusetts 02540, USA.
11
Department of
Integrative Biology, University of Guelph, Guelph, Ontario N1G 2W1, Canada.
12
Department of Renewable Resources, University of Alberta, Edmonton, Alberta T6G 2H1, Canada.
13
Geophysical Institute,
University of Alaska Fairbanks, Fairbanks, Alaska 99775, USA.
14
Tyumen State Oil and Gas University, Tyumen, Tyumen Oblast 625000, Russia.
15
National Snow and Ice Data Center, Boulder, Colorado
80309, USA.
16
Earth Systems Research Center, Institute for the Study of Earth, Oceans and Space, University of New Hampshire, Durham, New Hampshire 03824, USA.
17
Department of Earth Sciences,
Utrecht University, 3584 CD Utrecht, The Netherlands.
G2015 Macmillan Publishers Limited. All rights reserved
9 APRIL 2015 | VOL 520 | NATURE | 171
permafrost carbon is susceptible to future thaw
15
. The permafrost carbon
pool is now thought to comprise organic carbon in the top 3 m of surface
soil,carbonindepositsdeeperthan3m(includingthosewithintheyedoma
region, an area of deep sediment deposits that cover unglaciated parts of
Siberia and Alaska
16–18
), as well as carbon within permafrost that formed
on land duringglacial periods butthat is now found on shallow submarine
shelves in the Arctic. Recent researchhas expanded our knowledge consi-
derably while at the same time highlighting remaining gaps in our under-
standing of this vulnerable carbon pool
19
.
Surface carbon
The new northern permafrost zone carbon inventory reports the surface
permafrost carbon pool (0–3m) to be 1,03561 50 Pg carbon (mean 6
95% confidence interval, CI)
12,20
(where 1 Pg 51 billion tons) (Fig. 1a).
This estimate supported the original studies while improving precision
by increasing the number of deeper (.1 m) sampling locations tenfold.
This surface permafrost carbon pool is substantial. The rest of Earth’s
biomes, excluding the Arctic and boreal regions, are thought to contain
2,050 Pg carbon in the surface 3 m of soil
21
. Even though these northern
regions account for only 15% of global soil area, the 0–3m global soil
carbon pool is increased by 50% when fully accounting for the carbon
stored deeper in permafrost zone soil profiles.
Deep carbon in yedoma
Processes that accumulate carbon deep into permafrost soils do not stop
at 3 m depth, and our previously limited understanding of those deep carbon
deposits (.3 m depth) has been improved. In particular, several new esti-
mates have emerged for carbon that accumulated during, and since, the
last IceAge in the yedoma region in Siberia and Alaska
16–18
. These new data
support previous findings of relatively high carbon concentrations in per-
mafrost soil at depth, but revised the understanding of total carbon stock
by improving the estimates of spatial extent, type of deposit, sediment depth,
and ground ice content. These deep, perennially frozen sediments are par-
ticularly ice-rich, where ice occupies 50%–80% of the ground volume
22,23
.
Although this excess ice does not alter soil carbonconcentration, it affects
the total carbon inventory contained in a particular volume of soil, decreasing
carbon stocks per unit soil volume by 22%–50% compared to previous
estimates
24
. Because of the continueddifficulty of measuring total ground
ice content and total sediment depth, carbon pool estimatesfor the yedoma
region still range by twofold even as new data from this region have accu-
mulated. This region is now thought to contain between 210 670 Pg
carbon (ref. 16) and 456 645 Pg carbon (ref. 18), still supporting the
original accounts of several hundred billion tons of carbon stored deep
in the permafrost even when recalculated with new observations.
Deep carbon outside the yedoma region
While new measurements of deep carbon have been largely focused on
the 1.2 million square kilometres of the yedoma region in recent years,
other areas in the northern permafrost zone with thick loose sedimentary
material may also contain substantial organic carbon pools in permafrost
(Fig. 1b). The major Arctic river deltas are now thought to contain
91 639 Pgcarbon (95% CI)
12
, while carbon contained in the approxi-
mately 5 million square kilometres of thick (.5–10 m) sediments over-
lying bedrock outside the yedoma and river delta regions remain largely
unknown. Taking the spatial extent of these poorly known permafrost
areas, along with an estimated thickness in the tens of metres (similar
to that of yedoma), and average carbon content of a few deep borehole
soil samples, there could be an additional deep permafrost carbon pool of
350–465 PgC outsidethe yedoma region (calculated using a depth interval
of 3–10 m and carbon content of 11–14 kg C m
23
, which accounts for
ground ice
25
).
Subsea permafrost carbon
Much ofthe inventoryuntil this pointhas focused onterrestrialecosystems
where permafrost is currently sustained by cold winter air temperatures.
But permafrost also exists below Arctic Ocean continental shelves, in
particular the East Siberian Arctic Shelf, the largest and shallowest shelf
on Earth. This permafrost is an extension of the terrestrial permafrost
that existed during the last Ice Age, but became submerged when sea level
rose during thelate Pleistocene–Holocenetransition, and at the beginning
of the Holocene epoch. The shallow shelf area exposed as dry land in the
area around Alaska and Siberia during the last Ice Age (,125 m current
ocean depth), atalmost 3 million square kilometres, is about 2.5 times the
size of the current terrestrial yedomaregion
16,26
. But thequantity of organic
permafrost carbon stored beneaththe sea floor is even more poorly quan-
tifi ed than on land and could be lower than it once was
27,28
.Subseapermafrost
as a whole has been slowly degradingover thousands of years as relatively
warm ocean water has warmed the newly submerged sea floor. Frozen
sediments are thickest near the shore, where submergence with seawater
occurred more recentlythan on the outer shelf, which is now underlain by
discontinuous, patchy permafrost
29,30
. During this time of thaw, organic
carbon was mineralized by microbes within the sediment in low-oxygen
conditions that promote the formation of CH
4
, reducing the pool of
permafrost carbon remaining under the sea.
Taken together, the known pool of terrestrial permafrost carbon in the
northern permafrost zone is 1,330–1,580 Pgcarbon, accounting for sur-
face carbon as well as deep carbon in the yedoma region and river deltas,
with the potential for ,400 Pg carbon in other deep terrestrial permafrost
sediments that, along with an additional quantity of subsea permafrost
carbon, still remains largely unquantified.
Carbon decomposability
Permafrost carbon stocks provide the basis for greenhouse gas release to
the atmosphere, but the rate at which this can happen is also controlled
by the overall decomposability of organic carbon. Conceptual modelsand
initial data on decomposability suggested that a portion of permafrost
carbon is susceptible to rapid breakdown upon thaw
13,31
. But it has not
been clear to whatdegree this could be sustained on the decade-to-century
timescale of climate change, or what degree of variation exists within soils
across the vast landscape of the permafrost zone. New research has confirmed
that initial rates of permafrost carbon loss are potentially high, but continued
observation reported declines in carbon loss rates over time, which might
be expected as more labile carbon pools are exhausted
32
. This has high-
lighted the need for long-term observation under controlled conditions
to estimate the potential decomposability of permafrost carbon.New data
from a 12-year incubation of permafrost soil from Greenland showed
that 50%–75% of the initial carbon was lost by microbial decomposition
under aerobic and continuously unfrozen laboratory conditions over that
time frame
33
. This experiment, of unprecedented length for permafrost
soils compared to typical incubations that might be only weeks to months
long
34,35
, was then extended geographically in a newsynthesis of long-term
(.1 year) permafrost zone soil incubations. Soils from across the permafrost
region showed similarly high potential for microbial degradation of organic
carbon upon thaw in the laboratory, with a wider range of decade-long
losses projected to be 1%–76% (Fig. 2a) under laboratory conditions
36
.
