Small temperature benefits provided by realistic afforestation efforts

Abstract

Afforestation, the conversion of croplands or marginal lands into forests, results in the sequestration of carbon. As a result, afforestation is considered one of the key climate-change mitigation strategies available to governments by the United Nations1. However, forests are also less reflective than croplands, and the absorption of incoming solar radiation is greater over afforested areas. Afforestation can therefore result in net climate warming, particularly at high latitudes2-5. Here, we use a comprehensive Earth system model to assess the climate-change mitigation potential of five afforestation scenarios, with afforestation carried out gradually over a 50-year period. Complete (100%) and partial (50%) afforestation of the area occupied at present by crops leads to a reduced warming of around 0.45 and 0.25 °C respectively, during the period 2081-2100. Temperature benefits associated with more realistic global afforestation efforts, where less than 50% of cropland is converted, are expected to be even smaller, indicating that afforestation is not a substitute for reduced greenhouse-gas emissions. We also show that warming reductions per unit afforested area are around three times higher in the tropics than in the boreal and northern temperate regions, suggesting that avoided deforestation and continued afforestation in the tropics are effective forest-management strategies from a climate perspective.

Full text

LETTERS
PUBLISHED ONLINE: 19 JUNE 2011 | DOI: 10.1038/NGEO1182
Small temperature benefits provided by realistic
afforestation efforts
Vivek K. Arora1*and Alvaro Montenegro2
Afforestation, the conversion of croplands or marginal lands
into forests, results in the sequestration of carbon. As a
result, afforestation is considered one of the key climate-
change mitigation strategies available to governments by the
United Nations1. However, forests are also less reflective than
croplands, and the absorption of incoming solar radiation
is greater over afforested areas. Afforestation can therefore
result in net climate warming, particularly at high latitudes2–5.
Here, we use a comprehensive Earth system model to assess
the climate-change mitigation potential of five afforestation
scenarios, with afforestation carried out gradually over a 50-
year period. Complete (100%) and partial (50%) afforestation
of the area occupied at present by crops leads to a reduced
warming of around 0.45 and 0.25 ◦C respectively, during
the period 2081–2100. Temperature benefits associated with
more realistic global afforestation efforts, where less than
50% of cropland is converted, are expected to be even
smaller, indicating that afforestation is not a substitute
for reduced greenhouse-gas emissions. We also show that
warming reductions per unit afforested area are around three
times higher in the tropics than in the boreal and northern
temperate regions, suggesting that avoided deforestation and
continued afforestation in the tropics are effective forest-
management strategies from a climate perspective.
We analyse simulations made with the first-generation Canadian
Earth System Model6,7 (CanESM1), which consists of coupled
dynamical atmosphere and ocean models with full oceanic
and terrestrial carbon-cycle components. This model reproduces
twentieth-century observations of global mean atmospheric CO2,
as well as its seasonal cycle and interhemispheric gradient6. One
standard simulation with no land-use/cover change (LUCC) and
five afforestation simulations are carried out for the 2011–2100
period, all with emissions increasing according to the Special Report
on Emissions Scenarios A2 scenario. These future simulations are
continuations of an 1850–2010 historical simulation that is driven
with observation-based fossil-fuel emissions and changes in crop
area8,9 (see Methods) and has simulated CO2concentration of
381 ppm in 2010. CO2is the longest-lived and most important
of the major greenhouse gases and therefore we focus on
the afforestation response to CO2, which also simplifies the
interpretation of our simulations.
Afforestation is carried out in areas that are at present occupied
by croplands (Supplementary Fig. S1), and which, according to
estimates of potential vegetation, would be occupied by forests if
it were not for human activities (see Methods). The assumption
is that afforestation is not a viable strategy if artificial supply of
water or nutrients, or any other type of high-intensity management,
is required. In the 100% global afforestation simulation, the
1Canadian Centre for Climate Modelling and Analysis, Environment Canada, PO Box 3065, STN CSC, University of Victoria, Victoria, British Columbia,
V8W 3V6, Canada, 2Environmental Sciences Research Centre, Department of Earth Sciences, St Francis Xavier University, Antigonish, Nova Scotia, B2G
2W5, Canada. *e-mail: vivek.arora@ec.gc.ca.
