REVIEW ARTICLE

PUBLISHED ONLINE: 7 DECEMBER 2015 | DOI: 10.1038/NCLIMATE2870

Biophysical and economic limits to negative

CO2 emissions
Pete Smith1*, Steven J. Davis2, Felix Creutzig3,4, Sabine Fuss3, Jan Minx3,5,6, Benoit Gabrielle7,8,
Etsushi Kato9, Robert B. Jackson10, Annette Cowie11, Elmar Kriegler5, Detlef P. van Vuuren12,13,
Joeri Rogelj14,15, Philippe Ciais16, Jennifer Milne17, Josep G. Canadell18, David McCollum15,
Glen Peters19, Robbie Andrew19, Volker Krey15, Gyami Shrestha20, Pierre Friedlingstein21,
Thomas Gasser16,22, Arnulf Grübler15, Wolfgang K. Heidug23, Matthias Jonas15, Chris D. Jones24,
Florian Kraxner15, Emma Littleton25, Jason Lowe24, José Roberto Moreira26, Nebojsa Nakicenovic15,
Michael Obersteiner15, Anand Patwardhan27, Mathis Rogner15, Ed Rubin28, Ayyoob Sharifi29,
Asbjørn Torvanger19, Yoshiki Yamagata30, Jae Edmonds31 and Cho Yongsung32

To have a >50% chance of limiting warming below 2 °C, most recent scenarios from integrated assessment models (IAMs)
require large-scale deployment of negative emissions technologies (NETs). These are technologies that result in the net
removal of greenhouse gases from the atmosphere. We quantify potential global impacts of the different NETs on various fac-
tors (such as land, greenhouse gas emissions, water, albedo, nutrients and energy) to determine the biophysical limits to, and
economic costs of, their widespread application. Resource implications vary between technologies and need to be satisfactorily
addressed if NETs are to have a significant role in achieving climate goals.

D
espite two decades of effort to curb emissions of CO2 and options, to be able to decide which pathways are most desirable for
other greenhouse gases (GHGs), emissions grew faster dur- dealing with climate change.
ing the 2000s than in the 1990s1, and by 2010 had reached There are distinct classes of NETs, such as: (1) bioenergy with
~50 Gt CO2 equivalent (CO2eq) yr−1 (refs 2,3). The continuing rise carbon capture and storage (BECCS)11,12; (2) direct air capture of
in emissions is a growing challenge for meeting the international CO2 from ambient air by engineered chemical reactions (DAC)13,14;
goal of limiting warming to less than 2 °C relative to the pre-indus- (3) enhanced weathering of minerals (EW)15, where natural weath-
trial era, particularly without stringent climate policies to decrease ering to remove CO2 from the atmosphere is accelerated and
emissions in the near future2–4. As negative emissions technologies the products stored in soils, or buried in land or deep ocean16–19;
(NETs) seem ever more necessary 3,5–10, society needs to be informed (4) afforestation and reforestation (AR) to fix atmospheric carbon
of the potential risks and opportunities afforded by all mitigation in biomass and soils20–22; (5) manipulation of carbon uptake by the

1
Institute of Biological & Environmental Sciences, University of Aberdeen, 23 St Machar Drive, Aberdeen, AB24 3UU, UK. 2University of California, Irvine,
Department of Earth System Science, Irvine, California 92697-3100, USA. 3Mercator Research Institute on Global Commons and Climate Change,
Torgauer Street 12-15, 10829 Berlin, Germany. 4Technical University Berlin, Straße des 17, Junis 135, 10623 Berlin, Germany. 5Potsdam Institute for Climate
Impact Research (PIK), PO Box 60 12 03, 14412 Potsdam, Germany. 6Hertie School of Governance, Friedrichstrasse 180, 10117 Berlin, Germany. 7AgroParisTech,
UMR1402 ECOSYS, F-78850 Thiverval-Grignon, France. 8National de la Recherche Agronomique (INRA), Environment and Arable Crops Research Unit,
UMR1402 ECOSYS, F-78850 Thiverval-Grignon, France. 9The Institute of Applied Energy (IAE), Minato 105-0003, Tokyo, Japan. 10Department of Earth System
Science, Woods Institute for the Environment and Precourt Institute for Energy, Stanford University, Stanford, California 94305, USA. 11NSW Department
of Primary Industries, University of New England, Armidale NSW 2351, Australia. 12Copernicus Institute for Sustainable Development, Department of
Environmental Sciences, Utrecht University, Utrecht, 3584 CS, The Netherlands. 13PBL Netherlands Environmental Assessment Agency, PO Box 303 3720,
AH Bilthoven, The Netherlands. 14Swiss Federal Institute of Technology (ETH Zürich), Universitätstrasse 16, Zürich 8092, Switzerland. 15International Institute
for Applied Systems Analysis (IIASA), Schlossplatz 1, Laxenburg A-2361, Austria. 16Laboratoire des Sciences du Climat et de l’Environnement (LSCE),
Institut Pierre-Simon Laplace (IPSL), CEA-CNRS-UVSQ, CEA l’Orme des Merisiers, 91191 Gif-sur-Yvette Cedex, France. 17Stanford University 473 Via Ortega,
Stanford, California, 94305-2205, USA. 18Global Carbon Project, CSIRO Oceans and Atmosphere Research, GPO Box 3023, Canberra, Australian Capital
Territory 2601, Australia. 19Center for International Climate and Environmental Research-Oslo (CICERO), Gaustadalléen 21, Oslo 0349, Norway. 20US Carbon
Cycle Science Program, US Global Change Research Program, Washington, DC 20006, USA. 21University of Exeter, North Park Road, Exeter EX4 4QF, UK.
22
Centre International de Recherche sur l’Environnement et le Développement (CIRED), CNRS-PontsParisTech-EHESS-AgroParisTech-CIRAD, Campus
du Jardin Tropical, 45 bis avenue de la Belle Gabrielle, 94736 Nogent-sur-Marne Cedex, France. 23King Abdullah Petroleum Studies and Research Center,
PO Box 88550, Riyadh 11672, Saudi Arabia. 24Met Office Hadley Centre, FitzRoy Road, Exeter, Devon EX1 3PB, UK. 25University of East Anglia, Norwich
Research Park, Norwich NR4 7TJ, UK. 26Institute of Energy and Environment, University of Sao Paulo, Av. Prof. Luciano Gualberto, 1.289 – Cidade, Universitaria,
São Paulo 05508-010, Brazil. 27University of Maryland, 2101 Van Munching Hall, School of Public Policy, College Park, MD 20742, USA. 28Carnegie Mellon
University, Baker Hall 128A, Pittsburgh, Pennsylvania 15213, USA. 29Global Carbon Project — Tsukuba International Office, c/o NIES, 16-2 Onogawa, Tsukuba,
Ibaraki 305-8506, Japan. 30National Institute for Environmental Studies (NIES), 16-2 Onogawa, Tsukuba 305-8506, Ibaraki, Japan. 31Pacific Northwest National
Laboratory Joint Global Change Research Institute, 5825 University Research Court, Suite 3500, College Park, Maryland 20740, USA. 32Korea University, 5-ga,
Anam-dong, Seongbuk-gu, Seoul 136-701, Korea. *e-mail: pete.smith@abdn.ac.uk