A major cause of landscape-scale variation in decomposability across
soils was linked to the carbon to nitrogenratio of the organic matter, with
higher values leading to more greenhouse gas release. This simple metric
(the carbon to nitrogen ratio) is in part illustrated by grouping soils as organic
(.20% C) with mean decade-long losses of 17%–34% (lower-to-upper
97.5% CI) and mineral (,20% C) with mean decade-long losses of 6%–13%
(Fig. 2a). The metric takes into account the ability of microbes to process
permafrost carbon for metabolism by breaking down organic carbon for
energy, and to grow by acquiring nutrients such as nitrogen released
during the decompositionprocess. Because carbon and nitrogen are often
measured in soil surveys, maps of permafrost carbon pools can then be
combined with the findings from laboratory incubations to project potential
carbon emission estimates across the permafrost region to determine which
regions could be emission hotspots in a warming climate. The location of
such potential emission hotspots is expectedto be affected by both the total
pool of permafrost carbon and the potential for that carbon to be broken
RESEARCH REVIEW
G2015 Macmillan Publishers Limited. All rights reserved
172 | NATURE | VOL 520 | 9 APRIL 2015
down by microbes after thaw as controlled by the energy and nutrients
contained within the organic matter.
The inherent range of permafrost carbon decomposability across soil
types also intersects with environmental conditions, and aerobic decomposi-
tion is only part of the story for northern ecosystems. While temperature
controlover decompositionis implicit when considering permafrost thaw,
this region is characterizedby widespread lakes, wetlands, and soilswater-
logged as a result of surface drainage restricted by underlying permafrost.
The lack of oxygen in saturated anaerobic soils and sediments presents
anotherkey control overemissions from newly thawedpermafrost carbon.
Comparing the results from the aerobic permafrost soil incubation synthesis
36
with those from another circumpolar synthesis of anaerobic soil incubations
37
0.1–30 kg m–2
30–50 kg m–2
50–100 kg m–2
100–260 kg m–2
Soil organic carbon
storage (0–3 m):
#
*
Major river deltas
Yedoma largely unaffected by thaw cycles
Region of potential yedoma distribution
Thick sediments
Continuous permafrost
Discontinuous permafrost
a
b
Figure 1
|
Soil organic carbon maps. a, Soil organic carbon pool (kgC m
22
)
contained in the 0–3m depth interval of the northern circumpolar permafrost
zone
12
. Points show field site locations for 0–3 m depth carbon inventory
measurements; field sites with 1 m carbon inventory measurements numberin
the thousands and are too numerous to show. b, Deep permafrost carbon pools
(.3 m), including the location of major permafrost-affected river deltas (green
triangles), the extent of the yedoma region previously used to estimate the
carbon content of these deposits
13
(yellow), the current extent of yedoma region
soils largely unaffected by thaw-lake cycles that alter the original carbon
content
17
(red), and the extent of thick sediments overlying bedrock (black
hashed). Yedoma regionsare generally also thick sediments. The base map layer
shows permafrost distribution with continuous regions to the north having
permafrost everywhere (.90%), and discontinuous regions further south
having permafrost in some, but not all, locations (,90%)
96
.
REVIEW RESEARCH
G2015 Macmillan Publishers Limited. All rights reserved
9 APRIL 2015 | VOL 520 | NATURE | 173
shows that cumulative carbon emissions, over an equal one-year incubation
time frame, are, on average, 78%–85% lowerthanthosefromaerobicsoils
(Fig. 2b). Specialized microbes release CH
4
along with CO
2
in these envi-
ronments, and the more potent (that is, it affects climate change more power-
fully) greenhouse gas CH
4
in the atmosphere can partially offset a
decreased decomposition rate. While mean quantities of CH
4
are 3% (in
mineral soils) to 7% (in organic soils) that of CO
2
emitted from anaerobic
incubations (by weight of carbon), these mean CH
4
values represent 25%
(in mineral soil) to 45% (in organic soil) of the overallpotential impact on
climate over a 100-year timescale when accounting for CH
4
(ref. 38).
Across the mosaic of ecosystems in the permafrost region, controlled
laboratory observations brought together here imply that, in spite of
the more potent greenhouse gas CH
4
, a unit of newly thawed permafrost
carbon could have a greater impact on climate over a century if it thaws
and decomposes within a drier, aerobic soil as compared to an equivalent
amount of carbon within a waterlogged soil or sediment.
Controlled laboratory work is critical for identifying the key mechanisms
for potential greenhouse gas release from permafrost carbon, but some
important processes aredifficult to address withincubation experiments.
For example, CH
4
generated from permafrost carbon can be oxidized in
aerobic soil layers above the water tableand released to the atmosphere as
CO
2
instead. This effect can be modifiedby vegetation, for example, sedge
stems acting as pipes providea pathway for CH
4
to avoid oxidation and to
escape tothe atmosphere
39
. A synthesisof field CH
4
emissionrates showed
that sedge-dominated sites had emission rates 2–5 times higher
40
, due in
part to sedges allowing the physical escape of CH
4
, as well as providing
more decomposable carbon to the microbial community
41,42
. But even
with sedges, it is likely that CH
4
oxidation as a whole would decrease the
warming impact of permafrost carbon decomposing in a waterlogged envi-
ronment compared to what was measured from a laboratory potential.
Incubation results, while needing to be interpreted carefully, are useful for
scaling the potential of permafrost soils to release greenhouse gases upon
thaw, and also for helping to quantify the fraction of soil carbon that is
likely to remain relatively inert within the soil after thaw.
Projecting change
A number of ecosystem and Earth systemmodels have incorporateda first
approximation of global permafrost carbon dynamics. Recent key improve-
ments include the physical representation of permafrost soil thermody-
namics and the roleof environmentalcontrols, in particular the soil freeze/
thaw state, on decomposition of organic carbon
43–45
. These improved models,
which specifically address processes known to be important in permafrost
ecosystemsbut that were missingfrom earlier modelrepresentations,have
been key for forecasting the potential release of permafrost carbon with
warming, and the impact this would have on the rate of climate change.
Model scenarios show potential carbon release from the permafrost zone in
the range 37–174 Pg carbon by 2100 under the current climate warming
trajectory (Representative Concentration Pathway RCP8.5), with an
average across models of 92617 Pg carbon (mean 6s.e.) (Fig. 3)
45–52
.
Furthermore, thawing permafrost carbon is forecasted to impact global
climate for centuries, with models, on average, estimating that 59% of
total permafrost carbon emissions will occur after 2100. While carbon
releases over these time frames are understandably uncertain, they illus-
trate the momentum of a warming climate that thaws near-surface per-
mafrost, causing a cascading release of greenhouse gases as microbes
slowly decompose newly thawed permafrost carbon. At the scale of these
models not alldifferentiated between CO
2
and CH
4
loss, but expert assess-
ment, a method forsurveying expert knowledge, placed CH
4
losses at about
2.3% of total future emissions from the permafrost zone
53,54
. This has the
effect, in the expert assessment, of increasing the warming potential of
released carbon by 35%–48% when accounting for the more potent
greenhouse gas CH
4
over a 100-year timescale.