fractional coverage of woody tree plant functional types (PFTs)
is gradually increased, and the fractional coverage of crop PFTs
is gradually decreased, from 2011 to 2060, until the crop area
becomes zero by 2060. This unrealistic simulation provides an
upper limit to potential climatic changes when the land cover
is essentially returned to its preindustrial state. In the 50%-
afforestation simulation only 50% of the global crop area in 2010
is afforested over the 2011–2060 period. This more realistic, but
still somewhat extreme, scenario would require at least a doubling
of crop yield to feed the human population as crop area is
halved over time. The remaining three simulations are identical
to the 50% global afforestation simulation but afforestation is
carried out only over boreal (48.23◦N–90◦N), northern temperate
(22.24◦N–48.23◦N) or tropical (18.58◦S–22.24◦N) latitudinal
bands. The afforested areas in these simulations are summarized
in Table 1. The prescribed changes in the fractional coverage of
PFTs do not determine the structural attributes of vegetation, or
carbon sequestered over land, which are dynamically determined as
a function of simulated climate and atmospheric CO2concentration
(see Methods). The changes to land cover in our simulations
are less drastic than in earlier studies3–5,10 and so provide
insight into the effects of somewhat more realistic afforestation
efforts in conjunction with continuous increase in emissions. In
addition, afforestation is carried out gradually over the 2011–2060
(50 year) period rather than in a step change and its effects
are inferred directly and not by inversion of results from
deforestation simulations10.
The simulated CO2concentration in the standard no-LUCC
simulation increases from 381 ppm in 2010 to 760 ppm in
2100 (see Supplementary Fig. S2a). Afforestation leads to larger
land carbon uptake and consequently lower atmospheric CO2
concentration in all afforestation simulations compared with
the standard no-LUCC case (Fig. 1a,b). The 100% and 50%
global afforestation simulations yield 93 and 45 ppm lower CO2
in 2100, respectively, than the standard no-LUCC simulation.
The drawdown generated in northern-temperate- and tropical-
afforestation (∼20 ppm) simulations is more than double the
drawdown produced by boreal afforestation (9 ppm). This is partly
because the afforested area in the boreal simulation is about half that
in the northern-temperate-afforestation simulation (Table 1). The
tropical-afforestation simulation yields the same CO2drawdown as
the northern-temperate-afforestation simulation despite its lower
afforested area because carbon is sequestered at a faster rate per
unit afforested area in the tropics than in temperate regions. Land
carbon uptake increases by ∼240 and ∼120 Pg C in the 100% and
50% global afforestation simulations, respectively, compared with
the no-afforestation case (Fig. 1a and Supplementary Fig. S2b).
The 240 PgC of land carbon uptake in the 100% afforestation
514 NATURE GEOSCIENCE |VOL 4 |AUGUST 2011 |www.nature.com/naturegeoscience
© 2011 Macmillan Publishers Limited. All rights reserved.

NATURE GEOSCIENCE DOI: 10.1038/NGEO1182 LETTERS
Table 1 |Global and land-only averaged temperature differences between the standard no-LUCC and the five cropland
afforestation simulations for the 2081–2100 period.
Simulation Afforested
area (million
km2)
Temperature difference compared
with the no-afforestation case (◦C)
Statistical significance
Land only Global
1. 100% global afforestation 20.2 −0.63 −0.45 p<0.01
2. 50% global afforestation 10.1 −0.31 −0.25 p<0.01
3. 50% boreal afforestation 2.0 0.01 −0.04 p>0.25
4. 50% northern temperate afforestation 4.7 −0.16 −0.11 p>0.05
5. 50% tropical afforestation 2.7 −0.25 −0.16 p<0.01
The statistical significance of global temperature differencesis also shown on the basis of the unequal-variance Student t-test. Afforested areas in different simulations are also shown.