42 NATURE CLIMATE CHANGE | VOL 6 | JANUARY 2016 | www.nature.com/natureclimatechange

© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2870 REVIEW ARTICLE
ocean, either biologically (that is, by fertilizing nutrient-limited a Fossil fuel energy b Bioenergy
areas23,24) or chemically (that is, by enhancing alkalinity 25); (6) altered Atmosphere Atmosphere
agricultural practices, such as increased carbon storage in soils26–28; Fossil fuel Biogenic
Land Ocean Land Ocean
and (7) converting biomass to recalcitrant biochar, for use as a soil emissions emissions
amendment 29. In this Review, we focus on BECCS, DAC, EW and Geological Geological
AR, because there are large uncertainties with ocean-based strate-
gies (for example, ocean iron fertilization30), and other land-based c Carbon capture and storage (CCS) d Bioenergy + CCS (BECCS)
approaches (for example, soil carbon and biochar storage) have been Atmosphere Atmosphere
evaluated elsewhere31–33. Figure 1 depicts the main flows of carbon Fossil fuel Biogenic
among atmospheric, land, ocean and geological reservoirs for fossil Land Ocean emissions emissions Land Ocean
fuel combustion (Fig. 1a), bioenergy (Fig. 1b), carbon capture and Geological Geological
storage (CCS; Fig. 1c) and the altered carbon flows entailed by each
NET (Fig. 1d–h) when carbon is removed from the atmosphere.
e Direct air capture (DAC) f Enhanced weathering
Coupled energy–land-use analyses of NETs using IAMs have
so far focused primarily on BECCS7,34,35 and AR36–39 strategies, and Atmosphere Atmosphere

suggest that they may have considerable cost-competitive potential. Land Ocean
Capture
Land Ocean
Reaction
infrastructure with minerals
Although other NET options have also been studied13,19,40, they are
Geological Geological
not yet represented in most IAMs. The majority of IAMs allow bio-
mass-based NETs in the production of electricity and heat in power
stations as well as hydrogen generation, and sometimes for generating g Afforestation/changed h Ocean fertilization/alkalinization
agricultural practices
other transport fuels or bioplastics. The key distinguishing feature of
NETs is their ability to remove CO2 from the atmosphere. Depending Atmosphere Atmosphere

on the development of overall emissions, this may lead to: (1) a global Land Ocean Land Ocean
net removal of CO2 from the atmosphere by offsetting emissions that
Geological Geological
were released either in the past or in the near future41; or (2) offsetting
ongoing emissions from difficult-to-mitigate sources of CO2, such as
the transportation sector 42,43, as well as non-CO2 GHGs. Figure 1 | Schematic representation of carbon flows among atmospheric,
The Fifth Assessment Report (AR5) by the Intergovernmental land, ocean and geological reservoirs. a, Climate change results from
Panel on Climate Change (IPCC) database includes 116 scenar- the addition of geological carbon to the atmosphere through combustion
ios that are consistent with a >66% probability of limiting warm- or other processing of fossil fuels for energy. Carbon is indicated in red.
ing below 2 °C (that is, with atmospheric concentration levels of b, Bioenergy seeks to avoid the net addition of carbon to the atmosphere
430–480 ppm CO2eq in 2100)41. Of these, 101 (87%) apply global by instead using biomass energy at a rate that matches the uptake of
NETs in the second half of this century, as do many scenarios that carbon by re-growing bioenergy feedstocks. c, Carbon capture and storage
allow CO2 concentrations to grow between 480 and 720 ppm CO2eq (CCS) technologies intervene to capture most of the potential carbon
by 2100 (501/653 apply BECCS; with 235/653 (36%) delivering net emissions from fossil fuels, and return them to a geological (or possibly
negative emissions globally 41; see also Fig. 2). ocean) reservoir. d–h, NETs remove carbon from the atmosphere, either
Results from two recent modelling exercises10,35,44 show that through biological uptake (g,h), uptake by biological or industrial processes
median BECCS deployment of around 3.3 Gt C yr−1 (Supplementary with CCS (d,e) or enhanced weathering of minerals (f). Any atmospheric
Table 3) is observed for scenarios consistent with the <2 °C tar- perturbation will lead to the redistribution of carbon between the other
get (430–480 ppm CO2eq); we assess other NETs for deploy- reservoirs (but these homeostatic processes are not shown). Note that
ment levels that give the same negative emissions in 2100 (see there are significant differences in the materials and energy requirements
Supplementary Methods). for each process to remove (or avoid adding) a unit mass of carbon from
A key question is whether these rates of deployment of NETs (or to) the atmosphere.
can be achieved and sustained. Most of the NETs require the use of
land and water, some use fertilizer, and may also impact albedo. All by surface albedo) and cost, depending on both their character and
NETs are expected to have considerable costs8,10. Earlier studies have on the scale of their deployment. Figure 3 highlights the differences
examined a number of constraints to NETs7,37–39,45–50, but have not in these requirements expressed per t Ceq removed from the atmos-
assessed a range of different NET types together, or considered the phere. Geological storage capacity has recently been evaluated as a
range of impacts included here. We perform a ‘bottom-up’ implied potential limit to implementation for CCS (and hence BECCS)51,52,
resource use analysis rather than a ‘top-down’ potential efficacy so is not considered further here. Indirect effects of NETs through
analysis, using the best available data from the most recent literature. the reduced use of other technologies in pursuit of a given goal —
The evidence base for the values used varies greatly between NETs, for example, potentially fewer nuclear reactors, wind farms and
with some (for example, BECCS) having been the subject of a large solar arrays — are not considered here. The values we have used
body of research, whereas others (for example, EW) have received are estimated from analyses presented in the latest peer-reviewed
less attention. The data sources and a qualitative assessment of the literature (see Supplementary Methods).
confidence and uncertainty in the ranges we derive are described
in detail in the Supplementary Methods. We estimate the impacts Land area and GHG emissions. The area (and type) of land
of each NET per unit of negative emission, that is, per t C equiva- required per unit of Ceq removed from the atmosphere, also termed
lent (Ceq), then assess the global resource implications, focussing the land use intensity, is particularly important for land-based NETs
on the limits to large-scale NET deployment and how these differ (Fig. 3a). The land use intensity of BECCS is quite high, with values
between NETs. ranging from ~1–1.7 ha t−1 Ceq yr−1 where forest residues are used
as the BE feedstock, ~0.6 ha t−1 Ceq yr−1 for agricultural residues,
Impacts of NETs per unit of negative emissions and 0.1–0.4 ha t−1 Ceq yr−1 when purpose-grown energy crops are
NETs vary dramatically in terms of their requirements for land, used. Supplementary Table 2 shows the carbon and GHG emissions
GHG emissions removed or emitted, water and nutrient use, energy and removals associated with a range of energy crops and forest
produced or demanded, biophysical climate impacts (represented types, and the net negative emissions delivered (see Supplementary

NATURE CLIMATE CHANGE | VOL 6 | JANUARY 2016 | www.nature.com/natureclimatechange 43

© 2015 Macmillan Publishers Limited. All rights reserved

REVIEW ARTICLE NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2870

30 Nutrients. These are depleted when biomass is removed from a field

>1,000 ppm CO2eq or ecosystem for use as a BE feedstock. This is therefore an issue for
25 650–1000 ppm BECCS and for AR when biomass is removed from the site, but not
530–650 ppm for DAC or EW. Perennial energy crops typically contain around
20
480–530 ppm 10 kg N t−1 Ceq (and 0.8 kg P t−1 Ceq in the case of Miscanthus 57),
Net emissions (Gt C yr−1)

