Page 1
Biogeosciences, 11, 2341–2356, 2014
www.biogeosciences.net/11/2341/2014/
doi:10.5194/bg-11-2341-2014
© Author(s) 2014. CC Attribution 3.0 License.
Biogeosciences
Open Access
Changes in soil organic carbon storage predicted by Earth system
models during the 21st century
K. E. O. Todd-Brown
1
, J. T. Randerson
1
, F. Hopkins
1
, V. Arora
2
, T. Hajima
3
, C. Jones
4
, E. Shevliakova
5
, J. Tjiputra
6
,
E. Volodin
7
, T. Wu
8
, Q. Zhang
9
, and S. D. Allison
1,10
1
Department of Earth System Science, University of California, Irvine, California, USA
2
Canadian Centre for Climate Modelling and Analysis, Environment Canada, University of Victoria, Victoria, British
Columbia, Canada
3
Japan Agency for Marine-Earth Science and Technology, Yokohama, Japan
4
Met Office Hadley Centre, Exeter, UK
5
Department of Ecology and Evolutionary Biology, Princeton University, Princeton, New Jersey, USA
6
Uni Climate, Uni Research AS and Bjerknes Centre for Climate Research, Bergen, Norway
7
Institute of Numerical Mathematics, Russian Academy of Sciences, Moscow, Russia
8
Beijing Climate Center, China Meteorological Administration, Beijing, China
9
College for Global Change and Earth System Science, Beijing Normal University, Beijing, China
10
Ecology and Evolutionary Biology Department, University of California, Irvine, California, USA
Correspondence to: K. E. O. Todd-Brown (ktoddbro@uci.edu)
Received: 11 November 2013 – Published in Biogeosciences Discuss.: 4 December 2013
Revised: 21 February 2014 – Accepted: 4 March 2014 – Published: 25 April 2014
Abstract. Soil is currently thought to be a sink for car-
bon; however, the response of this sink to increasing lev-
els of atmospheric carbon dioxide and climate change is
uncertain. In this study, we analyzed soil organic carbon
(SOC) changes from 11 Earth system models (ESMs) con-
tributing simulations to the Coupled Model Intercompari-
son Project Phase 5 (CMIP5). We used a reduced complex-
ity model based on temperature and moisture sensitivities
to analyze the drivers of SOC change for the historical and
high radiative forcing (RCP 8.5) scenarios between 1850 and
2100. ESM estimates of SOC changed over the 21st cen-
tury (2090–2099 minus 1997–2006) ranging from a loss of
72 Pg C to a gain of 253 Pg C with a multi-model mean gain
of 65 Pg C. Many ESMs simulated large changes in high-
latitude SOC that ranged from losses of 37 Pg C to gains of
146 Pg C with a multi-model mean gain of 39 Pg C across
tundra and boreal biomes. All ESMs showed cumulative in-
creases in global NPP (11 to 59 %) and decreases in SOC
turnover times (15 to 28 %) over the 21st century. Most of
the model-to-model variation in SOC change was explained
by initial SOC stocks combined with the relative changes in
soil inputs and decomposition rates (
R
2
=
0
.
89,
p <
0
.
01).
Between models, increases in decomposition rate were well
explained by a combination of initial decomposition rate,
ESM-specific
Q
10
-factors, and changes in soil temperature
(
R
2
=
0
.
80,
p <
0
.
01). All SOC changes depended on sus-
tained increases in NPP with global change (primarily driven
by increasing CO
2
)
. Many ESMs simulated large accumula-
tions of SOC in high-latitude biomes that are not consistent
with empirical studies. Most ESMs poorly represented per-
mafrost dynamics and omitted potential constraints on SOC
storage, such as priming effects, nutrient availability, mineral
surface stabilization, and aggregate formation. Future models
that represent these constraints are likely to estimate smaller
increases in SOC storage over the 21st century.
1
Introduction
The global pool of soil organic carbon (SOC) is large rela-
tive to atmospheric CO
2
(Jobbagy and Jackson, 2000), and
changes in soil–atmosphere fluxes of carbon could gener-
ate either a positive or negative feedback to climate over
the next century. For example, the contemporary soil carbon
Published by Copernicus Publications on behalf of the European Geosciences Union.
Page 2
2342
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
sink in forests is estimated to be approximately 0.9 Pg C yr
−
1
(Pan et al., 2011). However, uncertainties in the response
of soil inputs and heterotrophic respiration to global change
drivers, along with difficulties in measuring heterogeneous
SOC pools, lead to relatively large uncertainties in below-
ground flux estimates (Houghton, 2003; Le Quéré et al.,
2009). The balance of SOC over the next century is even
less certain (Friedlingstein et al., 2006). Much of this uncer-
tainty is due to the environmental sensitivities of SOC input
and output fluxes that depend on environmental variables and
will likely change with climate over the next century (Giorgi,
2006).
Net primary production (NPP) provides the primary input
of carbon to soil and is sensitive to climate. NPP generally
increases with temperature, moisture, and CO
2
up to some
maximum, in turn providing increased carbon inputs to soil
(Chapin and Eviner, 2007; Körner, 2006). Free-air CO
2
en-
richment (FACE) studies have shown that NPP increases by
23 % on average across forest ecosystems in response to CO
2
increases from 365 to 550–580 ppm (Norby et al., 2005).
Similar increases in NPP have been observed in shrub- and
grassland-elevated CO
2
experiments (De Graaff et al., 2006;
Hungate et al., 2013). However, increases in NPP may be
constrained by available nutrients; the theory of progressive
nutrient limitation posits that NPP responses to elevated CO
2
will be limited by the supply of soil nutrients, particularly ni-
trogen (Hungate et al., 2009; Luo et al., 2004; Norby and
Zak, 2011; Norby et al., 2010; Nowak et al., 2004).
In addition, it remains unclear whether increases in NPP
will translate into increased SOC storage. FACE studies of-
ten observe no change in SOC despite increased NPP, pos-
sibly due to increased loss rates of C inputs (Hofmockel et
al., 2011; Hoosbeek and Scarascia-Mugnozza, 2009; Phillips
et al., 2012) or increased decomposition of SOC through
the priming effect (Carney et al., 2007; Cheng et al., 2014;
Talhelm et al., 2009). With priming, fresh carbon inputs as-
sociated with increasing NPP stimulate the microbial de-
composition of SOC, thereby preventing SOC accumulation
(Fontaine et al., 2004; Hungate et al., 2013; Kuzyakov et al.,
2000; Phillips et al., 2011; Wieder et al., 2014). Alternatively,
SOC accumulation under elevated CO
2
may be difficult to
measure due to spatial heterogeneity in SOC pools and the
short timescale of the experiments relative to soil turnover
times (Billings et al., 2010; Schlesinger and Lichter, 2001).
Heterotrophic respiration is the primary loss pathway for
SOC and is also sensitive to climate change. Heterotrophic
respiration generally increases with temperature (Davidson
and Janssens, 2006) and moisture levels in well-drained soils
(Cook and Orchard, 2008). Many studies have hypothesized
that rising temperatures will increase SOC losses through de-
creased soil turnover times (Davidson and Janssens, 2006;
Lloyd and Taylor, 1994). Such changes should increase het-
erotrophic respiration, especially in the high northern lati-
tudes (Schuur et al., 2008), and contribute to increases in
global atmospheric CO
2
(Koven et al., 2011). However, other
effects like aggregate formation and mineral–organic interac-
tions could stabilize SOC, limiting the response to increased
temperature (Dungait et al., 2012; Six et al., 2002; Torn et
al., 1997).
Earth system models (ESMs) are the primary tools for pre-
dicting climate impacts on carbon storage at the global scale.
The Coupled Model Intercomparison Project, currently in its
5th phase (CMIP5), enables direct comparison of ESMs by
providing a suite of common drivers and standardized output
formats (Taylor et al., 2011). Analysis of CMIP5 ESMs indi-
cates that the negative feedback between land carbon and at-
mospheric CO
2
concentration is larger than the positive feed-
back between land carbon and climate warming (Arora et al.,
2013). Also, inter-model variation in atmospheric CO
2
pro-
jections during the second half of the 21st century correlates
with biases that were present by 2005 (Hoffman et al., 2014).
Observational data on the sensitivity of atmospheric CO
2
growth rate to tropical temperature have been used to con-
strain the future climate response of tropical carbon stocks
projected by CMIP4 ESMs (Cox et al., 2013). Finally, the
IPCC 5th assessment report (Ciais et al., 2013) shows that
land carbon storage is the largest source of uncertainty in fu-
ture carbon cycle projections, and this uncertainty has not
changed since CMIP4 (Friedlingstein et al., 2006).
Previous work has shown that within and between ESMs,
variations in contemporary SOC stocks are mostly driven by
model estimates of NPP, parameterization of the intrinsic de-
composition rate (fitted at 15
◦
C and optimal soil moisture),
and the temperature dependency of heterotrophic respiration
(Todd-Brown et al., 2013). In high northern latitudes, soil
carbon stocks from CMIP5 ESMs deviate substantially from
observations mainly because ESMs do not directly repre-
sent permafrost carbon (Koven et al., 2011, 2013b; Todd-
Brown et al., 2013). Surprisingly, soil moisture variations
have very little effect on contemporary SOC stocks in most
CMIP5 models (Todd-Brown et al., 2013), in contrast to ex-
perimental studies (Cook and Orchard, 2008) and previous
work carried out with individual models (Exbrayat et al.,
2013; Falloon et al., 2011; Ise and Moorcroft, 2006). Overall,
ESMs perform poorly when global soil carbon distributions
are compared to benchmark data sets, calling into question
their ability to simulate future changes in SOC (Todd-Brown
et al., 2013). However, to date no studies have analyzed pat-
terns and drivers in ESM-simulated SOC change over the
21st century.
Because of the potential importance of SOC for future
carbon–climate feedbacks, the goal of our current study was
to evaluate ESM estimates of global SOC changes during the
21st century from ESM simulations contributed to CMIP5.
Specifically, we aimed to (1) compare SOC changes over the
21st century across ESMs, (2) identify the drivers of SOC
change within and between ESMs, and (3) assess the reliabil-
ity of ESM projections by evaluating these drivers and SOC
changes in the context of global data sets and empirical find-
ings available from global change studies. Our analyses were
Biogeosciences, 11, 2341–2356, 2014
www.biogeosciences.net/11/2341/2014/
Page 3
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
2343
conducted with the historical and RCP 8.5 future scenarios,
where RCP 8.5 assumes “business as usual” greenhouse gas
emissions leading to CO
2
concentrations of 935 ppm by 2100
(Riahi et al., 2007).
2
Methods
2.1
Earth system models
Outputs from ESMs that contributed to CMIP5 (Taylor et al.,
2011) were downloaded from the Earth System Grid Federa-
tion repository. The terrestrial decomposition sub-models of
these ESMs all use systems of first-order linear ordinary dif-
ferential equations with 1–9 substrate pools. The decomposi-
tion and transfer rates of the substrate pools have temperature
sensitivities that are either
Q
10
, Arrhenius, increase to an op-
timal point and then decrease, or some linear approximation
of these functions. In response to soil moisture, decompo-
sition rates either increase monotonically or increase to an
optimal point and then decrease (Table S1, extended from
Todd-Brown et al., 2013).