No LUCC 50% afforestation100% afforestation
50% boreal afforestation 50% temperate afforestation 50% tropical afforestation
Year Year
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
Year
Globally averaged screen temperature (°C)
Difference in cumulative land uptake compared
with the no-LUCC simulation (Pg C)
Difference in CO
2
concentration compared
with the no-LUCC simulation (ppm)
ab c
¬50
0
50
100
150
200
250
¬80
¬60
¬40
¬20
0
20
¬100
2010 2025 2040 2055 2070 2085 2100 2010 2025 2040 2055 2070 2085 2100
14.0
14.5
15.0
15.5
16.0
16.5
17.0
Figure 1 |Effect of cropland afforestation on land carbon uptake, atmospheric CO2and temperature. a,b, Differences in cumulative land uptake (a) and
CO2concentrations (b) in the five afforestation simulations compared with the standard no-LUCC simulation. c, Simulated globally averaged screen
temperature from all simulations.
simulation is larger than the 156 Pg C of cumulative land use change
emissions over the 1850–2005 period11 because the CO2fertilization
effect results in more carbon sequestered per unit afforested
area than there was in 1850. Ocean carbon uptake responds to
atmospheric CO2, with highest uptake in the no-LUCC simulation
and lowest in the 100% afforestation case (Supplementary Fig. S2c).
All afforestation simulations yield lower temperatures (that
is less warming) over the 2081–2100 period compared with
the standard no-LUCC case, although the differences are not
statistically significant (p>0.05) for the northern-temperate- and
boreal-afforestation simulations (see Fig. 1c and Table 1). Warming
reductions are larger over land, except for the boreal afforestation,
which yields a slight warming increase owing to the positive albedo
effect2. Overall, the net temperature benefits of afforestation are
small. Even the unfeasible scenario of afforesting all available
cropland yields reduced warming of only 0.45 ◦C, with the still
unrealistic 50% afforestation scenario resulting in a warming
reduction of 0.25 ◦C over the 2081–2100 period, compared with
2.2◦C warming realized over the 2010–2100 period in the no-
LUCC simulation (Fig. 1c). We define temperature effectiveness of
afforestation (TEA), ξ, as
ξ=1T
A
where 1T(◦C) is the globally averaged temperature difference
averaged over the 2081–2100 period between an afforestation and
the standard no-LUCC simulation and A(million km2) is the area
afforested. 1Tis used for the whole globe (1TG) as well as only
over land (1TL) to obtain ξGand ξLshown in Fig. 2 for the five
afforestation experiments. TEA values are negative, except for ξLfor
the 50%-boreal-afforestation experiment, because of the reduction
in warming that afforestation yields. Quantified in this manner of
reduced warming per unit afforested area, tropical afforestation is
around three times more effective than afforestation in boreal and
northern temperate regions.
The spatial patterns of the temperature response to afforestation
are shown in Fig. 3. The 100%-afforestation simulation, in which
all cropland area is afforested, leads to reduced warming almost
everywhere except regions of high-latitude Eurasia (Fig. 3a). Here
the regional enhanced warming associated with more radiation
absorbed by the forests dominates the global reduction in warming
associated with lower atmospheric CO2. The magnitude of reduced
warming is amplified in the Arctic because of the sea-ice albedo
feedback. The northern high-latitude warming enhancement as-
sociated with lower albedo of forests is more widespread in the
50%-global-afforestation simulation (Fig. 3b) owing to its lower
CO2drawdown, compared with the 100%-afforestation simulation.
In the 50%-northern-temperate- and boreal-afforestation simula-
tions warming enhancement is seen in northern mid/high-latitude
regions, as expected, because the albedo effect of afforestation in
these regions is strengthened by the presence of snow2(Supple-
mentary Fig. S3a and S3b). These experiments also show areas of
enhanced warming over parts of the Atlantic, Pacific and Southern
NATURE GEOSCIENCE |VOL 4 |AUGUST 2011 |www.nature.com/naturegeoscience 515
© 2011 Macmillan Publishers Limited. All rights reserved.

LETTERS NATURE GEOSCIENCE DOI: 10.1038/NGEO1182
50% global afforestation
100% global afforestation 50% temperate afforestation
50% tropical afforestation
50% boreal afforestation
(°C/million km2)
0
0.02
¬0.02
¬0.04
¬0.06
¬0.08
¬0.10
0
0.02
¬0.02
¬0.04
¬0.06
¬0.08
¬0.10
ab
G
ξ
(°C/million km2)
L
ξ
Figure 2 |Temperature effectiveness of afforestation (TEA). a,b, TEA
values for the five afforestation experiments are calculated using
temperature changes relative to the no-LUCC simulation and averaged over
the 2081–2100 period over the whole globe (a) and over land only (b).
oceans. Tropical afforestation generally leads to widespread reduced
warming but with enhancements over several oceanic regions
(Supplementary Fig. S3c).
