Article https://doi.org/10.

1038/s41467-024-52424-0

Technology availability, sector policies and

behavioral change are complementary
strategies for achieving net-zero emissions
Received: 31 January 2024 Jessica Strefler 1 , Leon Merfort 1,2, Nico Bauer 1, Miodrag Stevanović 1,
Dennis Tänzler3, Florian Humpenöder 1, David Klein1, Gunnar Luderer 1,2,
Accepted: 5 September 2024
Michaja Pehl1, Robert C. Pietzcker 1, Alexander Popp 1,4, Renato Rodrigues 1,
Marianna Rottoli1 & Elmar Kriegler 1,5

Check for updates

1234567890():,;
1234567890():,;

In this study, we analyze the effects of technology availability, political coor-

dination, and behavioral change on transformation pathways toward net-zero
greenhouse gas emissions in the European Union by 2050. We implemented an
iterative stakeholder dialogue to co-design the scenarios that were calculated
using a global multi-regional energy-economy-land-climate model. We find
that in scenarios without behavioral change and with restriction of technolo-
gies, the target of greenhouse gas neutrality in the European Union cannot be
reached. Already a target of 200 Mt CO2eq/yr requires CO2 prices above 100
€/tCO2 in 2030 across all sectors in all scenarios. The required CO2 price can
increase to up to 450 €/tCO2 by 2030 if technologies are constrained, if no
complementary regulatory measures are implemented, and if changes in
consumer behavior towards a more sustainable lifestyle do not materialize.

The Sixth Assessment Report of the Intergovernmental Panel on Cli- changes across all sectors are much less common. In this study, we aim
mate Change (IPCC)1 has shown that the majority of 1.5 °C scenarios to close this gap by bringing together different technology options,
achieve net-zero CO2 emissions globally by mid-century. To adhere to policy measures, and behavioral changes in a coherent scenario set.
this target, the European Union (EU) adopted the European Green Improved and iterative stakeholder dialogue is crucial to improve
Deal2,3 and the EU Long-term low greenhouse gas emission develop- the scenarios’ relevance for policy making and to obtain scenarios that
ment strategy of the European Union and its member states4 setting the are not only technically, but also politically and socially feasible11,12. In
more ambitious goal of net-zero greenhouse gas (GHG) emissions by recent years, multi-stakeholder engagement including business, gov-
20505,6. While the target is accepted and supported among member ernment, civil society, and science has become a key ingredient for
states, the implementation is still widely debated. For instance, policy tackling the challenges of climate change mitigation13,14. For this study,
measures to achieve especially emission reductions within the Effort we implemented such an iterative stakeholder dialogue process with
Sharing Regulation (ESR) can include bans, taxes, subsidies, or stan- representatives from the groups mentioned above to first identify the
dards, and may vary between member states. The availability of some most relevant transformation measures in the buildings, industry,
technologies like carbon capture and storage (CCS), wind power, transport, and land sectors. We then grouped the measures across all
nuclear power, or bioenergy depends not only on technological sectors into (i) technology and innovation, (ii) political coordination,
development but also on social acceptance and political feasibility. and (iii) behavioral change, which represent three key dimensions of
While the impacts of technology options on climate change mitigation the debate that are emphasized to a different degree by different
strategies have been extensively studied7–10, more comprehensive stu- stakeholders. From these three dimensions of transformation, we
dies also including an interplay with different policies and behavioral derived a set of scenarios that was again vetted with the stakeholders

1
Potsdam Institute for Climate Impact Research (PIK), Member of the Leibniz Association, Potsdam, Germany. 2Global Energy Systems, Technische Universität
Berlin, Berlin, Germany. 3adelphi consult GmbH, Alt-Moabit 91, Berlin, Germany. 4Faculty of Organic Agricultural Sciences, University of Kassel,
Kassel, Germany. 5Faculty of Economics and Social Science, University of Potsdam, Potsdam, Germany. e-mail: strefl[email protected]