Results from the historical and RCP 8.5 experiments (Tay-
lor et al., 2011) were analyzed. The historical simulations
were forced with observation-based estimates of CO
2
and
other greenhouse gas mixing ratios, aerosol emissions, and
land use change scenarios, where appropriate, from 1850–
2005. Some models also incorporated natural variability
through specified changes in solar radiation and volcanic ac-
tivity. All models started at 1850, except for GFDL-ESM2G
and HadGEM2-ES, which began at 1861 and 1860, respec-
tively. The RCP 8.5 is a high radiative forcing scenario with a
prescribed atmospheric CO
2
mole fraction ranging from 378
to 935 ppm for the period 2006–2100. The outputs from the
historical and RCP 8.5 experiments were merged to create a
continuous record from 1850 to 2100 (Table S1). We chose
to use model runs with prescribed CO
2
concentrations, as op-
posed to emissions-driven experiments, for consistent com-
parison of the strength of the CO
2
fertilization effects across
models and to avoid changes in projected temperature and
climate variability due to different CO
2
concentrations in the
models.
We used annual, globally gridded SOC, litter, coarse
woody debris carbon, soil temperature, total soil water, het-
erotrophic respiration, and NPP in our analysis (cSoil, cLit-
ter, cCwd, tsl, mrso, rh, and npp, respectively, from the
CMIP5 variable list). The reported monthly values for each
model were averaged to create annual gridded means. Not
all ESMs reported litter or coarse woody debris carbon; thus
soil, litter, and coarse woody debris carbon variables were
summed and are referred to as “soil organic carbon” (SOC)
throughout this analysis. If multiple ensembles were reported
for an individual model, then the ensembles were averaged.
Global totals and means were constructed using the land cell
area and the land surface fraction (areacella and sftlf, respec-
tively). Soil temperature used in this analysis was calculated
as the depth-weighted average from the top 10 cm. Soil wa-
ter for each model was expressed as a gridded fractional soil
water content, calculated from the total soil water content,
and divided by the maximal grid soil water content in the
first 10 years of the historical run. In general, ESMs did not
maintain soil carbon balance with the carbon flux variables
described above (additional carbon fluxes that consume or
diverted NPP in some ESMs included grazing, harvest, land
use change, and fire); thus we computed annual soil inputs
for each model as the annual change in SOC (
1C)
plus the
carbon lost due to heterotrophic respiration (
R)
summed for
each year, that is,
I
=
1C
+
R
. Soil turnover times (inverse
of the decomposition rates) were calculated annually from
gridded and global totals of SOC stocks and heterotrophic
respiration. Soil carbon fluxes were calculated from the grid-
ded and global annual differences in SOC stocks.
Many modeling centers submitted multiple ESMs to the
CMIP5 repository. In these cases, one model was selected
from each model center for analysis. If a model was recom-
mended by the modeling center, that model was used. Oth-
erwise, the model with the highest grid resolution was used.
Finally, if no model was preferred by the modeling center
and the models were of equal resolution, then a model was
randomly selected.
We selected one ESM per modeling center because there
were high correlations in the spatial distribution of 21st cen-
tury SOC change across models within modeling centers. In
general, the spatial patterns of SOC change during the 21st
century in models from the same center were more simi-
lar than the spatial patterns from different modeling centers
(Fig. S1), similar to previous results with spatial patterns in
modern SOC (Todd-Brown et al., 2013). Thus model out-
puts from the same center are non-independent, which could
inflate the significance of statistical tests (e.g., regressions)
applied to multiple models. In some cases, models from dif-
ferent centers also produced highly correlated predictions
because they used the same land carbon sub-model (e.g.,
CESM1(BGC), and NorESM1-ME). In these cases, we still
included one model from each center so as not to exclude
other factors that might differ across modeling centers, in-
cluding ocean and atmospheric components of the ESMs that
influence precipitation and land surface temperatures. We
emphasize that these model predictions are not completely
independent and that statistical tests based on them should
be interpreted with caution.
2.2
Biome definitions
We conducted biome-level analyses by constructing a com-
mon biome mask (similar to Todd-Brown et al. (2013)). We
downloaded vegetation type classifications for the period
2001–2009 from the MODIS satellite mission to construct
the biome map (Friedl et al., 2010; NASA Land Processes
Distributed Active Archive Center (LP DAAC), 2008). The
www.biogeosciences.net/11/2341/2014/
Biogeosciences, 11, 2341–2356, 2014
Page 4
2344
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
vegetative area coverage was regridded from the 0.05
×
0.05
MODIS grid to each individual ESM grid cell using an area-
weighted scheme. The maximal vegetation type for an indi-
vidual grid cell was then used to assign a biome.
2.3
Contribution of inputs and outputs to SOC change
We used two techniques to separate the relative contributions
of changes in carbon inputs and outputs to the total change in
SOC over time. For the first technique, we constructed two
possible temporal scenarios for gridded SOC stock evolution
to illustrate the relative impact of changes in inputs versus
decomposition rate on SOC. In the “constant decomposition
rate” scenario, soil carbon inputs evolved as predicted by the
model but the SOC decomposition rate was held constant at
the 1850 value. In the “constant soil inputs” scenario, de-
composition rate evolved as predicted by the model but soil
carbon input was held constant at the 1850 value. This is the
only analysis that used the entire 1850–2100 period. We ex-
amined this interval to check for consistency in ecosystem
responses over the historical and forecasted time periods.
For the second technique, we modeled ESM change in soil
carbon over the 21st century as a function of the change in
soil inputs and decomposition rate. First we assumed that the
change in soil carbon is proportional to the change in steady-
state soil carbon expected for soil inputs and decomposition
occurring at the start and end of the 21st century:
C
end
−
C
start
=
I
end
k
end
−
I
start
k
start
,
(1)
where end and start are averaged over 2090–2099 and 1997–
2006, respectively;
C
is the global soil carbon stock;
I
is the
associated average soil carbon input; and
k
is the decomposi-
tion rate calculated from global heterotrophic respiration and
soil carbon stocks. We can rearrange this equation to gener-
ate the following:
C
end
−
C
start
=
1
+
1I
I
start
1
+
1k
k
start
−
1
!
C
start
.
(2)
Using regression analysis with the terms on the right-hand
side of Eq. (2), we assessed the relative contributions of
changes in soil inputs, changes in decomposition rate, and
initial soil carbon stocks to changes in ESM soil carbon.
2.4
Drivers of changes in heterotrophic respiration and
decomposition rate
To investigate the drivers of heterotrophic respiration change
and decomposition, we used a reduced complexity model
(Todd-Brown et al., 2013). This model assumes that the
change in heterotrophic respiration is proportional to total
SOC:
R
=
kC,
(3)
where
R
is the heterotrophic respiration,
k
is the decompo-
sition rate (inverse of the turnover time), and
C
is the total
SOC. We then evaluated drivers of the change in decomposi-
tion rate (
k)
by assuming the decomposition rate is dependent
on the intrinsic decomposition rate (
κ
, spatially and tempo-
rally constant) times the temperature dependency (
Q
10
(T ))
and soil moisture dependency (
W
raised to the power
b)
of
decomposition for each model:
k
=
κQ
T
−
15
10
10
W
b
,
(4)
where the temperature (
T )
dependency function represents a
Q
10
-factor increase for each 10
◦
C of warming from a 15
◦
C
baseline, and the moisture dependency function monotoni-
cally increases (
b
greater than 0) with respect to relative wa-
ter content.
After substituting Eq. (4) into Eq. (3), we represented het-
erotrophic respiration as a function of SOC and the environ-
mental drivers:
R
=
κQ
T
−
15
10
10
W
b
C.
(5)
Equation (5) was then fitted to ESM variables as described
below to generate ESM-specific
κ
,
Q
10
, and
b
parameters.
We also used Eq. (4) to derive the change in decomposition
rate by taking the first-order derivative as follows:
d
k
d
t
=
k
ln
(Q
10
)
10
d
T
d
t
+
b
W
d
W
d
t
.
(6)
We then assumed that the rate of change can be approximated
by a single time step over the entire time period, giving the
following simplification:
1k
=
k
start
ln
(Q
10
)
1T
10
+
b
1W
W
start
,
(7)
where
k
start
, and
W
start
are the contemporary (1997–2006)
10-year mean of decomposition rate (soil respiration divided
by soil carbon) and relative soil water content, respectively;
1k
,
1W
, and
1T
are the changes from the contemporary
(1997–2006) to the final (2090–2099) 10-year means of de-
composition rate, soil water, and soil temperature. Note that
Eq. (7) is independent of the inferred intrinsic decomposition
rate (
κ
). In our analyses we simplified Eq. (7) to include just
the temperature term because soil water was not a signifi-
cant term in the model. Gridded values were used for within-
model comparisons. Global total SOC and heterotrophic res-
piration were used for between-model analyses with the de-
composition rate calculated from the previously mentioned
global totals. Area-weighted global mean soil temperature
and soil water content were used for the between-model com-
parisons.
Biogeosciences, 11, 2341–2356, 2014
www.biogeosciences.net/11/2341/2014/
Page 5
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
2345
Parameterization
The intrinsic decomposition rate (
κ
) and environmental de-
pendency parameters (
Q
10
,
b
) in Eq. (5) were fitted using
the ESM initial historical 10-year gridded mean (1850–1859
in most cases) of heterotrophic respiration, SOC, soil tem-
perature, and soil moisture. The parameters were fitted us-
ing a constrained Broyden–Fletcher–Goldfarb–Shanno op-
timization algorithm, a quasi-Newtonian method, as imple-
mented in R 2.13.1 (R Development Core Team, 2012).
This algorithm was selected for parameter fitting because
of its robust convergence and short run time. The follow-
ing variable ranges were considered in the parameterization:
k
10
−
4
,
10
,
Q
10
(
1
,
4
)
, and
b (
0
,
3
)
.
2.5
Benchmark constraints
Two modern benchmark constraints were used to select
models that best represent the modern carbon cycle with
the expectation that the best models would simulate a nar-
rower range of soil carbon change over the 21st century.
Modern global soil carbon stocks were constrained us-
ing estimates from the Harmonized World Soil Database
(FAO/IIASA/ISRIC/ISSCAS/JRC, 2012) with qualitative
uncertainties (Todd-Brown et al., 2013). Modern global NPP
estimates were taken from Ito (2011).
3
Results
Over the 21st century, ESMs predicted SOC changes rang-
ing from a 253 Pg C gain (HadGEM2-ES) to a 72 Pg C loss
(MIROC-ESM) with a multi-model mean gain of 65 Pg C
(Fig. 1, Tables 1 and 2). Five models (HadGEM2-ES, MPI-
ESM-MR, BCC-CSM1.1(m), BNU-ESM, and INM-CM4)
estimated SOC gains greater than 75 Pg C. Four other mod-
els (IPSL-CM5A-MR, GFDL-ESM2G, CESM1(BGC), and
NorESM1-ME) predicted moderate changes in SOC rang-
ing from a 15 Pg C gain to a 16 Pg C loss. Two models
(CanESM2 and MIROC-ESM) projected an SOC loss of
more than 50 Pg C. In many cases, large absolute changes
in SOC also translated into large relative changes, includ-
ing a 23 % gain by HadGEM2-ES, a 22 % gain by BCC-
CSM1.1(m), and a 14 % gain by BNU-ESM (Fig. 1, Table 2).