430–480 ppm trees around 4–5 kg P t−1 Ceq, and annual energy crops (such as fibre
sorghum) around 20 kg P t−1 Ceq. Nutrient removal therefore differs
15
several-fold among biomass sources (Fig. 3c), so large-scale transi-
2015 estimate
tion to using land for biomass production could deplete nutrients,
10 but this will depend on the vegetation (or other land use) that is
replaced. Additional nutrient requirements (that is, fertilization) are
5 Historical emissions difficult to estimate on a net basis, as fertilizer may also have been
used (with varying intensity) on the land before the switch to energy
0 crops58. Nutrient depletion further translates into agricultural inputs
and upstream GHG emissions and energy consumption.
−5
1980 2000 2020 2040 2060 2080 2100 Albedo. In addition to biogeochemical climate impacts (for exam-
ple, uptake of atmospheric carbon), changes in land use affect cli-
Year
mate by altering the physical characteristics of the Earth’s surface,
such as increased evapotranspiration59 and increased cloud cover in
Figure 2 | Scenarios including NETs for each of the scenario categories,
the tropics60. Important among these physical changes is albedo (here
corresponding to the ranges and median values shown in Supplementary
we focus on surface albedo), which is the reflectance of solar energy
Table 3. Scenarios with no technology constraints (that is, including NETs)
by the Earth’s surface. The albedo of lighter-coloured and less-dense
from the AMPERE10,44 and LIMITS35 modelling comparison exercises are
vegetation (for example, food crops and grasses) is much greater than
shown in colours, with all other scenarios from the IPCC AR5 database
that of trees53,61. The situation is further complicated in areas where
shown in grey. See the caption of Supplementary Table 3 for an explanation
shorter vegetation may be persistently covered by highly reflective
of the representation of gross positive and gross negative emissions.
snow in winter, while tall coniferous trees remain exposed and there-
Net land use change fluxes are included (note, the 1997 fluctuation is
fore much less reflective61. This snow-mediated effect is large enough
attributable to Indonesian peat fires). Sources: CDIAC94, IPCC AR5 scenario
to mean that AR in northerly latitudes may have a neutral or net
database (https://secure.iiasa.ac.at/web-apps/ene/AR5DB/)95 and the
warming effect (larger than the carbon sink provided by the vegeta-
Global Carbon Project.
tion)62–65. Figure 3d shows the change in albedo under different NETs
(focussing on the replacement of cropland or grassland with energy
Methods). EW and DAC have minimal land requirements, crops) or under AR, both with and without the effect of snow.
with land use intensities of <0.01 ha t−1 Ceq yr−1 (ref. 18) and
<0.001 ha t−1 Ceq yr−1 (ref. 14), respectively (Fig. 3a). Costs. The economic costs of deploying and operating NETs will vary
according to the specific technologies involved, the scale of deploy-
Water use. This is highly variable between different BE feedstocks ment and observed learning, the amount and value of co-products,
(including forest feedstocks) and is generally considered to be site-specific factors and the scale and cost of building and maintain-
higher for short-rotation coppice and C4 grasses than for annual ing any supporting infrastructure (the costs of capturing and stor-
crops and grassland (on an area basis)53, although when corrected ing a t Ceq are from studies using approximate 2005 to 2015 US$
for biomass productivity, the ranges are closer and overlap consider- values). In the case of BECCS and DAC, costs can be anticipated to
ably 54 (Fig. 3b). In calculating water implications of BECCS, water occur across three stages: (1) capture, (2) transport and (3) storage
use for CCS is added to the BE water use (Supplementary Methods). (including monitoring and verification). Recent estimates of the total
Where deployed, irrigation also has a dominant impact on water costs of DAC technologies40,66 are $1,600–2,080 per t Ceq, of which
use. Estimates of water required per t Ceq removed by DAC and roughly two-thirds are capital costs and one-third operating costs
EW are an order of magnitude or more lower than for BECCS (Fig. 3f). Although there are very wide ranges for costs of BECCS
(Fig. 3b). For EW of olivine, one molecule of water is required for technologies67, the mean price estimated across 6 IAMs for 210046 was
each molecule of CO2 removed, so each t Ceq would require 1.5 m3 $132 per t Ceq (Fig. 3f); costs of bioenergy without CCS are lower 54,55.
water (Fig. 3b). AR costs are estimated to be $65–108 per t Ceq for 2100, with a mean
of $87 per t Ceq. Estimated costs of EW are taken from Renforth56:
Energy input/output. This varies considerably between different $88–2,120 per t Ceq, with a mean of around $1,104 per t Ceq; these
NETs. BECCS has a positive net energy balance, with energy pro- estimates are uncertain and the relative balance between capital and
duction ranges of 3–40 GJ t−1 Ceq for energy crops55 (Fig. 3e). DAC operating costs has not yet been thoroughly examined.
and EW, on the other hand, require considerable energy input to
deliver C removal; the minimum theoretical energy input require- Global resource implications of NETs deployment
ment for the chemical reactions of DAC14 is 1.8 GJ t−1 Ceq removed We use global deployment of BECCS in the recent assessments
at atmospheric concentrations of CO2, and for EW of olivine is featured in Supplementary Table 3 to derive the corresponding
0.28–0.75 GJ t−1 Ceq (Fig. 3e). When also including other energy resource implications (Table 1), and focus on the scenario giving a
inputs for mining, processing, transport, injection and so on, the 2100 atmospheric CO2 concentration in the range of 430–­480 ppm
energy inputs for DAC and EW are much greater, perhaps as much (consistent with a 2 °C target). We compare DAC resource impli-
as 45 GJ t−1 Ceq and 46 GJ t−1 Ceq, respectively 14,56 (Fig. 3e). The cations at the same level of negative emission as BECCS (that is,
GHG implication of this additional energy use depends on the GHG 3.3 Gt Ceq yr–1 in 2100; Table 1). For other NETs, which are not able
intensity of the energy supply, which is likely to change over the rest to meet the same level of emissions removal, we use values compiled
of this century. Energy requirement is less important if low-carbon from an analysis of the recent literature to give mean and maxi-
energy is used (for example, using large areas of solar photovoltaic mum implementation levels (see Supplementary Methods). Mean
panels to power DAC plants45), but may still have additional impacts. values for carbon removals from AR are estimated to be around

44 NATURE CLIMATE CHANGE | VOL 6 | JANUARY 2016 | www.nature.com/natureclimatechange

© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2870 REVIEW ARTICLE
a b
0.6 3.0
Land Water

Water required (thousands of m3 per t Ceq)

0.5 2.5
Area required (hectares per yr per t Ceq)

0.4 2.0

CCS
Ded. crop (marginal)
0.3 1.5
Dedicated crops

Feedstock

Ded. crops (marginal)

1.0

DAC (e.g. amines)

0.2
DAC (e.g. amines)

Dedicated crops
Crop residues
Crop residues

EW (olivine)

EW (olivine)
0.1 0.5

Coppice
Tropical
Coppice
Tropical

Boreal
Boreal

Pine
Pine
0 0
Crop BECCS CDR Forest BECCS/AR Crop BECCS CDR Forest BECCS

c d
25 −70%
Nitrogen concentration in feedstock/biomass (kg N per t Ceq)

Nutrients Albedo

20
−50%
Change in albedo

15

Boreal (Summer)

Boreal (snow)
−30%

Ded. crop (marginal)

Dedicated crops
10

Pine
Crop residues

Coppice
Ded. crops (marginal)

Tropical
−10%
Dedicated crops

5
Crop residues

0
Pine

Coppice

Tropical
Boreal

0 +10%
Crop BECCS Forest BECCS/AR Crop BECCS Forest BECCS/AR

e f
55 2,250
Energy Cost
Energy required

2,000
35
Cost of negative emissions (US$ per t Ceq)

1,750
DAC (e.g. amines)

15
1,500
Net energy (GJ per t Ceq)

EW (olivine)

0
EW (olivine)
DAC (e.g. amines)

350
Ded. crops (marginal)

15
Energy produced

Pine

Tropical
Boreal

Forest
Crop

35 250
Coppice
Dedicated crops
Crop residues

55
150

75
50
95 0
Crop BECCS CDR Forest BECCS Crop BECCS CDR Forest BECCS

Figure 3 | The different requirements and impacts of NETs. a–f, Negative emissions technologies have different land (a), water (b) and nutrient (c)
requirements, different geophysical impacts on climate (for example, albedo; d), generate or require different amounts of energy (e), and entail different
capital and operating costs (f). For example, carbon dioxide removal (CDR) technologies such as DAC and EW of silicate rock tend to require much less
land and water than strategies that depend on photosynthesis to reduce atmospheric carbon (a,b), but the CDR technologies demand substantial energy
and economic investment per unit of negative emissions (e,f). Among BECCS options, forest feedstocks tend to require less nitrogen than purpose-grown
crops (c), but present greater risk of unwanted changes in albedo (d), and generate less energy (e). AR has been omitted from b,e,f to avoid confusion with
forest BECCS (where the CCS component is included). See Supplementary Methods and Table 1 for data sources.