The United Nations Framework Convention on Climate
Change (UNFCCC), through the Kyoto Protocol12, enables the
atmospheric carbon drawdown generated by afforestation to be
accounted as sequestered carbon for the emission budget of
the signatory nations. However, afforestation also changes the
physical state of the land surface, which affects the regional and
global energy balances. The effect of these radiative (related to
surface albedo changes) and non-radiative (related to changes in
evapotranspiration and surface roughness) biogeophysical changes
are not easily obtained (see, for example, ref. 13) and therefore
not at present taken into account by the UNFCCC. Here we
decompose the reduced warming caused by afforestation into
biogeophysical and biogeochemical components (see Methods) for
the five afforestation simulations (Table 2), which interestingly
add linearly to yield the net temperature response. Whereas
the biogeochemical component yields reduced warming, relative
to the no-LUCC simulation, owing to carbon sequestration
associated with afforestation, the response to the biogeophysical
changes in vegetation varies latitudinally, as is widely known2,3
(Table 2). Biogeophysical changes associated with afforestation
cause enhanced warming at mid–high latitudes associated with
more absorption of incoming solar radiation and reduced warming
in the tropics owing to higher evapotranspiration. The net result
is that afforestation in the tropics provides double the benefits
because both the biogeophysical and biogeochemical processes
act to reduce warming. This is the reason why afforestation
is around three times as effective in the tropics, in terms of
reducing warming, as in northern temperate and boreal regions.
Figure 3c–f shows the spatial pattern of the biogeophysical and
biogeochemical components of the net temperature response to
afforestation in the 100%- and 50%-afforestation simulations.
As expected, the biogeophysical processes result in widespread
enhanced warming, especially over northern mid/high-latitude
land areas and in the Arctic region, where this warming is
amplified owing to the sea-ice-albedo feedback. The effect of
CO2drawdown in offsetting warming is nearly global in both
simulations, with amplification in the Arctic region. In the
100%- and 50%-afforestation simulations the globally averaged
biogeophysical component is near zero (Table 2 and Fig. 3c,d) and
the net temperature response is almost solely determined by the
biogeochemical component. This is an artefact of the enhanced
biogeophysical warming at mid–high latitudes being compensated
by reduced biogeophysical warming in the tropics, compared with
the no-LUCC case.
Afforestation has been considered as a viable climate-change
mitigation strategy. Our simulations suggest that, although this
is true, the temperature benefits provided by afforestation are
marginal. Afforesting 50% of the existing crop area, everywhere
on the globe (an area much larger than the Amazon River
basin), yields a warming reduction of 0.25 ◦C in the last two
decades of the twenty-first century relative to the ∼3.0◦C
temperature increase over the 1850–2100 period obtained using
CanESM1 (for the A2 emission scenario). Temperature benefits
of any realistic afforestation efforts, with afforested area less
than that in the 50%-afforestation scenario, are expected to be
even lower, suggesting that afforestation is not a substitute for
reduced greenhouse-gas emissions. Moreover, in all afforestation
simulations the temperature benefits are not realized until
late in the twenty-first century. However, afforestation does
provide several other benefits and ecosystem services, including
wildlife habitat, provision of timber, pulp and paper, prevention
of soil erosion and, through its sequestration of atmospheric
CO2, reduced acidification of the oceans. When interpreted on
the basis of warming reduction per unit afforested area, the
model simulations suggest avoided deforestation and continued
afforestation in the tropics as more effective forest management
strategies. Quantitative results presented here are subject to
uncertainties, in particular those associated with the difference
in the albedo of forests and croplands, climate sensitivity and
the strength of the CO2fertilization effect, the latter two
of which vary widely between models14,15. Biogeophysical and
biogeochemical processes are influenced by all of these factors, so
both the sign and magnitude of the net temperature response to
afforestation are expected to vary between models. The climate
sensitivity of CanESM1 is similar to that of most Coupled
Climate Carbon Cycle Model Intercomparison Project models16
(Supplementary Fig. S4) although its CO2fertilization effect
is somewhat stronger15. Our finding, however, that the net
temperature effect of any realistic afforestation efforts is an order
of magnitude less than the warming realized over the 1850–2100
period is probably robust.