Within the wideuncertainty of forecasts, some broader patterns are just
beginning to emerge. Models vary widely when predicting the current
pool of permafrost carbon, which is the source of future carbon emissions
in a warmer world. The model average permafrost carbon pool size was
estimated at 77161 00 Pg carbon (mean 6s.e.), about half as much as the
measurement-based estimate, potentially related in part to the fact that
models mostly represented carbon to only 3 m depth. A smaller modelled
carbon pool could, in principle, constrain forecasted carbon emissions.
Normalizing the emissions estimates from the dynamic models by their
initial permafrost carbon pool size, 15% 63% (mean 6s.e.) of the initial
pool was expected to be lost as greenhouse gas emissions by 2100
55
. This
decrease in the permafrost carbon pool is similar, but somewhat higher,
than the 7%–11% (95% CI) loss predictedby experts
53,54
, and the relatively
constant fraction across model estimates does hint at the importance of pool
size in constraining carbon emissions. However, sensitivity to both modelled
Arcticclimate change, as wellas the responses of soiltemperature,moisture
and carbon dynamics, are important controls over emissions predictions
within these complex models, not pool size alone
44,56,57
. Full diagnosis of the
important parameters that regulate the permafrost carbon feedback is not
currently possible from the small number of modelling studies that exist,
but the estimates do seem to converge on a vulnerable fraction of permafrost
carbon that seems to be in line with other approaches.
Aerobic Anaerobic Aerobic Anaerobic
Cumulative C release (percentage of total C)
after 1 year of incubation
0
2
4
6
8
Or
g
anic soils Mineral soils
b
CO2-C
CO2-C
CH4-C in CO2-C equivalent
CH4-C in CO2-C equivalent
Incubation time (yr)
0246810
Cumulative C release (percentage of total C)
0
20
40
60
80
Mineral soils (<20% C)
Organic soils (>20% C)
a
Figure 2
|
Potentialcumulative carbon release. Data are given as a percentage
of initial carbon. a, Cumulative carbon release after ten years of aerobic
incubation at a constant temperature of 5 uC. Thick solid lines are averages for
organic (red, N543) and mineral soils (blue, N578) and thin solid lines
represent individual soils to show the response of individual soils. Dotted lines
are the averages of the 97.5% CI for each soil type. b, Cumulative carbon release
after one year of aerobic and anaerobic incubations (at 5 uC). Darker colours
represent cumulative CH
4
-carbon calculated as CO
2
-carbon equivalent (for
anaerobic soils) on a 100-year timescale according to ref. 38. Positive error bars
are upper 97.5% CI for CO
2
-carbon and negative error bars are lower 97.5% CI
for CH
4
-carbon. N528 for organic soils and N525 for mineral soils in
anaerobic incubations. Aerobic cumulative carbon release is redrawn from ref.
36 and anaerobic cumulative carbon release is calculated based on ref. 37.
RESEARCH REVIEW
G2015 Macmillan Publishers Limited. All rights reserved
174 | NATURE | VOL 520 | 9 APRIL 2015
These dynamic models also simultaneously assess the countering influence
of plant carbon uptake,which may in part offsetpermafrost carbonrelease.
Warmer temperatures, longer growing seasons,elevated CO
2
, and increased
nutrients released from decomposing organic carbon may all stimulate plant
growth
58
. New carbon can be stored in larger plant biomass or deposited
into surface soils
59
. A previous generation of Earthsystem models that did
not include permafrost carbon mechanisms but did simulate changes in
plant carbon uptake estimated that the vegetation carbon pool could increase
by 17 68 Pg carbon by 2100, with increased plant growth also contributing
to new soil carbon accumulation of similar magnitude
60
.Themodels
reviewed here that do include permafrost carbon mechanisms (as well
as many of the mechanisms that stimulate plantgrowth that were used in
the previous generation of models) generally indicate that increased plant
carbon uptake will more than offset soil carbon emissions from the per-
mafrost region for several decadesas climate becomes warmer
45,46,48
. Over
longer timescales and with continued warming, however, microbial
release of carbon overwhelms the capacity for plant carbon uptake, lead-
ing to net carbon emissions from permafrost ecosystems to the atmo-
sphere. Modelled carbon emissions projected under various warming
scenarios translate into a range of 0.13–0.27 uCadditionalglobalwarming
by 2100 and up to 0.42 uC by 2300, but currently remain one of the least
constrained biospheric feedbacks to climate
1
.
Abrupt permafrost thaw
Recentprogress towards predictingchange in permafrost carbon dynamics
focuses mostly on gradual top-down thawing of permafrost. However, increas-
ing evidence from the permafrost zone suggests that abrupt permafrost thaw
may be the norm for many parts of the Arctic landscape
17,18,61,62
(Fig. 4).
Abrupt permafrost thaw occurs when warming melts ground ice, causing
the land surface to collapse into the volume previously occupied by ice.
This process, called thermokarst, alters surface hydrology. Water is attracted
towards collapse areas, and pooling or flowing water in turn causes more
localized thawing and even mass erosion. Owing to these localized feedbacks
that canthaw through tens of metresof permafrostacross a hillslope within
only a few years, permafrost thaw occursmuch more rapidlythan would be
predicted from changes in air temperature alone. This raises the question
of whether key complexityis missing from large-scale modelprojections
that are based on first approximations of permafrost dynamics.
Abrupt thaw occurs only at point locations but often causes much deeper
permafrost thaw to occur more rapidly. This is in contrast to top-down
thawing, which occurs across the entire landscape but affects only the perma-
frost surface. New regional research is beginning to reveal that a large fraction
of permafrost carbon is vulnerable toabrupt thaw. For example, since the
end of the last Ice Age, thermokarst thaw-lake cycles have affected 70% of
the yedoma permafrost deposits in Siberian lowlands
17
. These cycles occur
when abrupt permafrost thaw forms lakes that can drain over time, allowing
sediments and carbon to refreeze into permafrost, while elsewhere new
thaw lakes form and repeat this cyclic process (Fig. 4a, c). Abrupt thaw in
upland regions, where waterdoes not generally pool and form lakes, often
creates gullies and slump features that can erode permafrost carbon into
streams, riversand lakes (Fig. 4b, d). These thaw features can also be wide-
spread but are not as well recognized as are thaw lakes; over 7,500 upland
thaw features were mappedwithin a 1,700-square-kilometrefoothillregion
of Alaskan tundra
49
. Studies such as these illustrate a widespread influence
of abrupt thaw in both upland and lowland permafrost landscapes, even
though they do not provide a chronology of change.
Climate change is expected to increase the initiation and expansion of
abrupt thaw features, potentially changing the rate of this historic disturbance
cycle
62–65
. Wetland expansion due to abrupt thaw has affected 10% of a
peatland landscape in northwestern Canada since the 1970s, with the fastest
expansion occurring in the past decade
66
. Landscape lake cover is also affected
by abrupt thaw, with net change being thesum of both lake expansion and
drainage. The area of small open-water features around Prudhoe Bay on
the Alaskan tundra has doubled since 1990 (ref. 67). In northwestern
Alaska, lake initiation has increased since 1950, while lake expansion rates
remained steady
68
. In general, landscape lake cover is currently believed to
be stable or increasing within the continuous permafrost zone, whereas
there is a tendency for lake drainage and vegetation infilling to dominate
over lake expansion in the discontinuous permafrost zone
68–72
.