NATURE CLIMATE CHANGE | VOL 6 | JANUARY 2016 | www.nature.com/natureclimatechange 45

© 2015 Macmillan Publishers Limited. All rights reserved

REVIEW ARTICLE NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2870

Table 1 | Global impacts of NETs for the average needed global C removals per year in 2100 in 2 °C-consistent scenarios
(430–480 ppm scenario category; Supplementary Table 3).
NET Global C removal Mean (max.) Estimated energy Mean (max.) Nutrient impact Albedo impact Investment needs
(Gt Ceq yr−1 in land requirement requirement water requirement (kt N yr−1 in in 2100 (BECCS for electricity/
2100) (Mha in 2100) (EJ yr−1 in 2100) (km3 yr−1 in 2100) 2100) biofuel; US$ yr−1 in 2050)
BECCS 3.3 380–700 −170 720 Variable Variable 138 billion /123 billion
DAC 3.3 Very low (unless 156 10–300 None None >>BECCS
solar PV is used
for energy)
EW* 0.2 (1.0) 2 (10) 46 0.3 (1.5) None None >BECCS
AR* 1.1 (3.3) 320 (970) Very low 370 (1,040) 2.2 (16.8) Negative, or <<BECCS
reduced GHG
benefit where not
negative
*NETs with lower maximum potential than the BECCS emission requirement of 3.3 Gt Ceq per year in 2100; their mean (and maximum) potential is given along with their impacts (see Supplementary Methods).
Wide ranges exist for most impacts, but for simplicity and to allow comparison between NETs (sign and order of magnitude), mean values are presented. See main text and Supplementary Methods for full details.
PV, photovoltaic.

1.1 Gt Ceq yr–1 by 2100, with a maximum value of 3.3 Gt Ceq yr–1 for Total agricultural land area in 2000 was ~4,960 Mha, with an area
very large-scale deployment 6,7,68 (Table 1). The potential of carbon of arable and permanent crops of ~1,520 Mha71, so area for BECCS
removal by EW (including adding carbonate and olivine to both (380–700 Mha) represents 7–25% of agricultural land, and 25–46%
oceans and soils) has been estimated to be as great as 1 Gt Ceq yr–1 of arable plus permanent crop area. AR (at 1.1–3.3 Gt Ceq yr−1
by 2100, but with mean annual removal an order of magnitude less68 negative emissions; 320–970 Mha, respectively) represents 6–20%
at 0.2 Gt Ceq yr–1. Combined with the bottom-up, per-t-Ceq impact of total agricultural land, and 21–64% of arable plus permanent
ranges (Supplementary Methods), we then assess the resource crop area. This range of land demands are 2–4 times larger than
implications, and the extent to which available resources may limit land identified as abandoned or marginal72. Thus, the use of BECCS
the deployment of NETs globally. and AR on large areas of productive land is expected to impact
the amount of land available for food or other bioenergy produc-
Land area. DAC has a small direct land footprint (Fig. 3a) and can tion12,37,73–75, as well as the delivery of other ecosystem services12,32,76,
be deployed on unproductive land that supplies few ecosystem ser- which may prove to be a limit to the implementation of BECCS77
vices14, although the land footprint could be considerable if solar and AR. One uncertainty is the future rate of increase of food crop
pholtovoltaic panels or wind turbines were used to provide the yields37,78 and whether this will meet future food demand79, thereby
energy required45. EW has a larger land footprint if the minerals are potentially freeing more cropland for BECCS or AR, even if at a
applied to the land surface (as opposed to the oceans, or if weather- higher price37.
ing reactions occur in industrial autoclaves), although crushed oli-
vine or carbonates could be spread on agricultural and forest land Water. Increasing global water stress is attributable to rising water
to allow the weathering to take place, with the added benefits of demands and reduced supplies, both of which can be exacerbated
raising the pH of acidic soils to make them more productive15. Thus, in some locations by climate change80. In particular, the evaporative
EW technologies may not always compete for land with other uses, demand of plants increases with temperature as vapour pressure
despite the large areas involved (for example, the estimated poten- deficit increases. Evaporative loss can be 20–30 mol H2O per mol
tial of 1 Gt Ceq yr−1 removed might require 10 Mha)15. CO2 absorbed by an amine DAC unit 14,81, giving a water use esti-
Assuming per-area carbon in biomass available for capture as mate of ~92 (mean; 73–110) m3 t−1 Ceq (Fig. 3b). Implementation
a feedstock for BECCS of widely applicable, high-productivity at levels of 3.3 Gt Ceq yr−1 in 2100 (Table 1) would therefore be
dedicated energy crops (willow and poplar short rotation cop- expected to use ~300 km3 yr−1 of water assuming current amine
pice (SRC) and Miscanthus; 4.7–8.6 t Ceq ha−1 yr−1; Supplementary technology, which is 4% of the total current evapotranspiration
Table 2), BECCS delivering 3.3 Gt Ceq yr−1 of negative emissions used for crop cultivation82. Sodium hydroxide for DAC, however,
would require a land area of approximately 380–700 Mha in 2100 uses 3.7 m3 t−1 Ceq (Fig. 3b)81, so equivalent levels of implemen-
(Table 2), with a wider possible range that is determined by pro- tation using sodium hydroxide in place of amines would result in
ductivity (Supplementary Table 2). This emissions removal is water use of ~10 km3 yr−1. For EW, with a water use of 1.5 m3 t−1 Ceq
equivalent to 21% of total current human appropriated net pri- (Fig. 3), deployment to remove 0.2 (mean) or 1 (maximum)
mary productivity (NPP) (15.6 Gt C yr−1 in 2000), or 4% of total Gt Ceq yr −1 would involve water use of 0.3 and 1.5 km3, respectively.
global potential NPP69. Areas for AR that are calculated assuming Water use for forests is estimated to be 1,765 (1,176–
a mean carbon uptake over the growth period of 3.4 t Ceq ha−1 yr−1 2,353) m3 t−1 Ceq yr−1, which includes both interception and tran-
(Supplementary Methods; Fig. 3a) give a land area corresponding to spiration (Fig. 3b). However, because trees replace other vegetation
1.1 and 3.3 Gt C yr−1 removed in 2100 of ~320 and ~970 Mha, respec- during AR, the total net impact must be calculated by subtracting
tively, similar to other estimates50. Estimates of land use by BECCS the water use of the previous land cover. Assuming a water use sim-
and AR are consistent with the values presented in previous stud- ilar to short vegetation of 1,450 (900–2,000) m3 t−1 Ceq yr−1 before
ies47 for three IAMs (Global Change Assessment Model (GCAM), AR (Fig. 3b), the additional water use from AR is estimated to be
Integrated Model to Assess the Global Environment (IMAGE) around 315 m3 t−1 Ceq yr−1, which is 1% of the total evapotranspi-
and Regional Model of Investments and Development/Model ration from current forests82. For AR delivering capture of 1.1 or
of Agricultural Production and its Impact on the Environment 3.3 Gt C yr−1 (Table 1), additional water use is thus estimated to be
(ReMIND/MAgPIE)), although other studies suggest larger areas39. ~370 or 1,040 km3 yr−1, respectively.
Without global forest protection, increased bioenergy deployment Similar calculations can be made for BECCS. For
would increase GHG emissions from land-use change70. unirrigated bioenergy, evaporative loss is estimated to be

46 NATURE CLIMATE CHANGE | VOL 6 | JANUARY 2016 | www.nature.com/natureclimatechange

© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2870 REVIEW ARTICLE
1,530 (1,176–1,822) m3 t−1 Ceq yr−1, which is 80 m3 t−1 Ceq yr−1 more 4.0
than for average short vegetation (Fig. 3b). Thus, deployment of
BECCS at 3.3 Gt Ceq yr−1 in 2100 would lead to additional water 3.5 BECCS DAC Land