Nature Communications | (2024)15:8440 1

Article https://doi.org/10.1038/s41467-024-52424-0

to capture the most relevant narratives. A more detailed description of European Union cannot be reached. Already a target of 200 Mt CO2eq/
the stakeholder engagement is provided in Supplementary Note 1. yr requires CO2 prices above 100 €2020/tCO2 in 2030 across all sectors
Each dimension has two possible realizations, which stand for in all scenarios. The required CO2 price can increase to up to 450 €2020/
different possible technological developments, actions in politics, or tCO2 by 2030 if technologies are constrained, if no complementary
developments in society. In the “technology and innovation” dimen- regulatory measures are implemented, and if changes in consumer
sion, we contrast a “focus on GHG mitigation”, where all mitigation behavior towards a more sustainable lifestyle do not materialize.
technology options are available, with a “focus on social acceptance”,
where technologies with a lack of (perceived) support in the population, Results
such as CCS, nuclear, but also wind power, are restricted. In the “poli- Strong near-term ambition and sufficient carbon prices are
tical coordination” dimension, we contrast a “market-oriented” necessary
approach, with a cross-sector CO₂-eq price as the central measure, with We find that in all five scenarios, GHG emissions in the EU are
a “sector-oriented” approach, in which the CO₂-eq price is com- strongly and immediately reduced (Fig. 1a). However, we find that in
plemented by targeted sector policies. These sector policies can be scenarios without behavioral change (“Price oriented”) but with
modeled explicitly, e.g. a ban on internal combustion engines for light- restriction of technologies like CCS (“Focus social acceptance”), the
duty vehicles or oil and gas heating, or implicitly, e.g. by assuming a target of GHG neutrality in the EU could not be reached (S1 and S4).
lower hydrogen or electricity price as a proxy for subsidies. In the To still be able to compare all scenarios, the climate target was
“behavioral change” dimension, it is either assumed that consumer relaxed to 200 Mt CO2eq/yr residual emissions in 2050, which would
behavior is only based on price signals (“price-oriented”) or that there is have to be offset by additional carbon dioxide removal (CDR)
a broad shift towards low-carbon and more sustainable consumption of options not represented in the model. For comparison, the results of
goods and services, including modal shifts or changing dietary choices scenarios S2, S3, and S5 achieving the full GHG neutrality target are
(“value-oriented”). More detail including a list of all transformation discussed in Supplementary Note 3.
measures in the three dimensions is provided in Supplementary Note 2. While the emission trajectories are similar in all scenarios, the
In close cooperation with the stakeholders, we derived five sce- underlying transformation pathways differ. The necessary carbon
nario narratives from the potential combinations of the different rea- prices in 2030 to achieve the climate target range from 125 €2020/
lizations, which represent different approaches to the transformation tCO₂ (S2) to more than 450 €2020/tCO₂ (S4) (Fig. 1b). We find that
challenge (Table 1). The “policy steering approach” (S1) assumes that scenarios with supporting sector policies (S1-S3, red colors) show
only the subset of technologies that is perceived to be socially and consistently lower carbon prices than the “market oriented” scenarios
ecologically sustainable is deployed at a large scale. Targeted sector (S4-S5, blue colors). Also moving from limited to full technology
policies complement a carbon pricing scheme, and while the public availability or from price-oriented to value-oriented behavior can both
supports stringent climate policies, they do not change their behavior reduce the carbon price to a similar extent. The highest carbon price of
beyond price signals. The “behavioral approach” (S2) follows a similar 456 €2020/tCO₂ in 2030 in the acceptance-oriented scenario S4 can be
narrative but adds behavioral changes toward sustainability. The reduced by about 60% by either full inclusion of all technologies (194
“technology-oriented approach” (S3) assumes supporting sector poli- €2020/tCO₂ in S5), or by complementing the carbon price with the
cies as well as no behavioral change but allows for the use of the full set targeted sector policies (170 €2020/tCO₂ in S1). Augmenting the policy
of mitigation options. The “acceptance-oriented approach” (S4) and steering approach S1 either with full technology availability (S1 to S3)
the “market economy approach” (S5) both achieve their climate targets or a value-oriented behavioral change (S1 to S2) both lead to a further
only using carbon pricing mechanisms, without additional sector significant reduction of carbon prices by about 25-35% from 170
policies, and without behavioral changes. In terms of technologies, the €2020/tCO₂ to 125 €2020/tCO₂ and 109 €2020/tCO₂, respectively.
latter uses the full set of mitigation options while the “acceptance- Depending on the scenario, this would mean a moderate to large
oriented approach” relies only on technologies perceived to be socially increase of emission prices in the emissions trading system (ETS) from
and ecologically sustainable. The narratives are translated to scenarios above 90 €2020/tCO₂ in 2020, and a much larger increase of carbon
that aim to be both globally consistent with the 1.5 °C target and meet prices in the buildings and transport sector in the new emissions
the EU climate target of net-zero GHG emissions in 2050. The state-of- trading system ETS2, which are currently discussed to be capped at 45
the-art multi-regional energy-economy-land model with global cover- €2020/tCO₂. Carbon prices in a similar range have been found in
age REMIND-MAgPIE13–15 is extended and adapted to represent these previous studies12,15,16.
transformations and derive deep decarbonization pathways for the EU
consistent with the net-zero strategy, within a world aiming for a 1.5 °C Large-scale deployment of technologies can significantly reduce
limit of global mean temperature increase. carbon prices
In this work, we provide insights into the interplay between poli- In the “technology and innovation” dimension, technologies with a
cies, technology availability, and behavioral change, and their impact on (perceived) lack of public support are excluded or restricted, such as
land and energy systems and on selected economic, social, and envir- wind power, nuclear power, bioenergy, or CCS (scenarios S1, S2, S4).
onmental indicators informing the feasibility of the net zero transition. Limiting CCS implies a limitation of carbon dioxide removal (CDR),
We find that in scenarios without behavioral change and with restriction which is mainly supplied via bioenergy with CCS (BECCS). Direct air
of technologies, the target of GHG greenhouse gas neutrality in the capture with CCS (DACCS) is available, but not used in the model in the