Abrupt thaw influences carbon emissions to the atmosphere by exposing
previously frozen carbonto microbial processes, and also by altering the
hydrology that is critical for determining the balance of CO
2
and CH
4
emissions. Some of the highest CH
4
emissions in the permafrost region
have been observed in lakes and wetlands formed through abrupt thaw
40,73
.
At the same time, accumulation of new carbon under anaerobic conditions
in peat
74
and in lake sediments
18
can be greater than permafrost carbon
losses, at least in some ecosystems. In this way, anaerobic environments
replace freezing temperatures as a mechanismfor soil carbon stabilization,
keeping greenhouse gas emissions lower than they would otherwise be
75
.
In contrast, abrupt thaw processes in other landscapes clearly accelerate
carbon loss. Drained lakes and loweredwater tables will expose previously
waterlogged carbon to microbial decomposition in aerobic conditions
with relatively higher rates of carbon emissions. Also, lateral movement
of permafrost carbon by leaching or erosion into lakes, rivers and the
ocean
76–78
can increase loss, as carbon may be more readily mineralized
through microbial and photochemical processes after mobilization
79,80
.
How carbon cycling at the landscape scale will change under a warming
climate will depend critically on how much of the landscape becomes wetter
or drier, a question difficult to answer. It is clear that abrupt thaw is an
important mechanism of rapid permafrost degradation, with widespread
but varying influences on hydrology and carbon cycling. Yet abrupt thaw
is not included in large-scale models, suggestingthat important landscape
transformations are not currently being considered in forecasts of permafrost
carbon–climate feedbacks. This is in part due to the fact that we do not know
at this stage what the relativeimportance of abrupt to gradual thaw across
the landscape is likely to be.
Subsea carbon emissions
A majority of the observations and all of the modellingto date has focused
on potential emissions from permafrost carbon on land. This is in part
because subsea permafrost is buffered from recent climate change by the
overlying ocean, and becauseocean incursion at the end of the Ice Age has
already been thawing and potentially reducing the pool of permafrost carbon
under the sea. However, aside from organic carbon stored in permafrost,
the sea bed underlying Arctic shelves also accumulated fossil CH
4
stored
either as free CH
4
gas or as clathrates (CH
4
-ice lattices that are stable at
pressures and temperatures found at depth in this region). Layers of perma-
frost may serve as a physical barrier to the release of this CH
4
gas from the
sediment into the water column and eventually the atmosphere. These
0 100 200 300 400
Ref. 48
Ref. 51
Ref. 46
Ref. 50
Ref. 47
Ref. 45
Ref. 49
Ref. 52
2100
2200
2300
Cumulative emissions (Pg carbon)
Figure 3
|
Model estimates of potential cumulative carbon release from
thawing permafrost by 2100, 2200, and 2300. All estimates except those of
refs 50 and 46 are based on RCP 8.5 or its equivalent in the AR4 (ref. 97), the A2
scenario. Error bars show uncertainties for each estimate that are based on an
ensemble of simulations assuming different warming rates for each scenario
and different amounts of initial frozen carbon in permafrost. The vertical
dashed line shows the mean of all models under the current warming trajectory
by 2100.
REVIEW RESEARCH
G2015 Macmillan Publishers Limited. All rights reserved
9 APRIL 2015 | VOL 520 | NATURE | 175
shallow shelves are also depositional areas for carbon from the erosion of
coastalpermafrost carbonand from inland permafrost carbon transported
by Arctic rivers
81
. Together, these processes form ocean hotspots that are
documented sources of high CH
4
emissions to the atmosphere
82,83
,similar
to hotspots formed in Arctic lakes on land
58
. New quantification has
estimated that 17 Tg of CH
4
peryear(where1Pg51,000 Tg) is emitted
from the East Siberian Arctic Shelf after accounting for both diffusive and
point-source bubble emissions
83
. Although this amount represents an increase
from what was previously estimated for thisr egion
27
,thisisprobablybecause
of improved observations of these emissions that may have been persistent
over the thousands of years of land submergence. Climate warming, sea-ice
decline, and increasingstorminess have been linked to a 2.1 uC increase in
bottom water (,10m depth) temperature since the mid-1980s in this
region
84
. Degradation of subsea permafrost from above by climate warming,
and also frombelow by ongoing geothermal heat,will tend to increase new
pathways between CH
4
storage areas deeper in the sediments and the sea
floor
30
. But it isnot known whethermeaningful increases in CH
4
emissions
via these processes could occur within this century, or whether they are more
likely to manifest over a century or over millennia
84
. What is clear is that it
would take thousands of years of CH
4
emissions at the current rate to
release the same quantity of CH
4
(50 Pg) that was used in a modelled ten-year
pulse to forecast tremendous global economic damage as a result of Arctic
carbon release
8
, making catastrophic impacts such as those appear highly
unlikely
85–87
.
Permafrost and the global carbon cycle
Carbon pools in permafrost regionsrepresent a large reservoir vulnerable
to change in a warmingclimate. While some of thiscarbon will continue to
persist in soils and sedimentsover the long term, our understanding that a
substantial fraction of this pool is susceptibleto microbial breakdown once
thawed has been verified at the landscape scale (Box 1 and the Box 1 Figure).
The exponential nature of microbial decomposition and CO
2
and CH
4
release over time means that the initial decades after thaw will be the most
important for greenhouse gas release from any particular unit of thawed
soil. Our expert judgement is that estimates made by independent approaches,
including laboratory incubations, dynamic models, and expert assessment,
seem to be converging on ,5%–15% of the terrestrial permafrost carbon pool
being vulnerable to release in the form of greenhouse gases during this century
under the current warming trajectory, with CO
2
-carbon comprising the
majority of the release. There is uncertainty, but the vulnerable fraction does
not appear to be twice as high or half as much as 5%–15%, based on this
analysis. Ten per cent of the known terrestrial permafrost carbon pool is
equivalent to ,130–160 Pg carbon. That amount, if released primarily in
the form of CO
2
at a constant rate over a century,would make it similar in
magnitude to otherhistorically importantbiosphericsources, suchas land-
use change (0.9 60.5 Pg carbon per year; 2003–2012 average), but far less
than fossil-fuel emissions
88
(9.7 60.5 Pg carbon per year in 2012).
Considering CH
4
as a fraction of permafrost carbon release would increase
the warming impact of these emissions. At these rates, the observed and
projected emissions of CO
2
and CH
4
from thawing permafrost are
unlikely to occur at a speed that could cause abrupt climate change over
a period of a few years to a decade
1,9
. A large pulse release of permafrost
carbon on this timescale could cause climate change that would incur
catastrophic costs to society
8
, but there is little evidence from either
current observations or model projections to support such a large and
rapid pulse. Instead, permafrost carbon emissions are likely to occur over
1 km
TKL 1950 TKL 2006
1 km
Feature type
ALD GTK RTS
ab
cd
Figure 4
|
Abundance of abrupt thaw features in lowland and upland
settings in Alaska. Left panels (a,c) show thermokarst lake (TKL) abundance,
expansion, and drainage on the Seward Peninsula, Northwest Alaska, between
1950 and 2006
68
, with collapsing permafrost banks (photo credit G.G.). Right
panels (b,d) show extensive distribution of ground collapse and erosion
features (ALD, active layer detachment slide; RTS, retrogressive thaw slump;
GTK, thermal erosion gullies) in upland tundra in a hill slope region in
Northwest Alaska
61
, and thawing icy soils in a retrogressive thaw slump (photo
credit E.A.G.S.).