For four of the five models with the highest gains (more
than 75 Pg C: HadGEM2-ES, MPI-ESM-MR, BNU-ESM,
and INM-CM4), most of the global SOC gain was located in
boreal and tundra biomes; BCC-CSM1.1(m) was an excep-
tion, showing roughly equal gains in high northern latitude
and other biomes (Fig. 2, Table 1). In contrast, the two mod-
els showing the greatest losses in SOC (more than 50 Pg C:
MIROC-ESM and CanESM2) showed most of those losses
in the tropical forest and grassland and savanna biomes (Ta-
ble 1). Three of the four models with moderate SOC changes
showed more SOC change in other biomes than in high
northern latitude biomes; GFDL-ESM2G was an exception
and showed a 37 Pg C loss in the northern latitudes balanced
by a 41 Pg C gain in the other latitudes (Table 1). These di-
verging patterns were also apparent in maps of SOC change
(Figs. 2 and 3) and tended to follow the distribution of SOC
at the beginning of the analysis period (Fig. S2).
SOC change was not well constrained by modern SOC and
NPP benchmarks (Fig. 4). Although 5 of the 11 models met
modern SOC benchmarks, SOC change predicted by these
models ranged from a 51 Pg C loss (CanESM2) to a 203 Pg C
gain (BCC-CSM1.1(m)). Further constraining models based
on modern SOC and NPP global totals reduced the number
of models considered from five to two; however, the range of
SOC change remained the same.
3.1
Changes in NPP, soil inputs, respiration, and
turnover times
All of the models had increases in NPP and thus soil in-
puts during the 21st century (Table 2, Fig. 1). NPP in-
creases occurred across the globe in most models, with ab-
solute changes ranging from 5 Pg C yr
−
1
(NorESM1-ME)
to 46 Pg C yr
−
1
(MPI-ESM-MR). Relative global NPP in-
creases varied between 11 % (NorESM1-ME) and 59 %
(HadGEM2-ES) (Table 2). NPP increases generally followed
contemporary distributions (Figs. S3, S4, S5); however, de-
creases in NPP were notable in the Amazon Basin in three
models (BCC-CSM1.1(m), HadGEM2-ES, and CanESM2)
and southern sections of North America, southern Africa,
and southwest South America in IPSL-CM5A-MR (Figs. S4,
S5). For most models, global SOC input for the period 1997–
2006 and 2090–2099 matched the respective mean annual
NPP relatively closely, with global differences of less than
7 Pg C yr
−
1
(approximately 15 % of NPP) over the 21st cen-
tury (Table 2). However, there were more substantial dif-
ferences of 13–36 Pg C yr
−
1
between NPP and soil inputs
in three models (IPSL-CM5A-MR, GFDL-ESM2G, MPI-
ESM-MR). These differences could be due to either accu-
mulation of carbon in vegetation (and thus time delays in the
delivery of this carbon to SOC pools) or losses through other
pathways such as land use change, harvesting, or fire.
Global
heterotrophic
respiration
also
increased
in
all
models,
with
absolute
increases
ranging
from
7 Pg C yr
−
1
(CESM1(BGC), NorESM1-ME) to 43 Pg C yr
−
1
(HadGEM2-ES) (Table 2). These increases in heterotrophic
respiration were caused by the increase in soil inputs
described above and, simultaneously, decreases in global
turnover times between 12 years (25 %; MIROC-ESM) and
2.1 years (15 %; CESM1(BGC)) (Fig. 1, turnover calculated
by dividing global soil carbon by global heterotrophic
respiration). In many ESMs, gridded turnover times de-
creased by 100 years or more in high northern latitudes
(Figs. S6, S7). For comparison, contemporary heterotrophic
respiration fluxes ranged from 43 Pg C yr
−
1
(CESM1(BGC))
to 76 Pg C yr
−
1
(MPI-ESM-MR) and global turnover
times ranged from 13 years (CESM1(BGC)) to 46 years
www.biogeosciences.net/11/2341/2014/
Biogeosciences, 11, 2341–2356, 2014
Page 6
2346
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
0
500
1000
1500
2000
2500
3000
3500
[Pg C]
Contemporary
Soil carbon
0
20
40
60
80
100
[Pg C yr
−
1
]
NPP
0
10
20
30
40
50
[yr]
T
ur
no
v
er time
CESM1(BGC)
NorESM1−ME
BNU−ESM
BCC−CSM1.1(m)
HadGEM2−ES
IPSL−CM5A−MR
GFDL−ESM2G
CanESM2
INM−CM4
MIR
OC−ESM
MPI−ESM−MR
−200
−100
0
100
200
300
[Pg C]
21st century absolute change
0
10
20
30
40
50
60
[Pg C yr
−
1
]
−14
−12
−10
−8
−6
−4
−2
0
[yr]
CESM1(BGC)
NorESM1−ME
BNU−ESM
BCC−CSM1.1(m)
HadGEM2−ES
IPSL−CM5A−MR
GFDL−ESM2G
CanESM2
INM−CM4
MIR
OC−ESM
MPI−ESM−MR
−10
0
10
20
30
[%]
21st century relative change
0
10
20
30
40
50
60
70
[%]
−30
−25
−20
−15
−10
−5
0
[%]
CESM1(BGC)
NorESM1−ME
BNU−ESM
BCC−CSM1.1(m)
HadGEM2−ES
IPSL−CM5A−MR
GFDL−ESM2G
CanESM2
INM−CM4
MIR
OC−ESM
MPI−ESM−MR
Fig. 1. Contemporary values, absolute changes, and relative changes of soil carbon, net primary production (NPP), and turnover times for
CMIP5 Earth system models. Contemporary values of soil carbon, NPP, and turnover times are reported for the 1997–2006 mean using
model output from the historical experiment. Changes for the 21st century were estimated from the difference between 2090–2099 and
1997–2006 mean model estimates from the RCP 8.5 experiment. Turnover times were calculated from global soil carbon stocks divided by
global heterotrophic respiration.
(MIROC-ESM) (Fig. 1, Table 2, and starting distributions in
Fig. S8).
3.2
Changes in net SOC flux
By the end of the 21st century, the net SOC flux was nega-
tive (carbon being lost from the soil) in most models, ranging
between
−
0.2 Pg C yr
−
1
(BNU-ESM and CESM1(BGC))
and
−
3.0 Pg C yr
−
1
(MIROC-ESM), with the excep-
tions of HadGEM1-ES (+3.0 Pg C yr
−
1
)
, BCC-CSM1(m)
(+2.1 Pg C yr
−
1
)
, and MPI-ESM-MR (+1.4 Pg C yr
−
1
)
. Mod-
ern net SOC flux had a much smaller range, between
−
0.5 Pg C yr
−
1
(CanESM2) and +1.9 Pg C yr
−
1
(MIROC-
ESM) with a multi-model mean of +0.7 Pg C yr
−
1
. The SOC
in six models switched from neutral or net sinks to net
sources of carbon (Table 2).
3.3
Changes in soil temperature and soil moisture
Predicted changes in soil temperature and moisture varied
widely across models, with consequences for SOC dynam-
ics. Across all models there was a warming trend over the
21st century in soil temperature (Table 2), with a multi-model
mean increase of 4.8
◦
C and a range between 3.1
◦
C (INM-
CM4) and 6.4
◦
C (HadGEM1-ES). Warming was most in-
tense in the high northern latitudes in most models (Fig. S9),
ranging between 3.1
◦
C (INM-CM4, BCC-CSM1.1(m)) and
8.0
◦
C (MIROC-ESM) when averaged across boreal and arc-
tic tundra biomes (Table S2). In most models, soil tempera-
tures increased in these biomes by a smaller amount than sur-
face air temperatures because loss of snow cover increased
winter heat fluxes to the atmosphere. Koven et al. (2013a)
show the importance of correctly simulating the relationship
between air and soil temperatures in this region.
The change in fractional soil water content over the en-
tire soil column of each model was even more variable
across models over the 21st century, with some of the
models becoming drier and others becoming wetter (Ta-
ble 2, Fig. S10) compared to starting distributions shown
in Fig. S11. However, the magnitude of change in global
mean soil water was relatively small, ranging from
−
0.046
Biogeosciences, 11, 2341–2356, 2014
www.biogeosciences.net/11/2341/2014/
Page 7
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
2347
Table 1. Change in soil carbon (Pg C) between 1997–2006 and 2090–2099 biome means; the change in the high northern latitude biomes
(tundra and boreal forests); the change in other biomes (tropical rainforest, temperate forest, desert and scrubland, grasslands and savanna,
and cropland and urban); and the change in global soil carbon stock across all ESMs. Multi-model means and standard deviations are given
in the last two columns.
CESM1(BGC)
NorESM1-ME
BNU-ESM
BCC-CSM1.1(m)
HadGEM2-ES
IPSL-CM5A-MR
GFDL-ESM2G
CanESM2
INM-CM4
MIR
OC-ESM
MPI-ESM-MR
Multi-model
mean
Multi-model
SD
Tundra
3
3
54
50
85
2
−
9
3
24
1
64
26
32
Boreal forest
−
1
−
3
18
47
61
−
5
−
28
−
6
22
−
22
62
13
32
Tropical rainforest
−
11
−
15
4
14
5
14
6
−
15
9
−
34
9
−
1
15
Temperate forest
−
2
−
4
0
7
11
0
2
−
1
0
−
5
6
1
5
Desert and shrubland
2
2
11
16
16
2
13
−
3
4
14
69
13
20
Grasslands and savanna
1
0
8
37
36
3
7
−
20
10
3
−
10
7
17
Cropland and urban
4
2
2
32
34
−
2
13
−
9
6
−
31
7
5
18
Permanent wetlands
0
0
0
0
2
0
−
2
0
0
−
2
3
0
1
Snow and ice
0
0
2
0
3
0
0
0
0
4
1
1
1
High northern latitude
2
0
72
97
146
−
2
−
37
−
3
46
−
21
126
39
63
Other
−
6
−
16
25
106
102
18
41
−
48
29
−
53
81
25
55
Total
−
4
−
16
99
203
253
15
1
−
51
76
−
72
211
65
113
to 0.007 (kg-water m
2
kg
−
1
max. water m
−
2
) across the 21st
century (Table 2).
3.4
Drivers of SOC change
Changes in global SOC were the net effect of large increases
in inputs and large decreases in turnover times (Fig. 5). For
all ESMs, most of the change in SOC inputs and turnover
is projected to occur in the 21st century, not over the histor-
ical period (1850–2006). In the constant turnover time sce-
nario (Fig. 5, green dashed lines), between 160 (NorESM1-
ME) and 1230 (MPI-ESM-MR) Pg C of SOC accumulated
from 1850 to 2100 across models due to increases in SOC
inputs associated with increased NPP. Conversely, in the con-
stant carbon input scenario (Fig. 5, red dotted lines), between
104 (MIROC-ESM) and 629 (MPI-ESM-MR) Pg C were lost
from global soil stocks as soils warmed and heterotrophic
respiration increased. For all models, the amount of carbon
potentially gained due to increasing inputs over time was
greater than the amount of carbon potentially lost through
increases in decomposition rates.
Between ESMs, the change in global SOC was signifi-
cantly explained by Eq. (2) using the relative changes in soil
inputs and decomposition rates, and the size of the initial
SOC pool (
R
2
=
0.89,
p <
0
.
01, Fig. 6a). With all models in-
cluded, the regression slope was 0.73, but it approached 1.0
when MIROC-ESM was removed. Explanatory power was
similarly high when SOC change was expressed proportion-
ally to the change in steady state (Eq. 1;
R
2
=
0.88,
p <
0
.