Table 1 | Scenarios are built from different combinations of the three dimensions of emission reduction measures
Scenario Name Technology & Political Behavioral
Innovation Coordination Change
S1 Policy steering approach Focus social acceptance Sector oriented Price oriented
S2 Behavioral approach Focus social acceptance Sector oriented Value oriented
S3 Technology-oriented approach Focus GHG mitigation Sector oriented Price oriented
S4 Acceptance-oriented approach Focus social acceptance Market oriented Price oriented
S5 Market economy approach Focus GHG mitigation Market oriented Price oriented

Nature Communications | (2024)15:8440 2

Article https://doi.org/10.1038/s41467-024-52424-0

Fig. 1 | Key characteristics of transformation scenarios. a Annual greenhouse gas scenarios S4-S5, with full (S3, S5) or restricted (S1, S2, S4) technology availability.
emissions and b carbon prices in the European Union for all five scenarios. Red See Table 1 for the description of scenarios S1-S5.
colors represent the sector-oriented scenarios S1-S3, blue colors market-oriented

Fig. 2 | Key trade-off for the technology and innovation dimension in the trade-off between limited vs. full technology availability for sector-oriented (S1 vs.
European Union. a Cropland for bioenergy and b carbon capture and storage S3) and market-oriented (S4 vs. S5) scenarios. See Table 1 for the description of
(values shown for 2050) increase, while c carbon price (values shown for 2030) scenarios S1-S5.
decreases. We also indicate the respective values in 2020. Black arrows indicate the

EU28 before 2050 due to higher costs. Lower CDR availability leaves €2020/tCO₂ in S5 to 456 €2020/tCO₂ in S4) than in “sector oriented”
less leeway for residual emissions requiring more costly mitigation scenarios (36% from 125 €2020/tCO₂ in S3 to 170 €2020/tCO₂ in S1).
measures and therefore increasing the carbon price. This carbon price The additional sector policies already reduce the residual emissions
increase is higher in “market oriented” scenarios (135% from 194 and therefore reduce the reliance on CDR.

Nature Communications | (2024)15:8440 3

Article https://doi.org/10.1038/s41467-024-52424-0

a 2020 2030 2050

b 2020 2030 2050
Sector
Market Market Sector Market Sector Market Market Sector Market

Final Energy (Buildings) [EJ/yr]

200 15
Vehicle stock (Million)

10

100

0
0
S1 S2 S3 S4 S5 S1 S2 S3 S4 S5 S1 S2 S3 S4 S5 S1 S2 S3 S4 S5

Battery Electric Vehicels Combustion Engine Vehicles Type

Hydrogen Heat Liquids
Hybrid Electric Vehicles Natural Gas/H2 Vehicles Electricity Gases Solids

c 2020 2030 2050

d 2020 2030 2050

Sector Market Market Sector Market Sector Market Market Sector Market

40
30
Oil
H2 Production Gas
Buildings 30
Fossil fuels [EJ/yr]
Electricity [EJ/yr]