RESEARCH REVIEW
G2015 Macmillan Publishers Limited. All rights reserved
176 | NATURE | VOL 520 | 9 APRIL 2015
decades and centuries as the permafrost region warms, making climate
change happen even faster than we project on the basis of emissions from
human activities alone. Because of momentum in the system and the
continued warming and thawing of permafrost, permafrost carbon emis-
sions are likely not only during this century but also beyond. Although
never likely to overshadow emissions from fossil fuel, each additional ton of
carbon released from thepermafrost region to the atmosphere will prob-
ably incur additional costs to society.
Next steps for model–data integration
The Earth system models analysed for the IPCC AR5
1
did not include perma-
frost carbon emissions, and there is a need for the nextassessment to make
substantive progress analysing this climate feedback. It is clear, even among
models that are currently capable of simulating permafrost carbon emissions,
that improvements are needed to the simulations of the physical and biological
processes that control the dynamics of permafrost distribution and soil
thermal regime
43,44,57
. The initial model projections we review here are based
on a range of different model formulations, many of which are known to
lack key structural features. Critical next steps that are being achieved by
the research community include a permafrost carbon model intercomparison
using standard driving variables to improve model formulations and con-
ceptualization. Initial intercomparison results point towards several key
structural features that should be implemented by models attempting to
forecast permafrost carbon emissions. These include explicitly defining
the vertical distribution of carbon in permafrost soils to account for the
way atmospheric warming at the surface propagates through the soil, causing
permafrost thaw and carbon decomposition at depth. Additionally, many
large-scale models do not distinguish CH
4
versus CO
2
release and project
only total carbon emissions. This partitioning depends on explicitly de-
scribing the interactions between permafrost thaw and surface hydrology
and is critical to produce credible projections of the effect of permafrost
carbon on climate. A first-order issue is whether the terrestrial landscape
in the permafrost region, already interspersed with thaw lakes, wetlands
and waterlogged soils, becomes wetter ordrier in a warmer world
89
.Lastly,
new modelling formulations for describing abrupt thaw arebeing developed.
These are needed to understand how gradual warming from the surface,
occurring across the entire landscape as currently modelled, compares to
hotspots on the landscape where permafrost undergoes catastrophic ground
collapse and rapid thaw. These issues go beyond temperature sensitivity
alone and are at the forefront of current ecosystem model development
and research.
Models are useful tools for making projections, but need to use observations
more effectively for benchmarking and parameterization. Current models
show a wide range of results when compared against benchmark data sets
of permafrost soil temperatures
44
,soilcarbonstocks
90
, and high-latitude
carbon fluxes
91
, emphasizing the high uncertainty in these projections.
Now, new data sets on decomposability (reviewed here) are available and
should be used to parameterize key aspects of model carbon feedbacks.
The databases on decomposability however, remain two orders of mag-
nitude smaller than surface (,1m) carbon pool data sets. Increasing the
number of laboratory incubations will help to constrain uncertainty regarding
the potential for permafrost carbon to remain stable under different environ-
mental conditions and will allow researchersto understand whichcontrols
over decomposition are most important for the slow turnover pools that
comprise a large fraction of the total permafrost carbon pool. At the same
time, further work is required to quantify the permafrost carbon pool itself
better. Despite substantial recent progress, remote regionssuch as the Canadian
High Arctic, central Siberia, and the subsea continental shelves remain
poorly represented,with very few data points deeper than 1 m. Other data
sets synthesizing field observations of CH
4
emissions and CO
2
exchange
provide process-level understanding available for model validation as
well
40,91–93
. Model–data fusion using these newly created databases from both
laboratory and field observations is urgently needed to evaluate which
models can credibly represent the permafrost region and thus help reduce
the uncertainty in forecasting the permafrost carbon feedback.
BOX 1
Permafrost carbon feedback to
climate change
AsshownintheBox1Figure,carbonstoredfrozeninpermafrost,once
thawed, can enter ecosystems that have either predominantly aerobic
(oxygen present) or predominantly anaerobic (oxygen limited) soil
conditions. Across the permafrost region, there is a gradient of water
saturation that ranges from mostly aerobic upland ecosystems to
mostly anaerobic lowland lakes and wetlands. In aerobic soils, CO
2
is
released by microbial decomposition of soil organic carbon, whereas
both CO
2
and CH
4
are released from anaerobic soils and sediments.
Microbial breakdown of soil organic carbon can happen in the surface
active layer, which thaws each summer and refreezes in the winter,
and in the subsurface as newly thawed carbon becomes available for
decomposition after it has emerged from the perennially frozen pool.
The decomposability of soil organic carbon varies across the
landscape depending in part on the plant inputs as well as the soil
environment, and also with depth in the soil profile. The landscape
mosaic of water saturation is also affected by permafrost thaw.
Gradual and abrupt thaw processes such as top-down thawing of
permafrost (increasing the thickness of the active layer) and lake
draining can expose more carbon to aerobic conditions. Alternatively,
abrupt thaw processes can create wetter anaerobic conditions as the
ground surface subsides, attracting local water. Carbon can also be
mobilized by erosion or by leaching from upland soils into aquatic
systems or sediments. Plant carbon uptake can be stored in increased
plantbiomass or depositedin the surfacesoils, whichin part can offset
losses from soils.
Atmospheric C
CO2CO2
CH4
Plant C
Soil drying
Lake drainage
Lake and wetland formation
Erosion into aquatic systems
Active layer C
Active layer C
Thawed C
Thawed C
Frozen C
AnaerobicAerobic
Box 1 Figure
|
Key features regulating the permafrost carbon
feedback to climate from new, synthesized observations.
REVIEW RESEARCH
G2015 Macmillan Publishers Limited. All rights reserved
9APRIL2015|VOL520|NATURE|177
High-latitude warming and the emission of permafrost carbon remains
a likely global carbon cycle feedback to climate change. The sheer size of these
frozen carbon pools and the rapid changes observed in the permafrost region
warrant focused attention on these remote landscapes. The observations
and modelling steps outlined here will help in forecasting future change.
At the same time, it is imperative to continue developing effective observation
networks, including remote sensing capability
94
, to adequately quantify
real-time CO
2
and CH
4
emissions from permafrost regions
95
. While increased
permafrost carbon emissions in a warming climate are more likely to be
gradual and sustained rather than abrupt and massive, such observation
networks are needed to detect the potential emissions predicted here, and
also to provide early warning of phenomena and potential surprises we do
not yet fully appreciate or understand. The combination of robust observations
with appropriate modelling tools for forecasting change is essential to properly
evaluate permafrost carbonsources. The quantification of carbon sources
in addition to those that are a direct result of human activity is necessary
when developing and evaluating climate change mitigation policies.
Received 14 July 2014; accepted 12 February 2015.
1. IPCC 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 Stocker, T. F. et al.) 1535 (Cambridge Univ. Press, 2013).
2. Brown, J. & Romanovsky, V. E. Report from the International Permafrost
Association: state of permafrost in the first decade of the 21st century. Permafr.
Periglac. Process. 19, 255–260 (2008).
3. Romanovsky, V. E. et al. Thermal state of permafrost in Russia. Permafr. Periglac.
Process. 21, 136–155 (2010).
4. Romanovsky, V. E. et al. Permafrost (Arctic Report Card 2011) http://
www.arctic.noaa.gov/report11/ (2013).
5. Zimov, S. A., Schuur, E. A. G. & Chapin, F. S. Permafrost and the global carbon
budget. Science 312, 1612–1613 (2006).