01,
Fig. S12a) or relative change in the inputs and decomposi-
tion rate (
R
2
=
0.83,
p <
0
.
01, Fig. S12b). The total change
in global SOC was not significantly correlated with the initial
SOC (
R
2
=
0.00,
p
=
0
.
86, Fig. 6b), nor the change decom-
position rate (
R
2
=
0.08,
p
=
0
.
39, Fig. 6c). Instead, much
of this change was attributed to the change in soil inputs
(
R
2
=
0.54,
p <
0
.
01, Fig. 6d), with the remaining variation
due to the interaction between initial SOC, changes in soil
inputs, and changes in decomposition rate.
Within ESMs, changes in SOC at each grid cell over the
21st century were not always consistent with Eq. (2). Ex-
planatory power was good for three models (0
.
88
> R
2
>
0
.
78,
p <
0
.
05), moderate for five models (0
.
50
> R
2
>
0
.
19,
p <
0
.
05), and not significant for three models. Nei-
ther the change in soil inputs nor the change in decomposi-
tion rate alone explained more than 20 % of the variation in
SOC changes within ESMs (Table S3).
3.5
Drivers of change in decomposition between models
Between-model variation in the change in global mean de-
composition rate was well explained by model parameteriza-
tion and environmental variables (Fig. 7). Together, the initial
decomposition rate, the temperature-dependency parameter
(
Q
10
)
of each ESM, and the ESM-simulated change in soil
temperature explained most of the variation between ESMs
as modeled by Eq. (7) (
R
2
=
0
.
80,
p <
0
.
01, Fig. 7a). The
change in decomposition rate was moderately explained by
the initial decomposition rate (
R
2
=
0
.
49,
p
=
0
.
02, Fig. 7b).
Other individual terms in Eq. (7), including variation in
Q
10
and ESM-simulated change in soil temperature, did not
www.biogeosciences.net/11/2341/2014/
Biogeosciences, 11, 2341–2356, 2014
Page 8
2348
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
Table 2. Starting value (1997–2006 mean), final value (2090–2099 mean), absolute change, and relative change of soil carbon, net primary
productivity (NPP), soil inputs, heterotrophic respiration, net SOC flux (positive values are carbon gains by the SOC pool, negative values
are carbon losses), area-weighted soil temperature, and area-weighted normalized soil water content. Soil water is mean gridded soil water
content divided by the grid maximum soil water content at the 10-year initial historical mean value.
CESM1(BGC)
NorESM1-ME
BNU-ESM
BCC-CSM1-1(m)
HadGEM2-ES
IPSL-CM5A-MR
GFDL-ESM2G
CanESM2
INM-CM4
MIR
OC-ESM
MPI-ESM-MR
Multi-model
mean
Multi-model
SD
SOC
2006
575
617
704
930
1121
1396
1411
1539
1683
2565
3101
1422
801
(Pg C)
2099
571
601
803
1134
1374
1412
1412
1488
1759
2493
3312
1487
816
Abs. Change
−
4
−
16
99
203
253
15
1
−
51
76
−
72
211
65
113
Rel. Change (%)
−
0.6
−
2.5
14.0
21.9
22.6
1.1
0.1
−
3.3
4.5
−
2.8
6.8
5.6
9.6
NPP
2006
46
47
50
58
75
84
78
65
68
63
93
66
15
(Pg C yr
−
1
)
2099
56
53
72
85
120
116
118
86
85
78
138
92
28
Abs. Change
10
5
22
27
45
31
40
21
17
15
46
25
14
Rel. Change (%)
22.1
11.0
43.8
46.9
59.3
37.0
50.4
32.4
25.1
23.7
49.1
36.5
14.8
SOC inputs
2006
43
44
49
57
75
69
65
65
67
57
77
61
12
(Pg C yr
−
1
)
2099
50
51
70
87
119
92
91
84
83
70
102
82
21
Abs. Change
7
7
21
29
44
23
26
20
16
13
25
21
11
Rel. Change (%)
17.0
15.2
44.0
51.6
59.3
32.8
39.8
30.3
24.7
22.8
32.4
33.6
14.0
Heterotrophic
2006
43
44
48
56
73
69
66
65
66
55
76
60
12
respiration
2099
50
51
70
85
116
93
92
85
83
73
101
82
20
(Pg C yr
−
1
)
Abs. Change
7
7
22
28
43
24
26
20
18
18
25
22
10
Rel. Change (%)
17.5
16.9
46.8
50.5
58.3
34.3
40.0
30.4
27.1
32.5
33.1
35.2
12.9
Net SOC flux
2006
0.0
0.3
0.8
0.9
1.4
0.5
−
0.4
−
0.5
1.0
1.9
1.5
0.7
0.8
(Pg C yr
−
1
)
2099
−
0.2
−
0.4
−
0.2
2.1
3.0
−
0.4
−
0.6
−
0.7
−
0.4
−
3.0
1.4
0.0
1.6
Abs. Change
−
0.2
−
0.7
−
1.0
1.2
1.6
−
0.9
−
0.3
−
0.2
−
1.3
−
4.9
−
0.1
−
0.6
1.7
Soil
2006
16.4
14.4
17.2
17.2
14.8
14.2
14.0
15.7
11.1
16.4
14.6
15.1
1.8
temperature
2099
20.2
18.5
21.5
20.9
21.3
20.3
17.8
21.8
14.2
22.6
19.7
19.9
2.4
(
◦
C)
Abs. Change
3.8
4.2
4.3
3.8
6.4
6.1
3.8
6.1
3.1
6.2
5.1
4.8
1.2
Soil water
2006
0.249
0.035
0.261
0.189
0.498
0.675
0.560
0.119
0.354
0.651
0.311
0.355
0.215
(kg m
2
water
2099
0.253
0.030
0.263
0.195
0.468
0.629
0.553
0.121
0.353
0.651
0.310
0.348
0.205
kg
−
1
m
−
2
Abs. Change
0.004
−
0.005
0.002
0.007
−
0.030
−
0.046
−
0.007
0.002
−
0.001
0.001
−
0.001
−
0.007
0.016
max. water)
explain any significant variation (Fig. 7c and d), nor did the
moisture term add any explanatory value.
4
Discussion
A key determinant of future climate-carbon cycle feedbacks
is whether increased inputs from NPP (Matthews et al., 2005)
or increased losses due to heterotrophic respiration (David-
son and Janssens, 2006) will dominate SOC responses to
global change. Our analysis shows that increases in both NPP
and heterotrophic respiration are large and relatively consis-
tent across ESMs. What varies substantially is the difference
between these two large fluxes under future conditions. Even
though all ESMs simulate 15 to 28 % declines in global SOC
turnover times (Fig. 1), many models (7 of 11) simulate in-
creases in global SOC over the 21st century. In these ESMs,
increased NPP more than offsets increased SOC decomposi-
tion. A major challenge moving forward will be to constrain
the uncertainty on these large fluxes enough to reliably esti-
mate the net change in SOC over time. This challenge also
applies to ESM vegetation pools where uncertainties in car-
bon turnover rates contribute to variation in vegetation car-
bon change across models (Friend et al., 2013).
4.1
SOC decomposition in ESMs
Given the importance of decomposition fluxes for SOC bal-
ance, we analyzed the factors underlying variation in decom-
position changes across ESMs. Cross-model differences in
decomposition change over the 21st century mainly depend
on the initial decomposition rate, temperature dependency
of decomposition, and the temperature change predicted by
the models (Fig. 7). Although these factors varied substan-
tially across ESMs, the range of variation is broadly consis-
tent with empirical observations. The intrinsic decomposi-
tion rate (
κ
; fitted at 15
◦
C and optimal soil moisture) varied
between 6 and 24 years (Table S4), a range consistent with
the parameterization of traditional biogeochemical models
(e.g., Parton et al., 1993).
Q
10
factors ranged between 1.4 and
Biogeosciences, 11, 2341–2356, 2014
www.biogeosciences.net/11/2341/2014/
Page 9
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
2349
Fig. 2. Absolute change in soil carbon density (kg m
−
2
) over the 21st century for all models (difference in 10-year means; 2090–2099 minus
1997–2006).
2.2 (Table S4), consistent with the observed distribution of
Q
10
values reported in the literature (Davidson and Janssens,
2006; Fierer et al., 2006; Mahecha et al., 2010). Modeled
temperature changes in the ESMs, especially in northern lat-
itudes (Koven et al., 2013a), are an additional source of un-
certainty that should be addressed in future studies (Knutti
and Sedláˇcek, 2012).
Moisture dependency did not provide any additional ex-
planation of variation in decomposition rate changes across
ESMs nor did it substantially improve reduced complexity
models of heterotrophic respiration within ESMs (Table S4),
suggesting that it is not a primary driver of SOC decompo-
sition change at the global scale. This result contrasts with
previous work showing that individual global climate mod-
els are sensitive to moisture function selection (Exbrayat et
al., 2013). Our analysis suggests that if ESMs are sensitive
to moisture functions, that dependency can be captured by
shifts in the intrinsic decomposition rate (
κ
) or
Q
10
parame-
ter. However, it is also likely that climate change will result
in small-scale variations in soil moisture that cancel out in
global-scale analyses, thereby obscuring moisture effects on
soil carbon.
4.2
NPP and SOC inputs in ESMs
All ESMs predicted substantial (11–59 %) increases in NPP
over the 21st century (Table 2). These increases are primar-
ily due to CO
2
fertilization effects, and secondarily due to
temperature and moisture change (Anav et al., 2013). Al-
though plausible based on FACE studies (Norby et al., 2010;
Piao et al., 2013), the increases in SOC inputs predicted by
ESMs may not occur for several reasons. First, FACE studies
have only been conducted in a limited number of ecosystems
that are not representative of the entire terrestrial biosphere.
FACE studies tend to be located in ecosystems with relatively
high CO
2
response, such as early- to mid-successional sys-
tems with high nutrient supply and rapidly growing vegeta-
tion (Körner, 2006; Norby and Zak, 2011). Ecosystems with
closed nutrient cycles and full canopy development occupy
the largest fraction of the global land surface, and current
studies show a more limited CO
2
fertilization response in
www.biogeosciences.net/11/2341/2014/
Biogeosciences, 11, 2341–2356, 2014
Page 10
2350
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
Fig. 3. Relative change in soil carbon density (%) over the 21st century for all models (difference in 10-year means divided by modern soil
carbon density; 2090–2099 minus 1997–2006 divided by 1997–2006).
these ecosystems (Bader et al., 2013). Second, nitrogen and
phosphorus are expected to become progressively more lim-
iting as biomass increases (Exbrayat et al., 2013), and nutri-
ents could be lost from ecosystems that experience increases
in precipitation and leaching.
On the other hand, some constraints on NPP could be
alleviated by other aspects of global change. For example,
increased nutrient mineralization under soil warming or in-
creased nutrient inputs from atmospheric deposition could
alleviate nutrient limitation (Bai et al., 2013; Rustad et al.,
2001). An amplifying seasonal cycle of CO
2
exchange in
the Northern Hemisphere suggests that boreal and temper-
ate forests are showing a stronger positive NPP response to
contemporary environmental change than assumed in ESMs
(Graven et al., 2013). However, some NPP constraints could
also be exacerbated; for instance, reduced precipitation or in-
creased evapotranspiration under climate change could lead
to increased moisture limitation of plant growth.