Methods
CanESM1 is a comprehensive coupled carbon–climate model6,7 based on the
third-generation atmospheric general circulation model of the Canadian Centre
for Climate Modelling and Analysis, with horizontal resolution defined by a
T47 triangular truncation for dynamical terms and ∼3.75◦horizontal grid for
physical terms. In the vertical, the model domain extends to 1 hPa atmospheric
pressure with the thicknesses of the model’s 32 layers increasing monotonically.
The physical ocean component is based on the National Center for Atmospheric
Research community ocean model with no flux adjustment. The ocean model is
implemented at a horizontal resolution of ∼1.86◦such that there are four ocean
grid boxes underlying each atmosphere grid box. There are 29 levels in the vertical
and the vertical resolution increases towards the ocean surface, from 300 m in
the deep ocean to 50 m in the top 200m. Atmospheric CO2is a fully prognostic
three-dimensional tracer in CanESM1 and carbon enters and leaves the atmosphere
in the form of anthropogenic emissions and fluxes to or from the underlying land
and ocean. The biological and inorganic ocean carbon component of CanESM1 is
the Canadian Model of Ocean Carbon, which incorporates an inorganic chemistry
module (solubility pump) and an ecosystem model (organic and carbonate
pumps)17 based on a nutrient–phytoplankton–zooplankton–detritus model18.
Terrestrial ecosystem processes in CanESM1 are modelled using the Canadian
Terrestrial Ecosystem Model6,19 for nine PFTs (needleleaf evergreen and deciduous
trees, broadleaf evergreen and cold and drought deciduous trees, and C3and
C4crops and grasses).
The historical simulation is carried out with prescribed fossil-fuel CO2
emissions and land-use change up to 2010. Observation-based fossil-fuel CO2
emissions20 are used for the 1850–2000 period and emissions for the 2001–2010
period are based on the A2 scenario. The land cover is reconstructed using a
historical crop-area data set8for the 1850–1992 period and its extension based
on the A2 scenario for the 1993–2010 period21,22. Land-use-change emissions are
calculated interactively in the model9.
The preindustrial state of the land cover in 1850 is first used to generate a
potential land-cover map22. Any grid-cell fraction occupied by crops in 1850 is
substituted by non-crop PFTs in proportion to their fractional coverage within the
grid cell at the time. This potential land-cover map, with no crop area, is then used
as a guide to determine which of the five Canadian Terrestrial Ecosystem Model’s
tree PFTs are used to replace croplands when afforestation is implemented. An area
is considered capable of being afforested if it is occupied by crops in 2010 and is, at
the same time, occupied by tree PFTs in the potential land-cover map. Land surface
516 NATURE GEOSCIENCE |VOL 4 |AUGUST 2011 |www.nature.com/naturegeoscience
© 2011 Macmillan Publishers Limited. All rights reserved.

NATURE GEOSCIENCE DOI: 10.1038/NGEO1182 LETTERS
Global average
¬0.45 °C
Temperature change during 2081–2100 owing to afforestation Temperature change during 2081–2100 owing to afforestation
(°C)
Biogeophysical component of the temperature change Biogeophysical component of the temperature change
Biogeochemical component of the temperature change Biogeochemical component of the temperature change
50% afforestation
100% afforestation
Global average
¬0.45 °C
Global average
0.00 °C
Global average
¬0.24 °C
Global average
¬0.01 °C
Global average
¬0.25 °C
Latitude (° N)
Longitude (° E)
¬50
0
50
Latitude (° N)
¬50
0
50
Latitude (° N)
¬50
0
50
Latitude (° N)
¬50
0
50
Latitude (° N)
¬50
0
50
Latitude (° N)
¬50
0
50
¬150 ¬100 ¬50 0 50 100 150
Longitude (° E)
¬150 ¬100 ¬50 0 50 100 150
Longitude (° E)
¬150 ¬100 ¬50 0 50 100 150
Longitude (° E)
¬150 ¬100 ¬50 0 50 100 150
Longitude (° E)
¬150 ¬100 ¬50 0 50 100 150
Longitude (° E)
¬150 ¬100 ¬50 0 50 100 150
¬3 ¬2 ¬1 ¬0.25 0 0.25 1 2 3
~
a
c
e
b
d
f
Figure 3 |Geographical pattern of temperature change owing to cropland afforestation and its separation into biogeophysical and biogeochemical
components. a,b, Temperature change over the 2081–2100 period in the 100%- (a) and 50%- (b) afforestation simulations compared with the standard
no-LUCC case. c–f, The biogeophysical component of this temperature change (c,d) and the biogeochemical component (e,f) for the 100%- and
50%-afforestation simulations, respectively. Negative values (blue colours) indicate reduced warming and positive values (red colours) indicate areas of
enhanced warming.