6. Tarnocai,C. et al. Soil organic carbonpools in the northerncircumpolar permafrost
region. Glob. Biogeochem. Cycles 23, GB2023 (2009).
7. Schuur, E. A. G. et al. Vulnerability of permafrost carbon to climate change:
implications for the global carbon cycle. Bioscience 58, 701–714 (2008).
8. Whiteman,G., Hope, C. & Wadhams, P. Climatescience: vast costs of Arcticchange.
Nature 499, 401–403 (2013).
9. National Research Council. Abrupt Impacts of Climate Change: Anticipating
Surprises (The National Academies Press, 2013).
10. Scha
¨del, C. et al. Short communication on network related activities: research
coordination network on thevulnerability of permafrost carbon. FrozenGround 37,
7(2013).
11. Schuur, E. A. G. et al. Research coordination network on the vulnerability of
permafrost carbon. Frozen Ground 35, 6 (2011).
12. Hugelius, G. et al. Estimated stocks of circumpolar permafrost carbon with
quantified uncertainty ranges and identified data gaps. Biogeosciences 11,
6573–6593 (2014).
Revised and updated current state of knowledge on permafrost soil organic
carbon stocks at circumpolar scales.
13. Zimov, S. A. et al. Permafrost carbon: stock and decomposability of a globally
significant carbon pool. Geophys. Res. Lett. 33, L20502 (2006).
14. Ping, C.-L. et al. High stocks of soil organic carbon in the North American Arctic
region. Nature Geosci. 1, 615–619 (2008).
15. Harden, J. W. et al. Field information links permafrost carbon to physical
vulnerabilities of thawing. Geophys. Res. Lett. 39, L15704 (2012).
Provides cumulative distributions of active layer thickness under current and
future climates and estimates the amounts of newly thawed carbon and
nitrogen.
16. Strauss, J. et al. The deep permafrost carbon pool of the yedoma region in Siberia
and Alaska. Geophys. Res. Lett. 40, 6165–6170 (2013).
Quantifies the organic carbon pool for yedoma deposits and thermokarst
deposits in Siberia and Alaska.
17. Grosse, G. et al. Distribution of late Pleistocene ice-rich syngenetic permafrost of
the Yedoma Suite in east and central Siberia, Russia. US Geol. Surv. Open File Rep.
2013–1078, 1–37 (2013).
18. Walter Anthony, K. M. et al. A shift of thermokarst lakes from carbon sources to
sinks during the Holocene epoch. Nature 511, 452–456 (2014).
19. Mishra, U. et al. Empirical estimates to reduce modeling uncertainties of soil
organic carbon in permafrost regions: a review of recent progress and remaining
challenges. Environ. Res. Lett. 8, 035020 (2013).
20. Hugelius, G. et al. The Northern Circumpolar Soil Carbon Database: spatially
distributed datasets of soil coverage and soil carbon storage in the northern
permafrost regions. Earth Syst. Sci. Data 5, 3–13 (2013).
21. Jobba
´gy, E. G. & Jackson, R. B. The vertical distribution of soil organic carbon and
its relation to climate and vegetation. Ecol. Appl. 10, 423–436 (2000).
22. Schirrmeister, L. et al. Late Quaternary paleoenvironmental records from the
western Lena Delta, Arctic Siberia. Palaeogeogr. Palaeoclimatol. Palaeoecol. 299,
175–196 (2011).
23. Kanevskiy, M., Shur, Y., Fortier, D., Jorgenson,M. T. & Stephani, E. Cryostratigraphy
of late Pleistocene syngenetic permafrost (yedoma) in northern Alaska, Itkillik
River exposure. Quat. Res. 75, 584–596 (2011).
24. Schirrmeister, L. et al. Fossil organic matter characteristics in permafrost deposits
of the northeast Siberian Arctic. J. Geophys. Res. Biogeosci. 116, G00M02 (2011).
25. Hugelius, G. et al. High-resolution mapping of ecosystem carbon storage and
potentialeffects of permafrostthaw in periglacial terrain,European Russian Arctic.
J. Geophys. Res. Biogeosci. 116, G03024 (2011).
26. Brosius, L. S. et al. Using the deuterium isotope composition of permafrost
meltwater to constrain thermokarstlake contributions to atmospheric CH
4
during
the last deglaciation. J. Geophys. Res. Biogeosci. 117, G01022 (2012).
27. McGuire, A. D. et al. Sensitivity of the carbon cycle in the Arctic to climate change.
Ecol. Monogr. 79, 523–555 (2009).
28. Walter, K. M., Edwards, M. E., Grosse, G., Zimov, S. A. & Chapin, F. S. Thermokarst
lakes as a source of atmospheric CH
4
during the last deglaciation. Science 318,
633–636 (2007).
29. Romanovskii, N. N., Hubberten, H. W., Gavrilov, A. V., Eliseeva, A. A. & Tipenko, G. S.
Offshore permafrost and gas hydrate stability zone on the shelf of East Siberian
seas. Geo-Mar. Lett. 25, 167–182 (2005).
30. Nicolsky, D. J. et al. Modeling sub-sea permafrost in the East Siberian Arctic Shelf:
the Laptev Sea region. J. Geophys. Res. Earth Surf. 117, F03028 (2012).
31. Dutta, K., Schuur, E. A. G., Neff, J. C. & Zimov, S. A. Potential carbon release from
permafrost soils of Northeastern Siberia. Glob. Change Biol. 12, 2336–2351
(2006).
32. Knoblauch, C., Beer, C., Sosnin, A., Wagner, D. & Pfeiffer, E.-M. Predicting long-term
carbon mineralization and trace gas production from thawing permafrost of
Northeast Siberia. Glob. Change Biol. 19, 1160–1172 (2013).
33. Elberling, B. et al. Long-term CO
2
production following permafrost thaw. Nature
Clim. Change 3, 890–894 (2013).
34. Kirschbaum, M. U. F. Will changes in soil organic carbon act as a positive or
negative feedback on global warming? Biogeochemistry 48, 21–51 (2000).
35. Ka
¨tterer,T., Reichstein, M., Andren,O. & Lomander, A. Temperature dependenceof
organic matterdecomposition: a criticalreview using literature data analyzedwith
different models. Biol. Fertil. Soils 27, 258–262 (1998).
36. Scha
¨del, C. et al. Circumpolar assessment of permafrost C quality and its
vulnerability over time using long-term incubation data. Glob. Change Biol. 20,
641–652 (2014).
Synthesizesthe decomposability of permafrost organic matterusing incubation
data and calculates potential carbon loss for high-latitude soils.
37. Treat, C. et al. A pan-Arctic synthesis of CH
4
and CO
2
production from anoxic soil
incubations. Glob. Change Biol. doi:10.1111/gcb.12875 (in the press).
38. Myhre, G. et al. in Climate Change 2013: The PhysicalScience Basis. Contributions of
Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on
Climate Change (edsStocker, T. F. et al.) 659–740 (Cambridge Univ. Press, 2013).
39. Verville, J. H., Hobbie, S. E., Iii, F. S. C. & Hooper, D. U. Response of tundra CH
4
and
CO
2
flux to manipulation of temperature and vegetation. Biogeochemistry 41,
215–235 (1998).
40. Olefeldt, D., Turetsky, M.R., Crill, P. M. & McGuire, A. D. Environmental and physical
controlson northern terrestrialmethane emissionsacross permafrost zones.Glob.