Even if the terrestrial biosphere does respond to elevated
CO
2
in line with FACE studies, it is unclear how long this re-
sponse can be maintained. In ESMs, the majority of predicted
change in NPP and SOC inputs occurs beyond 550 ppm
CO
2
, a range unexplored in field studies. Additionally, FACE
studies examine an unrealistic instantaneous change in CO
2
rather than the relatively gradual CO
2
change simulated in
ESMs. The constant turnover time scenarios in Fig. 5 suggest
that NPP sensitivity to global change (including CO
2
change)
is sustained throughout the 21st century in ESMs, well be-
yond the conditions simulated by most field experiments. Al-
though a sustained response cannot be ruled out based on cur-
rent experimental evidence, many photosynthetic processes
saturate at high CO
2
concentrations (Franks et al., 2013;
Körner, 2006). If NPP were to increase by nearly 50 % over
the 21st century, as predicted by some ESMs, negative feed-
backs on NPP would be expected at the ecosystem scale as
previously discussed.
Biogeosciences, 11, 2341–2356, 2014
www.biogeosciences.net/11/2341/2014/
Page 11
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
2351
!
30!
!
!
Abs.!
Δ
C
soil
!
[Pg!C]!
NPP!
[Pg!C!yr
31
]!
C
soil
!
[Pg
!
C]!
CESM1(BGC)!
34!
46!
575!
NorESM13ME!
316!
47!
617!
BNU3ESM!
99!
50!
704!
BCC3CSM1.1(m)!
203!
58!
930!
HadGEM23ES!
253!
75!
1121!
IPSL3CM5A3MR!
15!
84!
1396!
GFDL3ESM2G!
1!
78!
1411!
CanESM2!
351!
65!
1539!
INM3CM4!
76!
68!
1683!
MIROC3ESM!
372!
63!
2565!
MPI3ESM3MR!
211!
93!
3101!
Benchmark!range!
!
46373
!
89031660
!
Figure!4:!Metrics!of!ESM!performance!and!benchmarks!for!soil!C!change!(difference!in!103
year!means,!209032099!minus!199732006),!net!primary!production!(NPP;!103year!mean!19973
2006),!and!modern!soil!carbon!pools!(103year!mean!199732006).!Benchmark!data!are!shown!in!
the!bottom!row.!Blue!cells!indicate!ESM!outputs!that!fall!below!the!benchmarks,!yellow!cells!are!
consistent!with!benchmarks,!and!red!cells!are!above!the!benchmarks.!NPP!data!are!from!Ito!
(2011);!modern!soil!carbon!stocks!are!from!Todd3Brown!
et#al.
!(2013)!and!originally!from!
(FAO/IIASA/ISRIC/ISSCAS/JRC,!2012).
Fig. 4. Metrics of ESM performance and benchmarks for soil
C change (difference in 10-year means, 2090–2099 minus 1997–
2006), net primary production (NPP; 10-year mean 1997–2006),
and modern soil carbon pools (10-year mean 1997–2006). Bench-
mark data are shown in the bottom row. Blue cells indicate
ESM outputs that fall below the benchmarks, yellow cells are
consistent with benchmarks, and red cells are above the bench-
marks. NPP data are from Ito (2011); modern soil carbon
stocks are from Todd-Brown et al. (2013) and originally from
(FAO/IIASA/ISRIC/ISSCAS/JRC, 2012).
In addition to potentially overestimating the 21st century
NPP response, ESMs likely overestimate the increase of SOC
pools in response to increased carbon inputs. We found a
strong relationship between the change in carbon inputs to
soil and the change in ESM SOC (Fig. 6d). Even stronger re-
lationships have been established between the spatial pattern
of contemporary NPP and SOC in ESMs (Todd-Brown et al.,
2013). However, there is empirical and theoretical evidence
that increases in soil inputs, especially under elevated CO
2
,
may have little effect on SOC stocks (Bader et al., 2013;
Norby and Zak, 2011). CO
2
fertilization often increases the
labile fractions of plant litter and root exudation (Phillips et
al., 2011; Schlesinger and Lichter, 2001). These inputs enter
fast turnover carbon pools and are hypothesized to facilitate
the decomposition of slow turnover pools through priming
(Neill and Gignoux, 2006). All of these ESMs contain first-
order linear decomposition models of SOC and are thus un-
able to represent priming mechanisms (Wutzler and Reich-
stein, 2008). In addition, the availability of mineral surfaces
(Six et al., 2002) may constrain the amount of SOC that can
be stored in a given soil. These effects will likely weaken the
coupling between NPP and SOC changes.
Finally, land use change can significantly affect NPP and
land carbon storage. In one recent analysis with a subset of
CMIP5 models, land use change resulted in terrestrial carbon
losses of 25 to 205 Pg C compared with no land use change
(Brovkin et al., 2013). This change is on the same order of
magnitude as the 21st century change in soil carbon simu-
lated by many ESMs. Therefore land use change is a critical
component of the terrestrial carbon cycle and is likely the
CESM1(BGC)
0
500
1000
1500
2000
2500
Constant turnover time
Constant carbon inputs
ESM soil carbon
−2
+167
−121
NorESM1−ME
−13
+160
−129
BNU−ESM
+151
+684
−204
BCC−CSM1.1(m)
0
500
1000
1500
2000
2500
+261
+696
−260
HadGEM2−ES
+361
+1041
−335
IPSL−CM5A−MR
+25
+620
−367
GFDL−ESM2G
0
500
1000
1500
2000
2500
−18
+475
−353
CanESM2
−20
+589
−418
INM−CM4
1850
1950
2050
2150
+134
+520
−310
MIROC−ESM
1850
1950
2050
2150
2000
2500
3000
3500
4000
4500
−8
+572
−104
MPI−ESM−MR
1850
1950
2050
2150
+331
+1230
−629
Soil carbon [Pg C]
Fig. 5. Soil carbon over time (historical and RCP 8.5) under the
“constant turnover time” and “constant carbon inputs” scenarios.
The black solid line is the ESM-simulated global soil carbon stock
with the net change from 1850 indicated in black text. The green
dashed line is the soil carbon stock that would be predicted if
turnover time were held constant at the 1850 value and soil in-
puts were allowed to evolve as predicted by the ESM, with the net
change indicated in green text. The red dotted line is the soil carbon
stock that would be predicted if carbon inputs were held constant
at 1850 levels and the turnover time were allowed to evolve as pre-
dicted by the ESM, with the net change in red text. The vertical grey
line at 2006 indicates where the historical experiment ends and the
RCP 8.5 experiment begins.
driver of some of the variation in SOC not explained by pa-
rameterization and environmental variables in our analysis.
4.3
Assessing reliability of SOC predictions
One possible approach for narrowing the likely range of
SOC predictions is to select estimates of SOC change from
ESMs that are consistent with global benchmark data sets
(Fig. 4). Two models fell within empirical estimates (BCC-
CSM-1.1(m) and CanESM2), and two additional models
matched one empirical estimate and were close to the second
(HadGEM2-ES and INM-CM4). Models consistent with the
benchmarks did not have a markedly narrower range of sim-
ulated change in SOC; CanESM2 showed losses of 51 Pg C,
and BCC-CSM1.1(m) showed gains of 203 Pg C. This wide
range suggests that contemporary NPP and SOC benchmarks
are not strong constraints on ESM simulations of future soil
www.biogeosciences.net/11/2341/2014/
Biogeosciences, 11, 2341–2356, 2014
Page 12
2352
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
a
b
c
d
e
fg
h
i
j
k
−200
0
200
−100
0
100
200
300
y
=
0.73
x
+
30
,
R
2
=
0.89
(
p
<
0.01
)
Predicted change in soil carbon [Eq. 2; Pg C]
Change in soil carbon [Pg C]
a)
ab
c
d
e
fg
h
i
j
k
0
500
1500
2500
3500
−100
0
100
200
300
y
=
0.0083
x
+
50
,
R
2
=
0
(
p
=
0.86
)
Starting soil carbon [C
start
; Pg C]
Change in soil carbon [Pg C]
b)
a
b
c
d
e
f
g
h
i
j
k
0.70
0.75
0.80
0.85
−100
0
100
200
300
y
=
740
x
−
520
,
R
2
=
0.08
(
p
=
0.39
)
Relative change in decomposition rate
Change in soil carbon [Pg C]
c)
a
b
c
d
e
f
g
h
i
j
k
1.1
1.2
1.3
1.4
1.5
1.6
1.7
−100
0
100
200
300
y
=
590
x
−
720
,
R
2
=
0.54
(
p
<
0.01
)
Relative change in soil inputs
Change in soil carbon [Pg C]
d)
Fig. 6. Change in global soil organic carbon between 1997–2006
and 2090–2099 global means as a function of (a) Eq. (2), the rel-
ative change in input and decomposition rate times the initial soil
carbon stock,
1
+
1I
I
start
1
+
1k
k
start
−
1
×
C
start
, (b) starting soil carbon stock,
C
start
, (c) the inverse of the relative change in decomposition rate
(
k
),
1
1
+
1k
k
start
, and (d) relative change in soil inputs (
I
), 1
+
1I
I
start
. The
models are represented as follows; a: CESM1(BGC), b: NorESM1-
ME, c: BNU-ESM, d: BCC-CSM1.1(m), e: HadGEM2-ES, f: IPSL-
CM5A-MR, g: GFDL-ESM2G, h: CanESM2, i: INM-CM4, j:
MIROC-ESM, and k: MPI-ESM-MR.
carbon change. To have greater confidence in model projec-
tions, ESMs should probably match not only global bench-
marks but also observed spatial patterns in NPP and SOC at
biome- or grid scales (Todd-Brown et al., 2013).
The spatial pattern of SOC change provides another cri-
terion for evaluating the likelihood of ESM predictions. All
ESMs that predicted large SOC gains globally (i.e., greater
than 75 Pg C) also showed large gains in tundra and boreal
biomes (greater than 46 Pg C; Table 1). These models in-
clude BCC-CSM1.1(m) and INM-CM4, which were among
the three models most consistent with benchmarking data.
In contrast to these model predictions, empirical studies sug-
gest that high-latitude soils are unlikely to serve as a long-
term carbon sink (Schuur et al., 2009; Sistla et al., 2013).