Coal
Industry
20 Transport

20

10
10

0 0
S1 S2 S3 S4 S5 S1 S2 S3 S4 S5 S1 S2 S3 S4 S5 S1 S2 S3 S4 S5

Fig. 3 | Impacts of sector policies on the energy system in the European Union (yellow) and phase-out of liquids (blue) and gases (grey) by 2050 in the sector-
in 2020, 2030, and 2050. a Licensed light duty vehicles by mode for all five oriented scenarios S1-S3. c Electricity demand by sector and d fossil fuel use by
scenarios. In the sector-oriented scenarios S1-S3 no new internal combustion source. Energy is given in exajoules per year. See Table 1 for a description of
engines (grey) are permitted after 2030, leading to a phase-out until 2050. b Final scenarios S1-S5.
energy demand in the buildings sector by source, showing higher electrification

The fundamental trade-off in the “technology and innovation” Targeted sector policies can reduce the CO2 price
dimension is between the acceptance of technologies such as CCS and With regard to “political coordination” we analyze the effects of
larger use of agricultural land for bioenergy supply on the one hand, additional sector policies leading to faster uptake of electro-mobility
and a more than doubled CO₂ price on the other hand (Fig. 2). CCS is (Fig. 3a) and a phase-out of fossil fuels in the transport and buildings
used not only for mitigation of industry process emissions, but also for sectors (Fig. 3b). Though these policies lead to significant emission
CDR via BECCS. Higher CDR potentials allow for some residual fossil reductions (scenarios S1-S3), they are not sufficient to achieve the
fuel emissions, especially in the industry and transport sectors (see climate targets and still rely on being complemented by high carbon
Supplementary Fig. 3a). This reduces the necessary CO₂ price, but prices to cover all sectors.
requires a higher CCS deployment and larger land areas allocated to These targeted sector policies lead to more direct electrification.
bioenergy production (Fig. 2). Bioenergy imports are not allowed in Phasing out internal combustion engines for light-duty vehicles leads
the scenarios to avoid land-use change emissions (see Methods sec- to almost complete electrification of this transportation market seg-
tion). This trade-off can be mitigated by targeted sector policies, as ment by 2050 (Fig. 3a). In our scenarios, this cannot be achieved with
they reduce residual emissions and therefore the need for CDR carbon prices alone, even at more than 450 €2020/tCO₂ in 2030 as in
requiring CCS and bioenergy. scenario S4. Instead, at these high carbon prices the share of energy