Change Biol. 19, 589–603 (2013).
Synthesisof data on growing-season CH
4
emissionsfrom terrestrial ecosystems
across permafrost zones.
41. Treat, C. C. et al. Temperature and peat type control CO
2
and CH
4
production in
Alaskan permafrost peats. Glob. Change Biol. 20, 2674–2686 (2014).
42. Stro
¨m, L., Tagesson,T., Mastepanov,M. & Christensen,T. R. Presence of Eriophorum
scheuchzeri enhances substrate availability and methane emission in an Arctic
wetland. Soil Biol. Biochem. 45, 61–70 (2012).
43. Lawrence, D. M., Slater, A. G., Romanovsky, V. E. & Nicolsky, D. J. Sensitivity of a
model projection of near-surface permafrost degradation to soil column depth
and representation of soil organic matter. J. Geophys.Res. Earth Surf. 113, F02011
(2008).
44. Koven, C. D., Riley, W. J. & Stern, A. Analysis of permafrost thermal dynamics and
response to climate change in the CMIP5 Earth system models. J. Clim. 26,
1877–1900 (2013).
Analysis of Earth system models projections of permafrost change in response
to climate change scenarios.
45. Koven, C. D. et al. Permafrost carbon-climate feedbacks accelerate global
warming. Proc. Natl Acad. Sci. USA 108, 14769–14774 (2011).
46. Schaefer, K., Zhang, T., Bruhwiler, L. & Barrett, A. P. Amount and timing of
permafrost carbon release in response to climate warming. Tellus B 63, 165–180
(2011).
47. Schneider von Deimling, T. et al. Estimating the near-surface permafrost-carbon
feedback on global warming. Biogeosciences 9, 649–665 (2012).
48. MacDougall, A. H., Avis, C. A. & Weaver, A. J. Significant contribution to climate
warming from the permafrost carbon feedback. Nature Geosci. 5, 719–721
(2012).
49. Burke, E. J., Jones, C. D. & Koven, C. D. Estimating the permafrost-carbon climate
response in the CMIP5 climate models using a simplified approach. J. Clim. 26,
4897–4909 (2013).
50. Schaphoff, S. et al. Contribution of permafrost soils to the global carbon budget.
Environ. Res. Lett. 8, 014026 (2013).
51. Burke, E. J., Hartley, I. P. & Jones, C. D. Uncertainties in the global temperature
change caused by carbon release from permafrost thawing. Cryosphere 6,
1063–1076 (2012).
52. Zhuang, Q. et al. CO
2
and CH
4
exchanges between land ecosystems and the
atmosphere in northern high latitudes over the 21st century. Geophys. Res. Lett.
33, L17403 (2006).
RESEARCH REVIEW
G2015 Macmillan Publishers Limited. All rights reserved
178 | NATURE | VOL 520 | 9 APRIL 2015
53. Schuur, E. A. G., Abbott, B., &. the Permafrost Carbon Network. Climate change:
high risk of permafrost thaw. Nature 480, 32–33 (2011).
54. Schuur, E. A. G. et al. Expert assessment of vulnerability of permafrost carbon to
climate change. Clim. Change 119, 359–374 (2013).
State of knowledge on changes in permafrost distribution and soil organic
carbon stocks in response to climate warming based on expert survey.
55. Schaefer, K., Lantuit, H., Romanovsky, V. E., Schuur, E. A. G. & Witt, R. The impactof
the permafrost carbon feedback on global climate. Environ. Res. Lett. 9, 085003
(2014).
56. Lawrence, D. M., Slater, A. G. & Swenson, S. C. Simulationof present-day and future
permafrost and seasonally frozen ground conditions in CCSM4. J. Clim. 25,
2207–2225 (2012).
57. Slater, A. G. & Lawrence, D. M. Diagnosing present and future permafrost from
climate models. J. Clim. 26, 5608–5623 (2013).
58. Shaver, G. R. et al. Global warming and terrestrial ecosystems: a conceptual
framework for analysis. Bioscience 50, 871–882 (2000).
59. Sistla, S. A. et al. Long-term warming restructures Arctic tundra without changing
net soil carbon storage. Nature 497, 615–618 (2013).
60. Qian, H., Joseph, R. & Zeng, N. Enhanced terrestrial carbon uptake in the northern
high latitudes in the 21st century from the Coupled Carbon Cycle Climate Model
Intercomparison Project model projections. Glob. Change Biol. 16, 641–656
(2010).
61. Krieger, K. E. The topographic form and evolution of thermal erosion features: A first
analysis using airborne and ground-based LiDAR in Arctic Alaska. MSc thesis, Idaho
State Univ. (2012).
62. Jorgenson, M. T., Shur, Y. L. & Pullman, E. R. Abrupt increase in permafrost
degradation in Arctic Alaska. Geophys. Res. Lett. 33, L02503 (2006).
63. Christensen, T. R. et al. Thawing sub-arctic permafrost: effects on vegetation and
methane emissions. Geophys. Res. Lett. 31, L04501 (2004).
64. Johansson, T. et al. Decadal vegetation changes in a northern peatland,
greenhousegas fluxes and net radiative forcing.Glob. Change Biol. 12, 2352–2369
(2006).
65. Osterkamp, T. E. Characteristics of the recent warming of permafrost in Alaska.
J. Geophys. Res. Earth Surf. 112, F02S02 (2007).
66. Baltzer, J. L., Veness, T., Chasmer, L. E., Sniderhan, A. E. & Quinton, W. L. Forests on
thawing permafrost: fragmentation, edge effects,and net forest loss. Glob. Change
Biol. 20, 824–834 (2014).
67. Raynolds, M. K. et al. Cumulative geoecological effectsof 62 years of infrastructure
and climate change in ice-rich permafrost landscapes, Prudhoe Bay Oilfield,
Alaska. Glob. Change Biol. 20, 1211–1224 (2014).
68. Jones, B. M. et al. Modern thermokarst lake dynamicsin the continuouspermafrost
zone, northern Seward Peninsula, Alaska.J. Geophys. Res. Biogeosci.116, G00M03
(2011).
69. Smith, L. C., Sheng, Y., MacDonald, G. M. & Hinzman, L. D. Disappearing Arctic
lakes. Science 308, 1429 (2005).
70. Riordan, B., Verbyla, D. & McGuire,A. D. Shrinking ponds in subarctic Alaska based
on 1950–2002 remotely sensed images. J. Geophys. Res. Biogeosci.111, G04002
(2006).
71. Roach, J., Griffith, B., Verbyla, D. & Jones, J. Mechanisms influencing changes in
lake area in Alaskan boreal forest. Glob. Change Biol. 17, 2567–2583 (2011).
72. Sannel, A. B. K. & Kuhry, P. Warming-induced destabilization of peat plateau/
thermokarst lake complexes. J. Geophys. Res. Biogeosci. 116, G03035 (2011).
73. Walter, K. M., Zimov, S. A., Chanton, J. P., Verbyla, D. & Chapin, F. S. III. Methane
bubbling from Siberian thaw lakes as a positive feedback to climate warming.
Nature 443, 71–75 (2006).
74. Jones, M. C., Grosse, G., Jones, B. M. & Walter Anthony, K. Peat accumulation in
drained thermokarst lake basins in continuous, ice-rich permafrost, northern
Seward Peninsula, Alaska. J. Geophys. Res. 117, G00M07 (2012).