Even if tundra and boreal NPP increases substantially, SOC
storage will likely be constrained by permafrost thawing, in-
creases in fire frequency and severity, and high vulnerability
of old SOC to decomposition under global change (Flannigan
et al., 2009; Kasischke and Turetsky, 2006; Mack et al., 2004;
Turetsky et al., 2011). Thus all of the ESMs predicting large
gains in global SOC appear to depend on high levels of car-
bon storage in arctic ecosystems that may be more than offset
ab
c
d
e
f
g
h
i j
k
0.005
0.015
0.025
0.005
0.010
0.015
0.020
y
=
0.81
x
+
0.003
,
R
2
=
0.8
(
p
<
0.01
)
Starting decomposition modified by
temperature dependency and change [yr
−
1
]
Change in decomposition [yr
−
1
]
a)
k
start
ln
(
Q
10
)
∆
T
10
a
b
c
d
e
f
g
h
i
j
k
0.02
0.04
0.06
0.08
0.005
0.010
0.015
0.020
y
=
0.18
x
+
0.005
,
R
2
=
0.49
(
p
=
0.02
)
Starting decomposition rate [k
start
; yr
−
1
]
Change in decomposition [yr
−
1
]
b)
a
b
c
d
e
f
g
h
i j
k
1.4
1.6
1.8
2.0
2.2
2.4
0.005
0.010
0.015
0.020
y
=
0.0019
x
+
0.01
,
R
2
=
0.02
(
p
=
0.72
)
Temperature dependency [Q
10
]
Change in decomposition [yr
−
1
]
c)
a b
c
d
e
f
g
h
i
j
k
3
4
5
6
7
0.005
0.010
0.015
0.020
y
=
0.00025
x
+
0.013
,
R
2
=
0
(
p
=
0.85
)
Temperature change [
∆
T;
°
C]
Change in decomposition [yr
−
1
]
d)
Fig. 7. Change in decomposition rate between 1997–2006 and
2009–2099 global means due to variation in (a) starting decom-
position rate,
Q
10
, and soil temperature change from Eq. (7),
k
start
ln
(Q
10
)
1T
10
, (b) starting decomposition rate,
k
start
, (c) tem-
perature dependency,
Q
10
, and (d) change in soil temperature,
1T
. The models are represented as follows; a: CESM1(BGC), b:
NorESM1-ME, c: BNU-ESM, d: BCC-CSM1.1(m), e: HadGEM2-
ES, f: IPSL-CM5A-MR, g: GFDL-ESM2G, h: CanESM2, i: INM-
CM4, j: MIROC-ESM, and k: MPI-ESM-MR.
by the release of permafrost carbon (Burke et al., 2012). This
bias likely arises from the lack of permafrost carbon dynam-
ics in current CMIP5 ESMs. Although some ESMs represent
freeze–thaw dynamics, this broad-scale representation does
not capture permafrost dynamics that include highly local-
ized changes in hydrological conditions. ESMs that do not
predict SOC accumulation in the arctic tend to predict SOC
losses or much smaller gains at the global scale.
5
Conclusions
Our work shows that many current ESMs project substantial
carbon sequestration potential in global soils over the 21st
century. Although decomposition rates increase with climate
warming, this effect is largely offset by CO
2
-driven increases
in NPP and soil inputs. Unrecognized constraints on future
NPP response or limits on the conversion of plant inputs to
SOC could therefore substantially reduce soil carbon stor-
age. In particular, constraints on NPP are uncertain because
ecosystem-scale CO
2
manipulations have not been carried
out above 550 ppm, a value that may be exceeded by mid-
century. As a result, our analysis calls into question the cur-
rent majority of ESMs that show considerable SOC accumu-
lation in the face of 21st century global change. Although a
Biogeosciences, 11, 2341–2356, 2014
www.biogeosciences.net/11/2341/2014/
Page 13
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
2353
few ESMs are consistent with modern benchmark data sets,
we conclude that there is considerable uncertainty surround-
ing future soil carbon storage in soils related to processes that
are not represented within current ESMs.
New modeling efforts should consider quantifying con-
straints on the NPP response to global change and altering
SOC sub-models to represent mechanisms responsible for
SOC response to NPP change. For example, incorporating
nutrient dynamics into ESMs could help constrain the NPP
response to CO
2
fertilization (Exbrayat et al., 2013; Piao
et al., 2013; Thornton et al., 2009). Decomposition model
structures could be updated to account for microbial prim-
ing effects (Wieder et al., 2013) and mineral–SOC interac-
tions (Six et al., 2000, 2002; Torn et al., 1997; Wiesmeier
et al., 2014). Biogeochemical responses in ESMs should be
validated against global data sets (e.g., Luo et al., 2012;
Todd-Brown et al., 2013) and empirical results from global
change experiments across diverse ecosystems (Norby and
Zak, 2011; Rustad et al., 2001). In addition to soil carbon
stock and respiration measurements, isotopic data represent
an underutilized constraint on the residence times of different
soil carbon pools (Derrien and Amelung, 2011; Trumbore,
2009). These model development and analysis efforts will
help ensure that policymakers can develop mitigation strate-
gies for global change based on accurate projections of the
global carbon cycle.
Supplementary material related to this article is
available online at http://www.biogeosciences.net/11/
2341/2014/bg-11-2341-2014-supplement.pdf.
Acknowledgements. We thank the US National Science Foundation
(Advancing Theory in Biology and Decadal and Regional Climate
Prediction using Earth System Models (EaSM) programs) and
the US Department of Energy Office of Science Biological and
Environmental Research (BER) for funding this research. We
also acknowledge the World Climate Research Programme’s
Working Group on Coupled Modeling, which is responsible for
CMIP, and we thank the model groups (listed in Table S1 of this
paper) for producing and making available their model output. We
thank the US Department of Energy’s Program for Climate model
Diagnosis and Intercomparison for providing coordinating support
and leading development of software infrastructure in partnership
with Global Organization for Earth System Science Portals. We
would also like to thank Christian Reick for his contribution of the
MPI-ESM simulations and manuscript review. Chris Jones was
supported by the Joint UK DECC/Defra Met Office Hadley Centre
Climate Programme (GA01101). Jerry Tjiputra also acknowledges
the Biofeedback project of the Centre for Climate Dynamics.
Edited by: V. Brovkin
References
Anav, A., Friedlingstein, P., Kidston, M., Bopp, L., Ciais, P., Cox,
P., Jones, C., Jung, M., Myneni, R., and Zhu, Z.: Evaluating
the land and ocean components of the global carbon cycle in
the CMIP5 Earth system models, J. Climate, 26, 6801–6843,
doi:10.1175/JCLI-D-12-00417.1, 2013.
Arora, V. K., Boer, G. J., Friedlingstein, P., Eby, M., Jones, C.
D., Christian, J. R., Bonan, G., Bopp, L., Brovkin, V., Cad-
ule, P., Hajima, T., Ilyina, T., Lindsay, K., Tjiputra, J. F., and
Wu, T.: Carbon–concentration and carbon–climate feedbacks
in CMIP5 Earth system models, J. Climate, 26, 5289–5314,
doi:10.1175/JCLI-D-12-00494.1, 2013.
Bader, M. K.-F., Leuzinger, S., Keel, S. G., Siegwolf, R. T. W.,
Hagedorn, F., Schleppi, P., and Körner, C.: Central European
hardwood trees in a high-CO
2
future: synthesis of an 8-year for-
est canopy CO
2
enrichment project, J. Ecol., 101, 1509–1519,
doi:10.1111/1365-2745.12149, 2013.
Bai, E., Li, S., Xu, W., Li, W., Dai, W., and Jiang, P.: A
meta-analysis of experimental warming effects on terrestrial
nitrogen pools and dynamics., New Phytol., 19, 431–440,
doi:10.1111/nph.12252, 2013.
Billings, S. A., Lichter, J., Ziegler, S. E., Hungate, B. A., and
Richter, D. de B.: A call to investigate drivers of soil organic
matter retention vs. mineralization in a high CO
2
world, Soil
Biol. Biochem., 42, 665–668, doi:10.1016/j.soilbio.2010.01.002,
2010.
Brovkin, V., Boysen, L., Arora, V. K., Boisier, J. P., Cadule, P.,
Chini, L., Claussen, M., Friedlingstein, P., Gayler, V., van den
Hurk, B. J. J. M., Hurtt, G. C., Jones, C. D., Kato, E., de Noblet-
Ducoudré, N., Pacifico, F., Pongratz, J., and Weiss, M.: Effect
of anthropogenic land-use and land-cover changes on climate
and land carbon storage in CMIP5 projections for the twenty-
first century, J. Climate, 26, 6859–6881, doi:10.1175/JCLI-D-12-
00623.1, 2013.
Burke, E. J., Jones, C. D., and Koven, C. D.: Estimating the
permafrost-carbon climate response in the CMIP5 climate mod-
els using a simplified approach, J. Climate, 26, 4897–4909,
doi:10.1175/JCLI-D-12-00550.1, 2012.
Carney, K. M., Hungate, B. A., Drake, B. G., and Megonigal,
J. P.: Altered soil microbial community at elevated CO
2
leads
to loss of soil carbon, P. Natl. Acad. Sci., 104, 4990–4995,
doi:10.1073/pnas.0610045104, 2007.
Chapin, F. S. and Eviner, V. T.: Biogeochemistry of terrestrial net
primary production, in: Theatise on Geochemistry, vol. 8, edited
by: Holland, H. D. and Turekian, K. K., 1–35, Pergamon, Ox-
ford., 2007.
Cheng, W., Parton, W. J., Gonzalez-Meler, M. A., Phillips, R., Asao,
S., McNickle, G. G., Brzostek, E., and Jastrow, J. D.: Synthesis
and modeling perspectives of rhizosphere priming, New Phytol.,
201, 31–44, doi:10.1111/nph.12440, 2014.
Ciais, P., Sabine, C., Bala, G., Bopp, L., Brovkin, V., Canadell, J.,
Chhabra, A., DeFries, R., Galloway, J., Heimann, M., Jones, C.,
Le Quéré, C., Myneni, R. B., Piao, S., and Thornton, P.: Carbon
and Other Biogeochemical Cycles, in Climate Change 2013: The
Physical Science Basis. Contribution of Working Group I to the
Fifth Assessment Report of the Intergovernmental Panel on Cli-
mate Change, edited by: T. F. Stocker, Qin, D., Plattner, G.-K.,
Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex,
www.biogeosciences.net/11/2341/2014/
Biogeosciences, 11, 2341–2356, 2014
Page 14
2354
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
V., and Midgley, P. M., p. 1535, Cambridge University Press,
Cambridge, United Kingdom and New York, NY, USA, 2013.
Cook, F. J. and Orchard, V. A.: Relationships between soil respi-
ration and soil moisture, Soil Biol. Biochem., 40, 1013–1018,
doi:10.1016/j.soilbio.2007.12.012, 2008.
Cox, P. M., Pearson, D., Booth, B. B., Friedlingstein, P., Hunting-
ford, C., Jones, C. D., and Luke, C. M.: Sensitivity of tropical
carbon to climate change constrained by carbon dioxide variabil-
ity, Nature, 494, 341–344, doi:10.1038/nature11882, 2013.
Davidson, E. A. and Janssens, I. A.: Temperature sensitivity of soil
carbon decomposition and feedbacks to climate change, Nature,
440, 165–173, doi:10.1038/nature04514, 2006.
De Graaff, M.-A., Van Groenigen, K.-J., Six, J., Hungate, B., and
Van Kessel, C.: Interactions between plant growth and soil nutri-
ent cycling under elevated CO
2
: a meta-analysis, Glob. Change
Biol., 12, 2077–2091, doi:10.1111/j.1365-2486.2006.01240.x,
2006.
Derrien, D. and Amelung, W.: Computing the mean residence
time of soil carbon fractions using stable isotopes: impacts
of the model framework, Eur. J. Soil Sci., 62, 237–252,
doi:10.1111/j.1365-2389.2010.01333.x, 2011.
Dungait, J. A. J., Hopkins, D. W., Gregory, A. S., and Whit-
more, A. P.: Soil organic matter turnover is governed by acces-
sibility not recalcitrance, Glob. Change Biol., 18, 1781–1796,
doi:10.1111/j.1365-2486.2012.02665.x, 2012.
Exbrayat, J.-F., Pitman, A. J., Zhang, Q., Abramowitz, G., and
Wang, Y.-P.: Examining soil carbon uncertainty in a global
model: response of microbial decomposition to temperature,
moisture and nutrient limitation, Biogeosciences, 10, 7095–
7108, doi:10.5194/bg-10-7095-2013, 2013.