Nature Communications | (2024)15:8440 4

Article https://doi.org/10.1038/s41467-024-52424-0

that is not directly electrified is decarbonized via indirect electrifica- which suggests that a CO₂ price in combination with sector policies
tion using hydrogen or synthetic fuels. This increases hydrogen could be a more politically feasible option, even though it may not
demand beyond already high demands from the industry sector, which maximize economic efficiency.
raises questions regarding scalability and adds to the electricity
demand. Lifestyle changes have multiple benefits
In line with Luderer et al.10, enhanced electrification (Fig. 3c) In the “behavioral change” dimension we compare a scenario where
reduces the demand for fossil fuels (scenarios S1-S4 in Fig. 3d) and consumer choices are purely price-based (S1) with a scenario including
therefore reduces residual emissions as well as CDR demand (scenar- value-based consumer choices towards more sustainable options (S2)
ios S1–S4 in Supplementary Fig. 3). At the same time, energy security is (see Supplementary Table 1 for specific assumptions). These value-
increased due to lower dependence on oil and gas (Fig. 3d) which can based choices include dietary changes away from animal products
reduce import dependencies. Coal use is phased out almost com- towards more vegetables and nuts following the suggestions of the
pletely by 2030, and oil use is reduced by 85-94% by 2050 compared to EAT-Lancet commission20, lower energy consumption in households,
only 63% in scenario S5 (see also Supplementary Fig. 7). However, a and a switch from private cars towards public transport or bikes at
higher share of electrification requires more electricity and therefore scales similar to those discussed in the earlier literature21–24.
faster expansion of renewable energy (scenarios S1-S4 in Fig. 3c). In Comprehensive lifestyle changes result in significantly lower non-
these scenarios, wind energy annual new capacities would need to be CO₂ emissions from agricultural production (Supplementary Fig. 3),
about 4 times higher in 2030 than today (~70-85 GW/yr, Supplemen- leading to a lower demand for CO₂ removal from the atmosphere. The
tary Fig. 8a), and 10-11 times higher for solar energy (>100 GW/yr, lower non-CO₂ emissions reduce the CO₂ price required to achieve the
Supplementary Fig. 9a). climate target by one-third (Fig. 1b). Due to the reduced demand for
The reduction in CDR demand alleviates the trade-off between animal products, less pasture is needed so that they can be rededicated
cropland for bioenergy and CCS, and high carbon prices seen for the to cropland (Fig. 4) leading to lower food prices. Other additional
technology dimension (Fig. 2). As many of the supporting sector benefits include positive environmental and health effects associated
policies involve regulatory policies, the necessary carbon price to with agricultural production or diets that are less focused on animal
achieve the remaining emission reductions is decreased, even products, as well as lower consumer prices mainly due to the lower
though overall economic costs may be higher. The fundamental CO₂ price (Fig. 5). However, the low pressure on food systems only
trade-off between the two approaches is therefore primarily one of arises in combination with the assumption that energy produced from
social acceptance and political feasibility. A very high CO₂ price is biomass will be significantly curtailed due to lack of social acceptance.
accompanied by distributional effects that can jeopardize social If technologies to remove CO₂ from the atmosphere were fully sup-
acceptance and political feasibility17,18. These regressive distribu- ported and there was high demand for bioenergy crops, such low-
tional effects could be mitigated or even made progressive, e.g. by intensity agriculture would not be possible, and new trade-offs due to
appropriate redistribution of revenues from the CO₂ pricing18,19. In land-use competition could arise.
the present scenarios, targeted sector policies can significantly
reduce the CO₂ price needed to achieve the climate targets. How- Discussion
ever, they require stringent regulatory measures such as bans, which While all scenarios overachieve the EU climate target for 2030, they
must be socially accepted and politically implemented and are differ substantially in their challenges. Figure 5 summarizes the trade-
associated with distributional issues and legal risks. Yet some hard- offs between the three dimensions of transformation “technology &
to-abate emissions show little response to even very high CO₂ prices, innovation”, “political coordination”, and “behavioral change”.

a) b) c)
S1
Δ 2030
400 Δ 2030
S1
S1
Δ 2050
S2
S2
S2 S3
S3

300 Δ 2030 S4
S4
Δ 2050 S5
S5
Cropland
S3
Bioenergy areas
Δ 2030
200
Δ 2050 Δ 2050Pasture and rangeland

S1
S1
Other land
S4
Forest
S2
S2
Δ 2030
100 Δ 2050 S3
S3
S4
S4
S5
S5
S5
Δ 2030
Δ 2050 0 10 20 30
0
-20 -10 0 10 20
Land use area Agricultural price index
[Mha] Land use change [Mha] [change with respect to year 2020 in %]

Cropland Bioenergy areas Pasture and rangeland Other land Forest

Fig. 4 | Impacts of behavioral changes on the land-use sector in the rangeland (yellow) 65 Mha, other land (olive green) 69.4 Mha, forest (dark green)
European Union. a Land area in different land use pools in 2020 (cropland (brown) 163 Mha). b Change of land use in 2030 and 2050 and c Agricultural price index in
114.5 million hectares (Mha), bioenergy areas (bright green) 0.7 Mha, pasture and 2030 and 2050 compared to 2020. See Table 1 for a description of scenarios S1-S5.