75. Zona, D. et al. Increased CO
2
loss from vegetated drained lake tundra ecosystems
due to flooding. Glob. Biogeochem. Cycles 26, GB2004 (2012).
76. Olefeldt, D. & Roulet, N. T. Permafrost conditions in peatlands regulate magnitude,
timing, and chemical composition of catchment dissolved organic carbon export.
Glob. Change Biol. 20, 3122–3136 (2014).
77. Feng, X. et al. Differential mobilization of terrestrialcarbon pools in Eurasian Arctic
river basins. Proc. Natl Acad. Sci. USA 110, 14168–14173 (2013).
78. Vonk, J. E. & Gustafsson, O. Permafrost-carbon complexities. Nature Geosci. 6,
675–676 (2013).
79. Vonk, J. E. et al. High biolability of ancientpermafrost carbon upon thaw. Geophys.
Res. Lett. 40, 2689–2693 (2013).
80. Cory, R. M., Crump, B. C., Dobkowski, J. A. & Kling, G. W. Surface exposure to
sunlight stimulates CO
2
release from permafrost soil carbon in the Arctic. Proc.
Natl Acad. Sci. USA 110, 3429–3434 (2013).
81. Vonk, J. E. et al. Activation of old carbon by erosion of coastal and subsea
permafrost in Arctic Siberia. Nature 489, 137–140 (2012).
82. Shakhova, N. et al. Geochemicaland geophysical evidenceof methane release over
the East Siberian Arctic Shelf. J. Geophys. Res. Oceans 115, C08007 (2010).
83. Shakhova, N. et al. Ebullition and storm-induced methane release from the East
Siberian Arctic Shelf. Nature Geosci. 7, 64–70 (2014).
Quantitative assessment of bubble-induced CH
4
emissions resulting from
subsea permafrost degradation in the coastal area.
84. Dmitrenko, I. A. et al. Recent changes in shelf hydrography in the Siberian Arctic:
Potential for subsea permafrost instability. J. Geophys. Res. Oceans 116, C10027
(2011).
85. Parmentier, F.-J. W. et al. The impact of lowersea-ice extent on Arctic greenhouse-
gas exchange. Nature Clim. Change 3, 195–202 (2013).
86. Notz, D., Brovkin, V. & Heimann, M. Arctic: uncertainties in methane link. Nature
500, 529–529 (2013).
87. Parmentier, F.-J. W. & Christensen, T. R. Arctic: speed of methane release. Nature
500, 529–529 (2013).
88. Le Que
´re
´,C.et al. Global carbon budget 2013. Earth Syst. Sci. Data 6, 235–263
(2014).
89. Avis, C. A., Weaver, A. J. & Meissner, K. J. Reduction in areal extent of high-latitude
wetlands in response to permafrost thaw. Nature Geosci. 4, 444–448 (2011).
90. Todd-Brown, K. E. O. et al. Causes of variation in soil carbon simulations from
CMIP5 Earth system models and comparison with observations. Biogeosciences
10, 1717–1736 (2013).
91. McGuire, A. D. et al. An assessment of the carbon balance of Arctic tundra:
comparisons among observations, process models, and atmospheric inversions.
Biogeosciences 9, 3185–3204 (2012).
92. Belshe, E. F., Schuur, E. A. G. & Bolker,B. M. Tundra ecosystems observedto be CO
2
sources due to differential amplification of the carbon cycle. Ecol. Lett. 16,
1307–1315 (2013).
93. Ueyama, M. et al. Upscaling terrestrialcarbon dioxide fluxes in Alaska withsatellite
remote sensing and support vector regression. J. Geophys. Res. Biogeosci. 118,
1266–1281 (2013).
94. National Research Council. Opportunities to Use Remote Sensing in Understanding
Permafrost and Related Ecological Characteristics: Report of a Workshop http://
dels.nationalacademies.org/Report/Opportunities-Remote-Sensing/18711 (The
National Academies Press, 2014).
95. Schaefer, K., Lantuit, H., Romanovsky, V. E. & Schuur, E. A. G. Policy Implications of
Warming Permafrost (United Nations Environment Program, 2012).
96. Brown, J., Ferrians, O. J. J., Heginbottom, J. A. & Melnikov, E. S. Circum-Arctic Map of
Permafrost and Ground-Ice Conditions. Version 2, http://nsidc.org/data/docs/
fgdc/ggd318_map_circumarctic/index.html (National Snow and Ice Data Center,
2002).
97. IPCC in Climate Change 2007: The Physical Science Basis. Contribution of Working
Group I to the Fourth Assessment Report (AR4) of the Intergovernmental Panel on
Climate Change (eds Solomon, S. D. et al.) (Cambridge Univ. Press, 2007).
Acknowledgements Initial funding was provided by the National Science Foundation
Vulnerability of Permafrost Carbon Research Coordination Network Grant number
955713, with continued support from the National Science Foundation Research,
Synthesis,and Knowledge Transferin a Changing Arctic: ScienceSupport for the Study
of Environmental Arctic Change Grant number 1331083. Author contributions were
also supportedby grants to individuals: Departmentof Energy Office of Science, Office
of Biological and Environmental Sciences Division Terrestrial Ecosystem Sciences
program (DE-SC0006982) to E.A.G.S.; National Science Foundation Long Term
Ecological Research Program (1026415) to A.D.M.; Department of Energy (DE-AC02-
05CH11231,NGEE Arctic, BGC-FeedbacksSFA) to C.D.K.; Regional and GlobalClimate
Modeling Program (RGCM) of the US Department of Energy’s Office of Science (BER)
Cooperative Agreement (DE-FC02-97ER62402) to D.M.L.; European Research
Commission (338335) to G.G.; The Netherlands Organization for Scientific Research
(863.12.004) to J.E.V.; National Science Foundation Polar Programs (1312402) to
S.M.N.;National Science FoundationPolar Programs (856864and 1304271) to V.E.R.;
National Oceanic and AtmosphericAdministration (NA09OAR4310063) and National
Aeronauticsand Space Agency (NNX10AR63G)to K.S.; Nordforsk (DEFROST;23001),
EU FP7 (PAGE21; 282700) and FORMAS (Bolin Climate Research Centre; 214-2006-
1749) to G.H. and P.K.; Department of Energy Biological and Environmental Research
(3ERKP818) to D.J.H.; National ScienceFoundation, Division of Environmental Biology
(724514, 830997) to M.R.T. and A.D.M.; U.S. Geological Survey Climate and Land Use
Program to J.W.H. Any use of trade, firm, or product names is for descriptive purposes
only and does not imply endorsement by the US Government.
Author Contributions This manuscript arose from the collective effort of the
Permafrost Carbon Network (http://www.permafrostcarbon.org); all authors are
workinggroup leaders withinthe network. E.A.G.S.and A.D.M. wrote the initialdraft, with
additionalcontributionsfrom all authors. C.S. providedassistance with final editingand
submissionof the manuscript, and helped to organise the PermafrostCarbon Network
activities that made this possible. Figure 1 was prepared by G.H., Fig. 2 by C.S., Fig. 3
by K.S., Fig. 4 by G.G. and the Box 1 Figure by E.A.G.S.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Readers are welcome to comment on the online version of the paper. Correspondence
and requests for materials should be addressed to E.A.G.S. (ted.schuur@nau.edu).
REVIEW RESEARCH
G2015 Macmillan Publishers Limited. All rights reserved
9APRIL2015|VOL520|NATURE|179