Falloon, P., Jones, C. D., Ades, M., and Paul, K.: Direct soil mois-
ture controls of future global soil carbon changes: An impor-
tant source of uncertainty, Global Biogeochem. Cy., 25, GB3010,
doi:10.1029/2010GB003938, 2011.
FAO/IIASA/ISRIC/ISSCAS/JRC:
Harmonized
World
Soil
Database (version 1.10), FAO, Rome, Italy and IIASA,
Laxenburg, Austria, 2012.
Fierer, N., Colman, B. P., Schimel, J. P., and Jackson, R. B.: Pre-
dicting the temperature dependence of microbial respiration in
soil: A continental-scale analysis, Global Biogeochem. Cy., 20,
GB3026, doi:10.1029/2005GB002644, 2006.
Flannigan, M. D., Krawchuk, M. A., de Groot, W. J., Wotton, B. M.,
and Gowman, L. M.: Implications of changing climate for global
wildland fire, Int. J. Wildl. Fire, 18, 483, doi:10.1071/WF08187,
2009.
Fontaine, S., Bardoux, G., Abbadie, L., and Mariotti, A.: Carbon in-
put to soil may decrease soil carbon content, Ecol. Lett., 7, 314–
320, doi:10.1111/j.1461-0248.2004.00579.x, 2004.
Franks, P. J., Adams, M. A., Amthor, J. S., Barbour, M. M., Berry, J.
A., Ellsworth, D. S., Farquhar, G. D., Ghannoum, O., Lloyd, J.,
McDowell, N., Norby, R. J., Tissue, D. T., and von Caemmerer,
S.: Sensitivity of plants to changing atmospheric CO
2
concentra-
tion: from the geological past to the next century, New Phytol.,
197, 1077–1094, doi:10.1111/nph.12104, 2013.
Friedl, M. A., Sulla-Menashe, D., Tan, B., Schneider, A., Ra-
mankutty, N., Sibley, A., and Huang, X.: MODIS Collection
5 global land cover: Algorithm refinements and characteriza-
tion of new datasets, Remote Sens. Environ., 114, 168–182,
doi:10.1016/j.rse.2009.08.016, 2010.
Friedlingstein, P., Cox, P., Betts, R., Bopp, L., von Bloh, W.,
Brovkin, V., Cadule, P., Doney, S., Eby, M., Fung, I., Bala, G.,
John, J., Jones, C., Joos, F., Kato, T., Kawamiya, M., Knorr, W.,
Lindsay, K., Matthews, H. D., Raddatz, T., Rayner, P., Reick,
C., Roeckner, E., Schnitzler, K.-G., Schnur, R., Strassmann, K.,
Weaver, A. J., Yoshikawa, C., and Zeng, N.: Climate–carbon cy-
cle feedback analysis: Results from the C
4
MIP model Intercom-
parison, J. Climate, 19, 3337–3353, doi:10.1175/JCLI3800.1,
2006.
Friend, A. D., Lucht, W., Rademacher, T. T., Keribin, R., Betts, R.,
Cadule, P., Ciais, P., Clark, D. B., Dankers, R., Falloon, P. D., Ito,
A., Kahana, R., Kleidon, A., Lomas, M. R., Nishina, K., Ostberg,
S., Pavlick, R., Peylin, P., Schaphoff, S., Vuichard, N., Warsza-
wski, L., Wiltshire, A., and Woodward, F. I.: Carbon residence
time dominates uncertainty in terrestrial vegetation responses to
future climate and atmospheric CO
2
, P. Natl. Acad. Sci., 111,
3280–3285, doi:10.1073/pnas.1222477110, 2014.
Giorgi, F.: Climate change hot-spots, Geophys. Res. Lett., 33,
L08707, doi:10.1029/2006GL025734, 2006.
Graven, H. D., Keeling, R. F., Piper, S. C., Patra, P. K., Stephens, B.
B., Wofsy, S. C., Welp, L. R., Sweeney, C., Tans, P. P., Kelley, J.
J., Daube, B. C., Kort, E. A., Santoni, G. W., and Bent, J. D.: En-
hanced seasonal exchange of CO
2
by northern ecosystems since
1960, Science, 341, 1085–1089, doi:10.1126/science.1239207,
2013.
Hoffman, F. M., Randerson, J. T., Arora, V. K., Bao, Q., Cadule, P.,
Ji, D., Jones, C. D., Kawamiya, M., Khatiwala, S., Lindsay, K.,
Obata, A., Shevliakova, E., Six, K. D., Tjiputra, J. F., Volodin,
E. M., and Wu, T.: Causes and implications of persistent atmo-
spheric carbon dioxide biases in Earth System Models, J. Geo-
phys. Res.-Biogeo., 119, 141–162, doi:10.1002/2013JG002381,
2014.
Hofmockel, K. S., Zak, D. R., Moran, K. K., and Jastrow, J. D.:
Changes in forest soil organic matter pools after a decade of
elevated CO
2
and O
3
, Soil Biol. Biochem., 43, 1518–1527,
doi:10.1016/j.soilbio.2011.03.030, 2011.
Hoosbeek, M. and Scarascia-Mugnozza, G.: Increased litter build
up and soil organic matter stabilization in a Poplar plantation
after 6 years of atmospheric CO
2
enrichment (FACE): Final
results of POP-EuroFACE compared to other forest FACE ex-
periments, Ecosystems, 12, 220–239, doi:10.1007/s10021-008-
9219-z, 2009.
Houghton, R. A.: Why are estimates of the terrestrial car-
bon balance so different?, Glob. Change Biol., 9, 500–509,
doi:10.1046/j.1365-2486.2003.00620.x, 2003.
Hungate, B. A., Dijkstra, P., Wu, Z., Duval, B. D., Day, F. P., John-
son, D. W., Megonigal, J. P., Brown, A. L. P., and Garland, J. L.:
Cumulative response of ecosystem carbon and nitrogen stocks
to chronic CO
2
exposure in a subtropical oak woodland., New
Phytol., 200, 753–766, doi:10.1111/nph.12333, 2013.
Hungate, B. A., Van Groenigen, K.-J., Six, J., Jastrow, J. D., Luo, Y.,
De Graaff, M.-A., Van Kessel, C., and Osenberg, C. W.: Assess-
ing the effect of elevated carbon dioxide on soil carbon: a com-
parison of four meta-analyses, Glob. Change Biol., 15, 2020–
2034, doi:10.1111/j.1365-2486.2009.01866.x, 2009.
Ise, T. and Moorcroft, P. R.: The global-scale temperature and mois-
ture dependencies of soil organic carbon decomposition: an anal-
ysis using a mechanistic decomposition model, Biogeochemistry,
80, 217–231, doi:10.1007/s10533-006-9019-5, 2006.
Biogeosciences, 11, 2341–2356, 2014
www.biogeosciences.net/11/2341/2014/
Page 15
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
2355
Ito, A.: A historical meta-analysis of global terrestrial net primary
productivity: are estimates converging?, Glob. Change Biol., 17,
3161–3175, doi:10.1111/j.1365-2486.2011.02450.x, 2011.
Jobbagy, E. G. and Jackson, R. B.: The vertical distribution of soil
organic carbon and its relation to climate and vegetation, Ecol.
Appl., 10, 423–436, 2000.
Kasischke, E. S. and Turetsky, M. R.: Recent changes in the fire
regime across the North American boreal region – Spatial and
temporal patterns of burning across Canada and Alaska, Geo-
phys. Res. Lett., 33, L09703, doi:10.1029/2006GL025677, 2006.
Knutti, R. and Sedláˇcek, J.: Robustness and uncertainties in the new
CMIP5 climate model projections, Nat. Clim. Change, 3, 369–
373, doi:10.1038/nclimate1716, 2012.
Körner, C.: Plant CO
2
responses: an issue of definition, time and re-
source supply, New Phytol., 172, 393–411, doi:10.1111/j.1469-
8137.2006.01886.x, 2006.
Koven, C. D., Riley, W. J., and Stern, A.: Analysis of per-
mafrost thermal dynamics and response to climate change in
the CMIP5 Earth system models, J. Climate, 26, 1877–1900,
doi:10.1175/JCLI-D-12-00228.1, 2013a.
Koven, C. D., Riley, W. J., Subin, Z. M., Tang, J. Y., Torn, M. S.,
Collins, W. D., Bonan, G. B., Lawrence, D. M., and Swenson,
S. C.: The effect of vertically resolved soil biogeochemistry and
alternate soil C and N models on C dynamics of CLM4, Biogeo-
sciences, 10, 7109–7131, doi:10.5194/bg-10-7109-2013, 2013b.
Koven, C. D., Ringeval, B., Friedlingstein, P., Ciais, P., Cadule, P.,
Khvorostyanov, D., Krinner, G., and Tarnocai, C.: Permafrost
carbon-climate feedbacks accelerate global warming, P. Natl.
Acad. Sci., 108, 14769–14774, doi:10.1073/pnas.1103910108,
2011.
Kuzyakov, Y., Friedel, J. K., and Stahr, K.: Review of mechanisms
and quantification of priming effects, Soil Biol. Biochem., 32,
1485–1498, doi:10.1016/S0038-0717(00)00084-5, 2000.
Le Quéré, C., Raupach, M. R., Canadell, J. G., Marland, G., Bopp,
L., Ciais, P., Conway, T. J., Doney, S. C., Feely, R. A., Foster,
P., Friedlingstein, P., Gurney, K., Houghton, R. A., House, J. I.,
Huntingford, C., Levy, P. E., Lomas, M. R., Majkut, J., Metzl,
N., Ometto, J. P., Peters, G. P., Prentice, I. C., Randerson, J. T.,
Running, S. W., Sarmiento, J. L., U., S., Sitch, S., Takahashi, T.,
Viovy, N., van der Werf, G. R., and Woodward, F. I.: Trends in
the sources and sinks of carbon dioxide, Nat. Geosci., 2, 831–
836, doi:10.1038/ngeo689, 2009.
Lloyd, J. and Taylor, J. A.: On the temperature dependence of soil
respiration, Funct. Ecol., 8, 315–323, 1994.
Luo, Y. Q., Randerson, J. T., Abramowitz, G., Bacour, C., Blyth,
E., Carvalhais, N., Ciais, P., Dalmonech, D., Fisher, J. B., Fisher,
R., Friedlingstein, P., Hibbard, K., Hoffman, F., Huntzinger, D.,
Jones, C. D., Koven, C., Lawrence, D., Li, D. J., Mahecha, M.,
Niu, S. L., Norby, R., Piao, S. L., Qi, X., Peylin, P., Prentice, I.
C., Riley, W., Reichstein, M., Schwalm, C., Wang, Y. P., Xia, J.
Y., Zaehle, S., and Zhou, X. H.: A framework for benchmarking
land models, Biogeosciences, 9, 3857–3874, doi:10.5194/bg-9-
3857-2012, 2012.
Luo, Y., Su, B., Currie, W. S., Dukes, J. S., Finzi, A., Hartwig,
U., Hungate, B., Mc Murtrie, R. E., Oren, R., Parton, W. J.,
Pataki, D. E., Shaw, M. R., Zak, D. R., and Field, C. B.: Progres-
sive nitrogen limitation of ecosystem responses to rising atmo-
spheric carbon dioxide, Bioscience, 54, 731, doi:10.1641/0006-
3568(2004)054[0731:PNLOER]2.0.CO;2, 2004.
Mack, M. C., Schuur, E. A. G., Bret-Harte, M. S., Shaver, G. R., and
Chapin, F. S.: Ecosystem carbon storage in arctic tundra reduced
by long-term nutrient fertilization, Nature, 431, 440–443, 2004.