Negative emissions (Gt Ceq per year)

use of ~260 km3 yr−1 from the crop production phase. There is an requirements
3.0 720 1,040
opportunity cost of using soil moisture for sequestration and/or 300 (Mha per yr)
bioenergy production rather than for growing food. Our estimates 1000
2.5
for water use are an order of magnitude lower than other recent
AR 800
estimates for bioenergy crops48 and for AR50, as water use in those 2.0
studies were expressed as a total rather than additional water use 600
due to land use change, and those for bioenergy also considered 1.5
400
irrigation48. Irrigated bioenergy crops were estimated to double EW
agricultural water withdrawals in the absence of explicit water pro- 1.0 200
370
tection policies48, which could pose a severe threat to freshwater <1
0
ecosystems, as human water withdrawals are dominated by agri- 0.5
culture and already lead to ecosystem degradation and biodiversity
0
loss. Land requirements for bioenergy crops would greatly increase −200 −100 produced 0 100 required 200
(by ~40%, mainly from pastures and tropical forests) if irrigated Energy (EJ per year)
bioenergy production was excluded, meaning that there will be
a trade-off between water and land requirements if bioenergy is
implemented at large scales48. Figure 4 | The impacts and investment requirements of NETs to meet
For BECCS, additional water is required for CCS, adding about the 2 °C target. A schematic representation of the aggregate impacts of
450 m3 t−1 Ceq yr−1 to the evaporative loss relative to bioenergy alone14 NETs on land, energy and water, and relative investment needs, for levels
(Fig. 3b), equivalent to an additional water use of ~720 km3 yr−1 of implementation equivalent to BECCS (3.3 Gt C yr−1 negative emissions
due to BECCS (the sum of additional evaporative loss plus CCS in 2100) in scenarios consistent with a 2 °C target (or mean and maximum
water use), for the 3.3 Gt Ceq yr −1 by 2100 level of implementation attainable, where that level of negative emissions cannot be reached).
(Table 1). BECCS would thus require an additional quantity of water Water requirement is shown as water droplets, with quantities in km3 yr−1.
equivalent to ~10% of the current evapotranspiration from all crop- All values are for the year 2100 except relative costs, which are for 2050
land areas worldwide82. (see Supplementary Methods).
To put these figures in context, total global renewable freshwater
supply on land is 110,300 km3 yr−1, of which humans appropri- would amount to a global energy requirement of 156 EJ yr−1 if all
ate 24,980 km3 yr−1 (ref. 83), so the implementation of BECCS at energy costs are included (Table 1). This is equivalent to 29% of total
3.3 Gt Ceq yr−1 of negative emissions by 2100 represents an addi- global energy use in 2013 (540 EJ yr−1), and a significant propor-
tional use of ~3% of the freshwater currently appropriated for tion of total energy demand in 2100 (which the IPCC AR5 scenario
human use. AR implemented at 1.1 Gt Ceq yr−1 by 2100 would rep- database estimates will be~500–1,500 EJ yr−1), which will be a major
resent 1–2% of human-appropriated freshwater. Expressing addi- limitation unless low-GHG energy could be used, or the energy
tional water use as a proportion of runoff in a region would provide requirements significantly reduced.
a more accurate picture of the threat to water resources at a given
location — but this is not feasible without a spatially disaggregated Nutrients. DAC has no impact on soil nutrients, and EW may (in
analysis. Nevertheless, with human pressures on freshwater increas- some cases) provide beneficial minerals and pH adjustment that are
ing 80,84, water use could act as a significant limitation to implementa- difficult to quantify at the aggregate level. Nutrient concentrations in
tion of high-water-demand NETs such as BECCS. crop biomass are often higher than in tree biomass (Fig. 3c), but nutri-
ents are removed from cropland and grazing land in agricultural prod-
Energy. Bioenergy currently supplies about 10% of primary energy ucts, whereas AR on agricultural land is likely to increase the retention
worldwide55, that is, an estimated 44.5 EJ yr−1. Of this, 74% comes of nutrients within an ecosystem. However, nutrient limitation could
from fuel wood, 9% from forest and agricultural residues, 8% from limit productivity, which may limit carbon storage49. Nutrients are
recovered wood, 6% from industrial organic residues and 3% from also removed when bioenergy feedstocks are removed from the site
dedicated energy crops55. Most of this biomass, however, cannot cur- on which they are grown, resulting in the depletion of nutrients rela-
rently be used for BECCS, as the vast majority is used in small-scale tive to land uses where biomass is not removed, but not necessarily at
applications; for example, for household cooking and heating in the same level as agricultural land86. Bioenergy feedstocks with low
developing countries55. BECCS delivering 3.3 Gt Ceq yr−1 of negative nutrient concentrations, such as residue, forest and lignocellulosic bio-
emissions would deliver ~170 EJ yr−1 of primary energy in 210010,35,44 mass, should hence be favoured over feedstocks with higher nutrient
(Table 1). Estimates of future energy potential vary greatly; there is concentrations. Assuming the nutrient concentrations of forests are
high consensus that 100 EJ yr−1 could be attained, and a medium 2.0 to 5.1 kg N t−1 Ceq (Fig. 3c), and that most nutrients are removed
level of agreement that 100–300 EJ yr−1 could be attained — but there at harvest for energy and food crops, AR areas of ~320 and 970 Mha
is only low consensus that primary energy above 300 EJ yr−1 could (consistent with AR removing 1.1 (mean) and 3.3 (high) Gt Ceq yr−1
be supplied by bioenergy 12,32. Stabilization scenarios from the IAM (Table 1)) would increase global nitrogen retention in biomass by
literature suggest that bioenergy could supply from 10 to 245 EJ yr−1 2.2–5.6 and 6.6–16.8 kt N yr−1, respectively. Scaling values for imple-
of global primary energy by 205070,87, and deliver a sizable contribu- mentation of 1 Gt Ceq yr−1 of negative emissions50, P and N demand
tion to primary energy in 210041. to balance the carbon stored is estimated to be 220–990 kt P yr−1
The energy required by AR is very low (for site preparation only) and 100–1,000 kt N yr−1 for AR at 1.1–3.3 Gt Ceq yr−1 of negative
and is assumed here to be negligible. Other NETs have large energy emissions — although it must be noted that these values are absolute,
demands (Fig. 3e). Using our realistic estimate of 46 GJ of energy and do not account for the P and N in the vegetation replaced by AR.
required per t Ceq removed by EW (Fig. 3), the 0.2–1.0 Gt C yr−1
that might be captured (Supplementary Table 2) would entail up to Albedo. The effect of DAC and EW on the reflectivity of the Earth’s
46 EJ yr−1 of energy in 2100 (Table 1). The energy requirements of surface is assumed to be small (excluding possible use of solar pho-
amine DAC14 (Fig. 3e) deployed for net removal of ~3.3 Gt Ceq yr−1 tovoltaic panels to generate energy for DAC45; Fig. 3d). However,

NATURE CLIMATE CHANGE | VOL 6 | JANUARY 2016 | www.nature.com/natureclimatechange 47