Table 2 |Decomposition of the net globally averaged temperature response (◦C) of afforestation over cropland area into
biogeophysical and biogeochemical components.
Simulation Net temperature response Biogeophysical component Biogeochemical component
1. 100% global afforestation −0.45 0.00 −0.45
2. 50% global afforestation −0.25 −0.01 −0.24
3. 50% boreal afforestation −0.04 0.01 −0.05
4. 50% northern temperate afforestation −0.11 0.04 −0.15
5. 50% tropical afforestation −0.16 −0.07 −0.09
albedo in CanESM1 depends on the albedo of vegetation, bare soil and snow (if
present). A higher leaf area index of vegetation implies less visible bare-soil fraction
and/or snow on the ground. As vegetation grows on the afforested fraction of grid
cells the albedo changes rapidly (within ∼5–7 years) owing to increases in leaf area
index whereas carbon sequestration progresses slowly.
The decomposition of reduced warming, associated with afforestation, into
biogeophysical and biogeochemical components is carried out using five further
simulations, which use CO2concentration from the afforestation experiments but
the changes in land cover associated with afforestation are not included (referred
to as CO2-only simulations). The difference between these CO2-only and the
corresponding afforestation simulations yields the biogeophysical component of
the temperature effect of afforestation because they use the same CO2concentration
but different land covers. Similarly, the difference between the standard no-LUCC
and the corresponding CO2-only simulations yields the biogeochemical component
NATURE GEOSCIENCE |VOL 4 |AUGUST 2011 |www.nature.com/naturegeoscience 517
© 2011 Macmillan Publishers Limited. All rights reserved.

LETTERS NATURE GEOSCIENCE DOI: 10.1038/NGEO1182
of the temperature effect of afforestation (associated with CO2drawdown) because
they use different CO2concentrations but the same land cover.
Received 17 January 2011; accepted 13 May 2011; published online
19 June 2011
References
1. Nabuurs, G. et al. in IPCC Climate Change 2007: Mitigation. Contribution of
Working Group III to the Fourth Assessment Report of the Intergovernmental
Panel on Climate Change (eds Metz, B., Davidson, O. R., Bosch, P. R., Dave, R.
& Meyer, L. A.) (Cambridge Univ. Press, 2007).
2. Betts, R. A. Offset of the potential carbon sink from boreal forestation by
decreases in surface albedo. Nature 408, 187–190 (2000).
3. Gibbard, S., Caldeira, K., Bala, G., Phillips, T. J. & Wickett, M. Climate effects
of global land cover change. Geophys. Res. Lett. 32, L23705 (2005).
4. Schaeffer, M. et al. CO2and albedo climate impacts of extratropical carbon and
biomass plantations. Glob. Biogeochem. Cycles 20, GB2020 (2006).
5. Bala, G. et al. A. Combined climate and carbon-cycle effects of large-scale
deforestation. Proc. Natl Acad. Sci. 104, 6550–6555 (2007).
6. Arora, V. K. et al. The effect of terrestrial photosynthesis down-regulation
on the 20th century carbon budget simulated with the CCCma Earth System
Model. J. Clim. 22, 6066–6088 (2009).
7. Christian, J. R. et al. The global carbon cycle in the CCCma earth system
model CanESM1: Preindustrial control simulation. J. Geophys. Res. 115,
G03014 (2010).