Mahecha, M. D., Reichstein, M., Carvalhais, N., Lasslop, G.,
Lange, H., Seneviratne, S. I., Vargas, R., Ammann, C., Arain, M.
A., Cescatti, A., Janssens, I. A., Migliavacca, M., Montagnani,
L., and Richardson, A. D.: Global convergence in the tempera-
ture sensitivity of respiration at ecosystem level, Science, 329,
838–840, doi:10.1126/science.1189587, 2010.
Matthews, H. D., Eby, M., Weaver, A. J., and Hawkins,
B. J.: Primary productivity control of simulated carbon
cycle-climate feedbacks, Geophys. Res. Lett., 32, L14708,
doi:10.1029/2005GL022941, 2005.
NASA Land Processes Distributed Active Archive Center (LP
DAAC): Land Cover Type Yearly L3 Global 0.05Deg CMG
(MCD12C1), USGS/Earth Resour. Obs. Sci. Center, Sioux Falls,
South Dakota, available at: https://lpdaac.usgs.gov/products/
modis_products_table/mcd12c1, 2008.
Neill, C. and Gignoux, J.: Soil organic matter decomposition
driven by microbial growth: A simple model for a complex
network of interactions, Soil Biol. Biochem., 38, 803–811,
doi:10.1016/j.soilbio.2005.07.007, 2006.
Norby, R. J., DeLucia, E. H., Gielen, B., Calfapietra, C., Giar-
dina, C. P., King, J. S., Ledford, J., McCarthy, H. R., Moore,
D. J. P., Ceulemans, R., De Angelis, P., Finzi, A. C., Karnosky,
D. F., Kubiske, M. E., Lukac, M., Pregitzer, K. S., Scarascia-
Mugnozza, G. E., Schlesinger, W. H., and Oren, R.: For-
est response to elevated CO
2
is conserved across a broad
range of productivity, P. Natl. Acad. Sci., 102, 18052–18056,
doi:10.1073/pnas.0509478102, 2005.
Norby, R. J., Warren, J. M., Iversen, C. M., Medlyn, B. E., and
McMurtrie, R. E.: CO
2
enhancement of forest productivity con-
strained by limited nitrogen availability, P. Natl. Acad. Sci., 107,
19368–19373, doi:10.1073/pnas.1006463107, 2010.
Norby, R. J. and Zak, D. R.: Ecological lessons from free-
air CO
2
enrichment (FACE) experiments, Annu. Rev. Ecol.
Evol. Syst., 42, 181–203, doi:10.1146/annurev-ecolsys-102209-
144647, 2011.
Nowak, R. S., Ellsworth, D. S., and Smith, S. D.: Func-
tional responses of plants to elevated atmospheric CO
2
–
do photosynthetic and productivity data from FACE experi-
ments support early predictions?, New Phytol., 162, 253–280,
doi:10.1111/j.1469-8137.2004.01033.x, 2004.
Pan, Y., Birdsey, R. A., Fang, J., Houghton, R., Kauppi, P. E., Kurz,
W. A., Phillips, O. L., Shvidenko, A., Lewis, S. L., Canadell,
J. G., Ciais, P., Jackson, R. B., Pacala, S. W., McGuire, A. D.,
Piao, S., Rautiainen, A., Sitch, S., and Hayes, D.: A large and
persistent carbon sink in the world’s forests, Science, 333, 988–
993, doi:10.1126/science.1201609, 2011.
Parton, W. J., Scurlock, J. M. O., Ojima, D. S., Gilmanov, T. G., Sc-
holes, R. J., Schimel, D. S., Kirchner, T., Menaut, J., Seastedt, T.,
Garcia Moya, E., Kamnalrut, A., and Kinyamario, J. I.: Observa-
tions and modeling of biomass and soil organic matter dynamics
for the grassland biome worldwide, Global Biogeochem. Cy., 7,
785–809, doi:10.1029/93GB02042, 1993.
Phillips, R. P., Finzi, A. C., and Bernhardt, E. S.: Enhanced root
exudation induces microbial feedbacks to N cycling in a pine
forest under long-term CO
2
fumigation, Ecol. Lett., 14, 187–194,
doi:10.1111/j.1461-0248.2010.01570.x, 2011.
www.biogeosciences.net/11/2341/2014/
Biogeosciences, 11, 2341–2356, 2014
Page 16
2356
K. E. O. Todd-Brown et al.: Soil carbon changes in Earth system models
Phillips, R. P., Meier, I. C., Bernhardt, E. S., Grandy, A. S., Wick-
ings, K., and Finzi, A. C.: Roots and fungi accelerate carbon and
nitrogen cycling in forests exposed to elevated CO
2
, Ecol. Lett.,
15, 1042–1049, doi:10.1111/j.1461-0248.2012.01827.x, 2012.
Piao, S., Sitch, S., Ciais, P., Friedlingstein, P., Peylin, P., Wang, X.,
Ahlström, A., Anav, A., Canadell, J. G., Cong, N., Huntingford,
C., Jung, M., Levis, S., Levy, P. E., Li, J., Lin, X., Lomas, M.
R., Lu, M., Luo, Y., Ma, Y., Myneni, R. B., Poulter, B., Sun,
Z., Wang, T., Viovy, N., Zaehle, S., and Zeng, N.: Evaluation of
terrestrial carbon cycle models for their response to climate vari-
ability and to CO
2
trends, Glob. Change Biol., 19, 2117–2132,
doi:10.1111/gcb.12187, 2013.
R Development Core Team: R: A Language and Environment for
Statistical Computing, Vienna, Austria, 2012.
Riahi, K., Grübler, A., and Nakicenovic, N.: Scenarios of long-
term socio-economic and environmental development under cli-
mate stabilization, Technol. Forecast. Soc. Change, 74, 887–935,
doi:10.1016/j.techfore.2006.05.026, 2007.
Rustad, L., Campbell, J., Marion, G., Norby, R., Mitchell, M., Hart-
ley, A., Cornelissen, J., and Gurevitch, J.: A meta-analysis of
the response of soil respiration, net nitrogen mineralization, and
aboveground plant growth to experimental ecosystem warming,
Oecologia, 126, 543–562, doi:10.1007/s004420000544, 2001.
Schlesinger, W. H. and Lichter, J.: Limited carbon storage in soil
and litter of experimental forest plots under increased atmo-
spheric CO
2
, Nature, 411, 466–469, 2001.
Schuur, E. A. G., Bockheim, J., Canadell, J. G., Euskirchen, E.,
Field, C. B., Goryachkin, S. V, Hagemann, S., Kuhry, P., Lafleur,
P. M., Lee, H., Mazhitova, G., Nelson, F. E., Rinke, A., Ro-
manovsky, V. E., Shiklomanov, N., Tarnocai, C., Venevsky, S.,
Vogel, J. G., and Zimov, S. A.: Vulnerability of permafrost car-
bon to climate change: Implications for the global carbon cycle,
Bioscience, 58, 701–714, doi:10.1641/B580807, 2008.
Schuur, E. A. G., Vogel, J. G., Crummer, K. G., Lee, H., Sickman,
J. O., and Osterkamp, T. E.: The effect of permafrost thaw on
old carbon release and net carbon exchange from tundra, Nature,
459, 556–559, doi:10.1038/nature08031, 2009.
Sistla, S. A., Moore, J. C., Simpson, R. T., Gough, L., Shaver, G. R.,
and Schimel, J. P.: Long-term warming restructures Arctic tundra
without changing net soil carbon storage, Nature, 497, 615–618,
2013.
Six, J., Paustian, K., Elliott, E. T., and Combrink, C.: Soil struc-
ture and organic matter, Soil Sci. Soc. Am. J., 64, 681–689,
doi:10.2136/sssaj2000.642681x, 2000.
Six, J., Conant, R. T., Paul, E. A., and Paustian, K.: Stabilization
mechanisms of soil organic matter: Implications for C-saturation
of soils, Plant Soil, 241, 155–176, 2002.
Talhelm, A. F., Pregitzer, K. S., and Zak, D. R.: Species-specific re-
sponses to atmospheric carbon dioxide and tropospheric ozone
mediate changes in soil carbon, Ecol. Lett., 12, 1219–1228,
doi:10.1111/j.1461-0248.2009.01380.x, 2009.
Taylor, K. E., Stouffer, R. J., and Meehl, G. A.: An overview of
CMIP5 and the experiment design, B. Am. Meteorol. Soc., 93,
485–498, doi:10.1175/BAMS-D-11-00094.1, 2011.
Thornton, P. E., Doney, S. C., Lindsay, K., Moore, J. K., Ma-
howald, N., Randerson, J. T., Fung, I., Lamarque, J.-F., Fed-
dema, J. J., and Lee, Y.-H.: Carbon-nitrogen interactions regu-
late climate-carbon cycle feedbacks: results from an atmosphere-
ocean general circulation model, Biogeosciences, 6, 2099–2120,
doi:10.5194/bg-6-2099-2009, 2009.
Todd-Brown, K. E. O., Randerson, J. T., Post, W. M., Hoffman, F.
M., Tarnocai, C., Schuur, E. A. G., and Allison, S. D.: Causes
of variation in soil carbon simulations from CMIP5 Earth system
models and comparison with observations, Biogeosciences, 10,
1717–1736, doi:10.5194/bg-10-1717-2013, 2013.
Torn, M. S., Trumbore, S. E., Chadwick, O. A., Vitousek, P. M., and
Hendricks, D. M.: Mineral control of soil organic carbon storage
and turnover, Nature, 389, 170–173, 1997.
Trumbore,
S.:
Radiocarbon
and
soil
carbon
dynam-
ics,
Annu.
Rev.
Earth
Planet.
Sci.,
37,
47–66,
doi:10.1146/annurev.earth.36.031207.124300, 2009.
Turetsky, M. R., Kane, E. S., Harden, J. W., Ottmar, R. D., Ma-
nies, K. L., Hoy, E., and Kasischke, E. S.: Recent acceleration of
biomass burning and carbon losses in Alaskan forests and peat-
lands, Nat. Geosci., 4, 27–31, 2011.
Wieder, W. R., Bonan, G. B., and Allison, S. D.: Global soil carbon
projections are improved by modelling microbial processes, Nat.
Clim. Change, 3, 909–912, 2013.
Wieder, W. R., Grandy, A. S., Kallenbach, C. M., and Bonan, G.
B.: Integrating microbial physiology and physiochemical prin-
ciples in soils with the MIcrobial-MIneral Carbon Stabiliza-
tion (MIMICS) model, Biogeosciences Discuss., 11, 1147–1185,
doi:10.5194/bgd-11-1147-2014, 2014.
Wiesmeier, M., Hübner, R., Spörlein, P., Geuß, U., Hangen, E.,
Reischl, A., Schilling, B., von Lützow, M., and Kögel-Knabner,
I.: Carbon sequestration potential of soils in southeast Germany
derived from stable soil organic carbon saturation, Glob. Change
Biol., 20, 653–665, doi:10.1111/gcb.12384, 2014.
Wutzler, T. and Reichstein, M.: Colimitation of decomposition by
substrate and decomposers – a comparison of model formula-
tions, Biogeosciences, 5, 749–759, doi:10.5194/bg-5-749-2008,
2008.
Biogeosciences, 11, 2341–2356, 2014
www.biogeosciences.net/11/2341/2014/