© 2015 Macmillan Publishers Limited. All rights reserved

REVIEW ARTICLE NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2870
the land areas required for BECCS and AR can dramatically affect These biophysical and economic resource implications may directly
albedo (Fig. 3d). Because the effect is greatly amplified by the pres- impose limits on the implementation of NETs in the future, but they
ence of snow, the exact location (latitude and elevation) of the may also indirectly constrain NETs by interacting with a number of
BECCS or AR, and the vegetation it replaces, is critical in assessing societal challenges facing humanity in the coming decades, such as
the impact on albedo (Fig. 3d). Albedo can significantly reduce62 or food, water and energy security, and thereby sustainable develop-
even reverse net radiative forcing from AR at northern latitudes63. ment. In addition to the biophysical and economic limits to NETs
This observation could limit the value of AR for climate mitigation considered here, social, educational and institutional barriers, such
in northerly regions. For BECCS, the replacement of short veg- as public acceptance of and safety concerns about new technologies
etation with taller vegetation (for example, Miscanthus and SRC), and related deployment policies, could limit implementation. The
could have similar effects on albedo, although probably less than drivers, risks, and limitations of the supply of NETs, showing activi-
the impact of AR with coniferous forest (Fig. 3d). Because AR is ties thought to increase the potential supply of NETs, as well as the
more likely to occur at high latitudes than production of BECCS risks and geophysical and societal limits to the potential of NETs,
feedstocks, BECCS should not have a deleterious impact on albedo. are shown in Supplementary Fig. 1. Commercialization and deploy-
At low to mid latitudes, AR could increase radiative forcing by ment at larger scales will also allow more to be learnt about these
decreasing albedo; but, without a regional distribution, the scale of technologies, in order to improve their efficiency and reduce cost.
these impacts cannot be assessed. To inform society of the potential risks and opportunities
afforded by all mitigation options available, more research on NETs
Investment needs. The deployment of NETs (specifically BECCS) is clearly required. Although we have collated the best available data
in IAM scenarios is an outcome of an optimization of costs over on NET impacts and have reflected changes related to deployment
time. The existence of large-scale gross negative emissions even scale as accurately as possible, it is clear that common modelling
in less-ambitious stabilization pathways indicates that BECCS is frameworks are required to implement learning, cost, supply and
selected as a cost-effective component of the energy mix, allow- efficiency curves for all NETs. By implementing such curves, future
ing higher residual emissions elsewhere, which would otherwise models will be able to develop portfolios of trajectories of NET
be more expensive to abate. Investments in BECCS provide an development, allowing least-cost options to be selected, and learn-
additional indicator for assessing the scale and speed of BECCS ing and efficiency improvements to be reflected. The inconsistency
deployment over the next several decades. Supplementary Table 4 in coverage of NETs and their impacts highlights this key knowl-
summarizes investment estimates from six global integrated assess- edge gap; this analysis will help to frame these developments in the
ment models that assessed 2 °C scenarios within the context of the modelling community.
LIMITS model intercomparison87 for 2030 and 2050: US$36.2 and For BECCS, research and development is required to deliver
29.4 billion yr−1, respectively, worth of investment is estimated as high-efficiency energy conversion and distribution processes for
optimal by 2030 for scaling up biomass electricity and biofuels pro- the lowest-impact CCS, and the cost of infrastructure to transport
duction technologies worldwide on average. By 2050, these invest- CO2 from BECCS production areas to storage locations needs to
ment levels grow to US$138.3 and 122.6 billion yr−1, respectively 87. be further evaluated. To this end, early deployment of CCS would
This represents 5 and 4%, respectively, of the projected total global enhance understanding of the risks and possible improvements
energy system investments required by 2050 of US$2,932 (inter- of the technology. Integrated pilot plants need to be built (storing
model range: $1,889–4,338) billion yr−1 (ref. 87). Investment needs ~1 Mt CO2 per year) to examine how combined BECCS functions90;
for DAC, EW and AR are not known, but given the much higher the capital cost of 5–10 full-size demonstrations of BECCS or CCS
unit costs (per t Ceq) for DAC, and the higher costs of EW and the would require the investment of approximately US$5–10 billion90.
lower unit costs of AR described above, the investment needs are There is also a need to develop socio-economic governance systems
estimated qualitatively (relative to BECCS; Table 1). for all NETs, to provide incentives to fund this research and devel-
The aggregate impacts of NETs on land, energy and water, and the opment, and implementation of infrastructure in the most sustaina-
relative investment needs for levels of implementation equivalent to ble manner, to limit adverse impacts in the transition to low-carbon
BECCS in scenarios consistent with a 2 °C target (3.3 Gt Ceq yr−1, energy systems, and to manage the risks associated with CCS (such
or the mean and maximum attainable where that level of negative as leakage, seismic action and environmental impacts)91. Priorities
emissions cannot be reached) described in this section are summa- include investing in renewable and low-carbon technologies, effi-
rized schematically in Fig. 4. ciency and the integration of energy systems (to make the most of
waste heat, excess electrons from photovoltaic panels and wind, and
Discussion to close the carbon cycle of fossil sources by capturing and reusing
Biophysical, biogeochemical (that is, nutrients), energy and eco- CO2 by catalysis), and the realization of additional environmental
nomic resource implications of large-scale implementation of NETs benefits. In the meantime, emission reductions must continue to be
differ significantly. For DAC, costs and energy requirements are the central goal for addressing climate change.
currently prohibitive and can be anticipated to slow deployment. Addressing climate change remains a fundamental challenge for
Research and development is needed to reduce costs and energy humanity, but there are risks associated with relying heavily on any
requirements. For EW, the land areas required for spreading and/ technology that has adverse impacts on other aspects of regional or
or burying crushed olivine are large, such that the logistical costs planetary sustainability. Although deep and rapid decarbonization
may represent an important barrier, compounded by the fact that may yet allow us to meet the <2 °C climate goal through emissions
the plausible potential for carbon removal is lower than for other reduction alone8, this window of opportunity is rapidly closing 8,92
NETs. In contrast, AR is relatively inexpensive, but the unintended and so there is likely to be some need for NETs in the future41,93. Our
impacts on radiative forcing through decreased albedo at high lati- analysis indicates that there are numerous resource implications
tudes, and increased evapotranspiration increasing the atmospheric associated with the widespread implementation of NETs that vary
water vapour content, could limit effectiveness; likewise, increased between technologies and that need to be satisfactorily addressed
water requirements could be an important trade-off, particularly in before NETs can play a significant role in achieving climate change
dry regions. Competition for land is also a potential issue, as it is goals. Although some NETs could offer added environmental
for BECCS50,88,89. BECCS may also be limited by nutrient demand, benefits (for example, improved soil carbon storage28), a heavy reli-
or by increased water use, particularly if feedstocks are irrigated ance on NETs in the future, if used as a means to allow continued
and when the additional water required for CCS is considered. use of fossil fuels in the present, is extremely risky, as our ability