Nature Communications | (2024)15:8440 5

Article https://doi.org/10.1038/s41467-024-52424-0

a) b)

c)

Technology and Political Behavioral

Innovation Coordination Change

Policy steering Focus social

S1 Sector-oriented Price-oriented
approach acceptance
Behavioral Focus social
S2 Sector-oriented Value-oriented
approach acceptance
Technology- Focus
S3 Sector-oriented Price-oriented
oriented approach GHG-Mitigation
Acceptance- Focus social
S4 Market-oriented Price-oriented
oriented approach acceptance
Market economy Focus
S5 Market-oriented Price-oriented
approach GHG-Mitigation

Fig. 5 | Transformation indicators for all five scenarios. The colored bars show electricity price, where the change between 2020 and 2025 is shown as electricity
a the CO2 price in 2030, b carbon removal in 2050 (disaggregated into bioenergy prices peak in 2025. The colored dots show the percentage change between 2020
with carbon capture and storage, re- and afforestation, and other), and c the per- and 2050. See Table 1 for the description of scenarios S1–S5.
centage change between 2020 and 2030 for all other indicators except for the

In general, behavioral changes of consumers have entirely posi- 740-1180 MtCO₂eq/yr must be balanced with CDR, including the 200 Mt
tive effects, corroborating earlier results21,22. They reduce not only the CO2/yr residual emissions that remain uncompensated in the main sce-
necessary CO2 price, but also the residual emissions especially from narios. If CDR is limited due to limited CCS, residual emissions have to be
non-CO2 gases which are hard to reduce otherwise, and therefore the reduced further, e.g. via lifestyle changes. If both, CDR and the reduction
reliance on CDR. This leads to lower consumer prices, lower energy of residual emissions are limited, GHG neutrality cannot be achieved.
demand, and positive environmental and health effects. Though these CO2 prices above 100 €2020/t CO2 in 2030 are needed in all
changes result from individual consumer choices and are difficult to scenarios to achieve the climate target. The required CO2 price can be
influence at sufficient breadth across society by policies, policies can limited to this level by allowing technologies with limited social
support consumption choices by structural changes and political acceptance such as CCS, by regulatory measures such as banning
action to enable the uptake of low-carbon choices, e.g. by reducing combustion engines and oil and gas heating, or by changing consumer
barriers like lack of information25. behavior towards a more sustainable lifestyle. If none of this occurs,
None of the scenarios can achieve the climate target without CDR. very high CO2 prices of over 450 €2020/t CO2 will be needed as early as
To achieve GHG neutrality, residual emissions in 2050 in the range of 2030. To achieve their climate target, the EU therefore needs to either