8. Ramankutty, N. & Foley, J. A. Estimating historical changes in global
land cover: Croplands from 1700 to 1992. Glob. Biogeochem. Cycles 13,
997–1027 (1999).
9. Arora, V. K. & Boer, G. J. Uncertainties in the 20th century carbon budget
associated with land use change. Glob. Change Biol. 16, 3327–3348 (2010).
10. Claussen, M., Brovkin, V. & Ganopolski, A. Biogeophysical versus
biogeochemical feedbacks of large-scale land cover change.
Geophys. Res. Lett. 28, 1011–1014 (2001).
11. Houghton, R. A. in TRENDS: A Compendium of Data on Global Change
(Carbon Dioxide Information Analysis Center, Oak Ridge National
Laboratory, US Department of Energy, 2008).
12. United Nations Framework Convention on Climate Change. Decision 11/CP.7:
Land Use, Land-use Change and Forestry. Publication FCCC/CP/2001/13/Add.1
(2001).
13. Davin, E. L. & de Noblet-Ducoudre, N. Climatic impact of global-scale
deforestation: Radiative versus nonradiative processes. J. Clim. 23,
97–112 (2010).
14. Roe, G. H. & Baker, M. B. Why is climate sensitivity so unpredictable? Science
318, 629–632 (2007).
15. Arora, V. K. & Matthews, H. D. Characterizing uncertainty in modeling
primary terrestrial ecosystem processes. Glob. Biogeochem. Cycles 23,
GB2016 (2009).
16. Friedlingstein, P. et al. Climate–carbon cycle feedback analysis: Results from
the C4MIP model intercomparison. J. Clim. 19, 3337–3353 (2006).
17. Zahariev, K., Christian, J. R. & Denman, K. L. Preindustrial, historical,
and fertilization simulations using a global ocean carbon model with new
parameterizations of iron limitation, calcification, and N2fixation. Prog.
Oceanogr. 77, 56–82 (2008).
18. Denman, K. L. & Peña, M. A. A coupled 1-D biological/physical model of the
northeast subarctic Pacific Ocean with iron limitation. Deep-Sea Res. II 46,
2877–2908 (1999).
19. Arora, V. K. & Boer, G. J. A parameterization of leaf phenology for the
terrestrial ecosystem component of climate models. Glob. Change Biol. 11,
39–59 (2005).
20. Marland, G., Boden, T. A. & Andres, R. J. Trends: A Compendium of Data
on Global Change. (Carbon Dioxide Information Analysis Center, Oak Ridge
National Laboratory, 2008).
21. Alcamo, J., Leemans, R. & Kreileman, E. Global Change Scenarios of the 21st
Century: Results From the IMAGE 2.1 Model (Elsevier, 1998).
22. Wang, A., Price, D. & Arora, V. Estimating changes in global vegetation
cover (1850–2100) for use in climate models. Glob. Biogeochem. Cycles 20,
GB3028 (2006).
Acknowledgements
We would like to thank G. Flato, J. Fyfe and D. Blain and the two anonymous reviewers
for their helpful comments. A.M. is grateful for funding from the Natural Sciences and
Engineering Research Council. We also acknowledge the work of Canadian Centre for
Climate Modelling and Analysis members who developed CanESM1 including, as well
as the first author, G. J. Boer, C. L. Curry, J. R. Christian, K. Zahariev, K. L. Denman,
G. M. Flato, J. F. Scinocca, W. J. Merryfield, W. G. Lee and D. Yang for help with
processing model output.
Author contributions
V.K.A. carried out the model simulations, analysed model output, conceived the CO2-only
experiments and wrote most of the paper. A.M. conceived the primary experiments, put
together land-cover data and helped with the manuscript text.
Additional information
The authors declare no competing financial interests. Supplementary information
accompanies this paper on www.nature.com/naturegeoscience. Reprints and permissions
information is available online at http://www.nature.com/reprints. Correspondence and
requests for materials should be addressed to V.K.A.
518 NATURE GEOSCIENCE |VOL 4 |AUGUST 2011 |www.nature.com/naturegeoscience
© 2011 Macmillan Publishers Limited. All rights reserved.