48 NATURE CLIMATE CHANGE | VOL 6 | JANUARY 2016 | www.nature.com/natureclimatechange

© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2870 REVIEW ARTICLE
to stabilize the climate at <2 °C declines as cumulative emissions 25. Kheshgi, H. S. Sequestering atmospheric carbon dioxide by increasing ocean
increase8,35,92. A failure of NETs to deliver expected mitigation in the alkalinity. Energy 20, 915–922 (1995).
future, due to any combination of biophysical and economic limits 26. Smith, P. Soils and climate change. Curr. Opin. Environ. Sust. 4, 539–544 (2012).
27. Powlson, D. S. et al. Limited potential of no-till agriculture for climate change
examined here, leaves us with no ‘plan B’45. As this study shows, there mitigation. Nature Clim. Change 4, 678–683 (2014).
is no NET (or combination of NETs) currently available that could 28. Smith, P. et al. Greenhouse gas mitigation in agriculture. Phil. Trans. R. Soc. B
be implemented to meet the <2 °C target without significant impact 363, 789–813 (2008).
on either land, energy, water, nutrient, albedo or cost, and so ‘plan A’ 29. Woolf, D., Amonette, J. E., Street-Perrott. A., Lehmann, J. & Joseph, S.
must be to immediately and aggressively reduce GHG emissions. Sustainable biochar to mitigate global climate change. Nature Commun.
1, 56 (2010).
Received 23 July 2015; accepted 21 October 2015; 30. Schiermeier, Q. Convention discourages ocean fertilization. Nature
http://dx.doi.org/10.1038/news.2007.230 (2007).
published online 7 December 2015 31. Smith, P. et al. How much land based greenhouse gas mitigation can be
achieved without compromising food security and environmental goals?
References Global Change Biol. 19, 2285–2302 (2013).
1. Le Quéré, C. et al. The global carbon budget 1959–2011. Earth Syst. Sci. Data 32. Smith, P. et al. in Climate Change 2014: Mitigation of Climate Change
5, 165–185 (2013). (eds Edenhofer, O. et al.) Ch. 11 (IPCC, Cambridge Univ. Press, 2014).
2. Peters, G. P. et al. The challenge to keep global warming below 2 °C. 33. Smith, P. Soil carbon sequestration and biochar as negative emission
Nature Clim. Change 3, 4–6 (2013). technologies. Global Change Biol. (in the press).
A short article outlining the enormous challenge of meeting a 2 °C climate 34. Azar, C. et al. The feasibility of low CO2 concentration targets and the role of
stabilization target. bio-energy carbon-capture and storage. Clim. Change 100, 195–202 (2010).
3. IPCC Climate Change 2014: Mitigation of Climate Change 35. Kriegler, E. et al. What does the 2°C target imply for a global climate agreement
(eds Edenhofer, O. et al.) (Cambridge Univ. Press, 2014). in 2020? The LIMITS study on Durban Platform scenarios. Clim. Change Econ.
The latest IPCC Assessment Report on the mitigation options that are 04, 1340008 (2013).
available to stabilize the climate. 36. Strengers, B. J., Minnen, J. G. V. & Eickhout, B. The role of carbon
4. Tavoni, M. et al. Post-2020 climate agreements in the major economies assessed plantations in mitigating climate change: potentials and costs. Clim. Change
in the light of global models. Nature Clim. Change 5, 119–126 (2015). 88, 343–366 (2008).
5. Krey, V., Luderer, G., Clarke, L. & Kriegler, E. Getting from here to 37. Wise, M. et al. Implications of limiting CO2 concentrations for land use and
there — energy technology transformation pathways in the EMF27 scenarios. energy. Science 324, 1183–1186 (2009).
Clim. Change 123, 369–382 (2014). 38. Reilly, J. et al. Using land to mitigate climate change: hitting the target,
6. Edmonds, J. et al. Can radiative forcing be limited to 2.6 Wm−2 without recognizing the trade-offs. Environ. Sci. Technol. 46, 5672–5679 (2012).
negative emissions from bioenergy and CO2 capture and storage? Clim. Change 39. Humpenöder, F. et al. Investigating afforestation and bioenergy CCS as climate
118, 29–43 (2013). change mitigation strategies. Environ. Res. Lett. 9, 064029 (2014).
7. Van Vuuren, D. P. et al. The role of negative CO2 emissions for reaching 40. Chen, C. & Tavoni, M. Direct air capture of CO2 and climate stabilization:
2 °C — insights from integrated assessment modelling. Clim. Change a model based assessment. Clim. Change 118, 59–72 (2013).
118, 15–27 (2013). 41. Fuss, S. et al. Betting on negative emissions. Nature Clim. Change
8. Rogelj, J., McCollum, D. L., Reisinger, A., Meinshausen, M. 4, 850–853 (2014).
& Riahi, K. Probabilistic cost estimates for climate change mitigation. Nature 42. Kriegler E. et al. Is atmospheric carbon dioxide removal a game changer for
493, 79–83 (2013). climate change mitigation? Clima. Change 118, 45–57 (2013).
9. Clarke, L. et al. in Climate Change 2014: Mitigation of Climate Change 43. Sims, R. et al. in Climate Change 2014: Mitigation of Climate Change
(eds Edenhofer, O. et al.) Ch. 6 (IPCC, Cambridge Univ. Press, 2014). (eds Edenhofer, O. et al.) Ch. 8 (IPCC, Cambridge Univ. Press, 2014).
10. Riahi, K. et al. Locked into Copenhagen pledges — implications of short- 44. Kriegler, E. et al. Making or breaking climate targets: The AMPERE study on
term emission targets for the cost and feasibility of long-term climate goals. staged accession scenarios for climate policy. Technological Forecasting and
Technol. Forecast. Soc. 90, 8–23 (2015). Social Change, Part A 90, 24–44 (2015).
11. Obersteiner, M. et al. Managing climate risk. Science 294, 786–787 (2001). A study showing the impact of delay in implementation of mitigation on
12. Creutzig F. et al. Bioenergy and climate change mitigation: an assessment. climate stabilization over the course of the twenty-first century.
Global Change Biol. Bioenergy 7, 916–944 (2015) . 45. National Academy of Sciences. Climate Intervention: Carbon Dioxide Removal
13. Keith, D. Why capture CO2 from the atmosphere. Science 325, 1654–1655 (2009). and Reliable Sequestration (National Academies Press, 2015).
14. Socolow, R. et al. Direct air capture of CO2 with chemicals: a technology An in-depth report on carbon dioxide removal (equivalent to negative
assessment for the APS Panel on Public Affairs. (American Physical emissions) technologies.
Society, 2011). 46. Calvin, K. et al. A multi-model analysis of the regional and sectoral roles of
An in-depth assessment of direct air-capture technologies. bioenergy in near-and long-term CO2 emissions reduction. Clim. Change Econ.
15. Schuiling, R. D. & Krijgsman, P. Enhanced weathering: an effective and cheap 4, 1340014 (2013).
tool to sequester CO2. Climatic Change 74, 349–354 (2006). 47. Popp, A. et al. Land-use transition for bioenergy and climate stabilization:
16. Rau, G. H., Knauss, K. G., Langer, W. H. & Caldeira, K. Reducing energy- model comparison of drivers, impacts and interactions with other land use
related CO2 emissions using accelerated weathering of limestone. Energy based mitigation options. Clim. Change 123, 495–509 (2014).
32, 1471–1477 (2007). 48. Bonsch, M. et al. Trade-offs between land and water requirements
17. Köhler, P., Hartmann, J. & Wolf-Gladrow, D. A. Geoengineering potential of for large-scale bioenergy production. Global Change Biol. Bioenergy
artificially enhanced silicate weathering of olivine. Proc. Natl Acad. Sci. USA http://dx.doi.org/10.1111/gcbb.12226 (2014).
107, 20228–20233 (2010). 49. Wieder, W. R., Cleveland, C. C., Smith, W. K. & Todd-Brown, K. Future
18. Hartmann, J. & Kempe, S. What is the maximum potential for CO2 sequestration productivity and carbon storage limited by terrestrial nutrient availability.
by “stimulated” weathering on the global scale? Naturwissenschaften Nature Geosci. 8, 441–444 (2015).
95, 1159–1164 (2008). 50. Smith, L. J. & Torn, M. S. Ecological limits to terrestrial biological carbon
19. Kelemen, P. B. & Matter, J. M. In situ carbonation of peridotite for CO2 storage. dioxide removal. Clim. Change 118, 89–103 (2013).
Proc. Natl Acad. Sci. USA 105, 17295–17300 (2008). A study examining some ecological limits to land-based negative
20. Arora, V. K. & Montenegro, A. Small temperature benefits provided by realistic emission technologies.
afforestation efforts. Nature Geosci. 4, 514–518, (2011). 51. Scott, V., Haszeldine, R. S., Tett, S. F. B. & Oschlies, A. Fossil fuels in a trillion
21. Canadell, J. G. & Raupach, M. R. Managing forests for climate change tonne world. Nature Clim. Change 5, 419–423 (2015).
mitigation. Science 320, 1456–1457 (2008). 52. Benson, S. M. et al. in Global Energy Assessment Toward a Sustainable Future
22. Jackson, R. B. et al. Protecting climate with forests. Environ. Res. Lett. Ch. 13 (Cambridge Univ. Press, 2012).
3, 044006 (2008). 53. Zhao, K. & Jackson, R. B. Biophysical forcings of land-use changes
23. Sarmiento, J. L., Gruber, N., Brzezinski, M. A. & Dunne, J. P. High-latitude from potential forestry activities in North America. Ecol. Monogr.
controls of thermocline nutrients and low latitude biological productivity. 84, 329–353 (2014).
Nature 427, 56–60 (2004). 54. Zhuang, Q., Qin, Z & Chen, M. Biofuel, land and water: maize, switchgrass or
24. Joos, F., Sarmiento, J. L. & Siegenthaler, U. Estimates of the effect of Southern Miscanthus? Environ. Res. Lett. 8, 015020 (2013).
Ocean iron fertilization on atmospheric CO2 concentrations. Nature 55. IPCC Special Report on Renewable Energy Sources and Climate Change
349, 772–775 (1991). Mitigation (eds Edenhofer, O. et al.) (Cambridge Univ. Press, 2011).

NATURE CLIMATE CHANGE | VOL 6 | JANUARY 2016 | www.nature.com/natureclimatechange 49

© 2015 Macmillan Publishers Limited. All rights reserved

REVIEW ARTICLE NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2870
56. Renforth, P. The potential of enhanced weathering in the UK. 82. Oki, T. & Kanae, S. Global hydrological cycles and world water resources.
Int. J. Greenh. Gas Con. 10, 229–243 (2012). Science 313, 1068–1072 (2006).
57. Christian, D., Riche, A. B. & Yales, N. E. Growth, yield and mineral content 83. Postel, S. L., Daily, G. C. & Ehrlich, P. R. Human appropriation of renewable
of Miscanthus × giganteus grown as a biofuel for 14 successive harvests. freshwater. Science 271, 785–788 (1996).
Ind. Crop Prod. 28, 320–327 (2008). 84. Rockstrom, J. et al. The unfolding water drama in the Anthropocene: towards
58. St Clair, S., Hiller, J. & Smith, P. Estimating the pre-harvest greenhouse gas a resilience-based perspective on water for global sustainability. Ecohydrol.
costs of energy crop production. Biomass Bioenergy 32, 442–452 (2008). 7, 1249–1261 (2014).
59. Swann, A. L., Fung, I. Y., Levis, S., Bonan, G. B. & Doney, S. C. Changes in 85. Creutzig, F. et al. Reconciling top-down and bottom-up modelling on future
Arctic vegetation amplify high-latitude warming through the greenhouse bioenergy deployment. Nature Clim. Change 2, 320–327 (2012).
effect. Proc. Natl Acad. Sci. USA 107, 1295–1300 (2010). 86. Cowie, A. L., Smith, P. & Johnson, D. Does soil carbon loss in biomass
60. Ma, D., Notaro, M., Liu, Z. Y., Chen, G. S. & Liu, Y. Q. Simulated impacts of production systems negate the greenhouse benefits of bioenergy?
afforestation in East China monsoon region as modulated by ocean variability. Mitigation Adapt. Strateg. Global Chang. 11, 979–1002 (2006).
Clim. Dynam. 41, 2439–2450 (2013). 87. McCollum, D. et al. Energy investments under climate policy: a comparison of
61. Betts, A. K. & Ball, J. H. Albedo over the boreal forest. J. Geophys. Res. global models. Clim. Change Econ. 4, 1340010 (2013).
102, 28901–28910 (1997). 88. Creutzig, F. Economic and ecological views on climate change mitigation
62. Betts R. A. Offset of the potential carbon sink from boreal forestation by with bioenergy and negative emissions. Global Change Biol. Bioenergy
decreases in surface albedo. Nature 408, 187–190 (2007). http://dx.doi.org/10.1111/gcbb.12235 (2015).
63. Betts R. A., Falloon, P. D., Goldewijk, K. K. & Ramankutty, N. Biogeophysical 89. Kato, E. & Yamagata, Y. BECCS capability of dedicated bioenergy crops under a
effects of land use on climate: Model simulations of radiative forcing and large- future land-use scenario targeting net negative carbon emissions. Earth’s Future
scale temperature change. Agr. Forest Meteorol. 142, 216–233 (2007). 2, 421–439 (2014).
64. Schaeffer, M. et al. CO2 and albedo climate impacts of extratropical carbon and 90. Scott, V., Gilfillan, S., Markusson, N., Chalmers, H., Haszeldine, R. S. Last
biomass plantations. Global Biogeochem. Cy. 20, GB2020 (2006). chance for carbon capture and storage. Nature Clim. Change 3, 105–111 (2012).
65. Jones, A. D. et al. Greenhouse gas policy influences climate via direct effects of 91. Blackford, J. et al. Detection and impacts of leakage from sub-seafloor deep
land-use change. J. Clim. 26, 3657–3670 (2013). geological carbon dioxide storage. Nature Clim. Change 4, 1011–1016 (2014).
66. Mazzotti, M., Bociocchi, R., Desmond, M. J. & Socolow, R. Direct air capture 92. Luderer, G. et al. Economic mitigation challenges: how further delay closes the
of CO2 with chemicals: optimization of a two-loop hydroxide carbonate system door for achieving climate targets. Environ. Res. Lett. 8, 034033 (2013).
using a countercurrent air‑liquid contactor. Clim. Change 118, 119–135 (2013). A study showing the urgency of climate mitigation action.
67. Klein, D. et al. The value of bioenergy in low stabilization scenarios: an 93. Gasser, T., Guivarch, G., Tachiiri, K., Jones, C. D. & Ciais, P. Negative emissions
assessment using REMIND-MAgPIE. Clim. Change 123, 705–718 (2014). physically needed to keep global warming below 2°C. Nature Commun.
68. Lenton, T. M. in Geoengineering of the Climate System (eds Harrison, R. M. 6, 7958 (2015).
& Hester, R. E.) (Royal Society of Chemistry, 2014). 94. Boden, T. A., Marland, G. & Andres, R. J. Global, regional, and national fossil-
69. Haberl, H. et al. Quantifying and mapping the human appropriation of net fuel CO2 emissions. (Carbon Dioxide Information Analysis Center, 2015).
primary production in earth’s terrestrial ecosystems. Proc. Natl Acad. Sci. USA 95. Krey, V. et al. in Climate Change 2014: Mitigation of Climate Change
104, 12942–12947 (2007). (eds Edenhofer, O. et al.) Annex Ch. 2, 1308–1318 (IPCC, Cambridge Univ.
70. Schueler, V., Weddige, U., Beringer, T., Gamba, L. & Lamers, P. Global Press, 2014).
biomass potentials under sustainability restrictions defined by the European
Renewable Energy Directive 2009/28/EC. Global Change Biol. Bioenergy
5, 652–663 (2013). Acknowledgements
71. FAOSTAT (accessed 25 June 2015); http://faostat3.fao.org/home/E The views expressed herein are those of the authors, and do not represent those of a
particular governmental agency or interagency body. This analysis was initiated at a
72. Canadell, J. G., Schulze, E.-D. Global potential of biospheric carbon
Global Carbon Project meeting on NETs in Laxenburg, Austria, in April 2013 and
management for climate mitigation. Nature Commun. 5, 5282 (2014).
contributes to the MaGNET program (http://www.cger.nies.go.jp/gcp/magnet.html).
73. Smith, P. et al. Competition for land. Phil. Trans. R. Soc. B
G.P.P. was supported by the Norwegian Research Council (236296). C.D.J. was
365, 2941–2957 (2010).
supported by the Joint UK DECC/Defra Met Office Hadley Centre Climate Programme
74. Thompson, B. & Cohen, M. J. The impact of climate change and bioenergy on
(GA01101). J.G.C. acknowledges support from the Australian Climate Change Science
nutrition (Springer, 2012). Program. E.Ka. and Y.Y. were supported by the ERTDF (S-10) from the Ministry of the
75. Valentine, J., Clifton-Brown, J., Hastings, A., Robson, P, Allison G. & Environment, Japan.
Smith, P. Food vs. fuel: the use of land for lignocellulosic ‘next generation’
energy crops that minimise competition with primary food production.
Global Change Biol. Bioenergy 4, 1–19 (2012). Author contributions
76. Bustamante, M. et al. Co-benefits, trade-offs, barriers and policies for P.S. led the writing of the paper, with contributions from all authors in the inception of
greenhouse gas mitigation in the Agriculture, Forestry and Other Land Use the study and in writing the drafts. P.S. led the analysis with significant contributions
(AFOLU) sector. Global Change Biol. 20, 3270–3290 (2014). from S.J.D., F.C., S.F., J.M., B.G., R.B.J., A.C., E.Kr., D.M. and D.V.V. Figures were
77. Powell, T. W. R. & Lenton, T. M. Future carbon dioxide removal via conceptualized and produced by S.J.D., J.R., P.C., S.F., P.S., G.P., R.A. and J.M.
biomass energy constrained by agricultural efficiency and dietary trends.
Energy Environ. Sci. 5, 8116–8133 (2012).
78. Smith, P. Delivering food security without increasing pressure on land.
Additional information
Supplementary information is available in the online version of the paper. Reprints
Global Food Sec. 2, 18–23 (2013).
and permissions information is available online at www.nature.com/reprints.
79. Bajželj, B. et al. The importance of food demand management for climate
Correspondence and requests for materials should be addressed to P.S.
mitigation. Nature Clim. Change 4, 924–929 (2014).
80. Vörösmarty, C. J. Global water resources: vulnerability from climate change
and population growth. Science 289, 284–288 (2000). Competing financial interests
81. Stolaroff, J. K., Keith, D. W. & Lowry, G. V. Carbon dioxide capture from M.O. was given a share in Biorecro, a company that cooperates with BECCS projects
atmospheric air using sodium hydroxide spray. Environ. Sci. Technol. globally, honouring his pioneering work on BECCS. The other authors declare no
42, 2728–2735 (2008). competing financial interests.

50 NATURE CLIMATE CHANGE | VOL 6 | JANUARY 2016 | www.nature.com/natureclimatechange

© 2015 Macmillan Publishers Limited. All rights reserved