Nature Communications | (2024)15:8440 6

Article https://doi.org/10.1038/s41467-024-52424-0

accept the use of technologies like CCS, facilitate comprehensive can be traded internationally. For the determination of bioenergy
lifestyle changes, or accept very high costs. However, CO2 prices prices, as well as for the mapping of emissions from agriculture,
occurring in the real world will differ from the prices derived in this REMIND is coupled to the land use model MAgPIE (see below).
study for two reasons. First, the model assumes a first-best solution In addition to technologies for energy production and conver-
without market failures, with full global cooperation, and with perfect sion, there are technologies for CO2 capture and utilization (e.g., for
foresight. Second, the CO2 price depends on many assumptions, the production of synthetic fuels), as well as the option for CO2 sto-
especially technology costs, which are still uncertain for a range of new rage, needed e.g. for carbon dioxide removal technologies, specifically
technologies like many CDR options. bioenergy with carbon capture and storage (BECCS) and direct air
High CO2 prices lead to substantial increases in energy prices and capture with carbon storage (DACCS).
thus expenditures for many actors, but especially for poorer house- The transformation measures listed above are implemented
holds. The impacts on poor households can be addressed, e.g. via either explicitly as direct constraints (e.g. limits on CCS, bioenergy
redistribution of the CO2 revenues, but this requires additional policies production or bioenergy trade; phase out of fossil heating), implicitly
that are potentially difficult to implement. Nevertheless, a CO2 price has via costs (e.g. subsidies on electricity prices or hydrogen, or higher
the advantage that revenue is generated that can be redistributed to costs for curtailment and storage requirements for renewables as a
reduce or even convert regressive effects18,19. While removing limitations proxy for inhibited wind energy expansion), or by changing exogenous
for technologies and changes in consumer choices reduce economic assumptions (e.g. dietary changes, reduced energy demands, or modal
costs and therefore the CO2 price, regulatory sector policies may reduce switches in the transport sector).
the CO2 price, but still have economic costs if these sector policies do System boundaries and limitations: Since REMIND is a global
not address existing market failures. However, these economic costs do model, the spatial resolution is limited and the dynamics of individual
not show explicitly in the CO2 price, and there is no additional fiscal countries from the EU + UK region are not represented. Model results
income generated that could be used to mitigate distributional effects. It such as demand, prices, and emissions are therefore mean or total
would be an important area for future research to quantify the economic values for the entire EU + UK region. Trade of secondary and useful
costs associated with sector policies and compare them to the costs energy carriers is not represented. For example, Europe cannot import
resulting from higher CO2 prices. In the end, the choices between using hydrogen in the scenarios.
technologies like CCS and accepting higher CO2 prices, between chan- Furthermore, it is assumed that the policies implemented in the
ging consumer choices and accepting higher CO2 prices as well as EU under the various scenarios do not directly affect the rest of the
energy prices and greater reliance on CDR, and between very high CO2 world. For example, other countries do not respond to electrification
prices and more moderate CO2 prices in combination with regulatory policies, which means that in certain scenarios, for example, the EU
sector policies will be up to policymakers and societies. already has a very high share of electric cars, while countries such as
the U.S. and China still largely rely on internal combustion vehicles.
Methods This in turn influences the European transformation via world market
REMIND-MAgPIE integrated assessment modelling framework prices of primary energy sources.
For this study, we use the global multi-regional energy-economy-land-
climate model REMIND-MAgPIE 2.2-4.3.226–29 to derive cost-efficient The MAgPIE model
emission and technology pathways. In addition to the REMIND model MAgPIE (Model of Agricultural Production and its Impacts on the
with version 2.2, we implemented several new policy options specifi- Environment) is a modular, open-source framework developed for
cally tailored to Europe that represent the specific policies used in this modeling global land systems28,29. It integrates agro-economic and
study that are described in the introduction and in Supplementary biophysical constraints within the land-use sector to project optimal
Table 1. These changes, however, do not affect the fundamental model spatial patterns of agricultural production under global scenarios until
dynamics. REMIND-MAgPIE represents 12 subregions, namely the the end of the 21st century. Additionally, MAgPIE assesses the envir-
European Union including the United Kingdom, four individual coun- onmental implications of the land-use system. By doing so, it offers a
tries (China, India, Japan, United States of America) and seven aggre- comprehensive framework for investigating future pathways of land
gate regions (Canada, Australia, New Zealand; Latin America; Middle- system transformation. The holistic approach considers the synergies
East and North Africa; non-EU Europe; other Asia; reforming econo- and trade-offs between ecosystem services and sustainable develop-
mies; Sub-Saharan Africa). ment, allowing for a nuanced exploration of the complex interactions
within the land-use system.
The REMIND model MAgPIE is a flexible global land system model that operates at
REMIND26 is an open source27 global multi-regional general equilibrium multiple spatial resolutions. It encompasses three distinct spatial scales:
economic growth model. The macroeconomic core of the model is hard- First, World regions, which can be defined based on aggregated coun-
coupled to a detailed representation of the energy sector. The model tries, allowing for a broader regional perspective; second, spatial clus-
assumes that economic agents (i.e., private and government investors in ters, identified by similar local characteristics, aggregating input data
particular) have perfect foresight. For example, future price develop- from a 0.5° x 0.5° spatial grid resolution; and third, grid-level with a
ments - especially the level of the CO2 price - are anticipated. Under the resolution of 0.5°x0.5°, used for detailed biophysical inputs and dis-
premise that certain climate targets are achieved (here, residual emis- aggregation of outputs for land use patterns. As its input, MAgPIE
sions of 200 Mt CO2eq/yr for Europe in 2050, as well as a global incorporates detailed information on terrestrial carbon content, water
cumulative carbon budget of 500 GtCO2 from 2018 onwards, which is availability, and potential crop and pasture yields from a global gridded
consistent with a 1.5 °C target), REMIND determines an intertemporal vegetation and hydrology model LPJmL (Lund-Potsdam-Jena model with
Pareto optimum of global welfare. This means that the different sce- managed Land)30. The gridded inputs are integrated into a spatial cluster
narios are not future projections, but possible economically optimal based on proximity in potential crop yield projections for the purpose of
transformation paths based on a variety of assumptions. a non-linear program31. Furthermore, MAgPIE incorporates regional
To achieve the climate target, a wide range of different technol- economic characteristics, such as agricultural product demand32, tech-
ogies is available to convert primary energy into useful energy in the nological advancements33, and production costs.
end-use sectors. The energy sector is subdivided into the detailed end- MAgPIE is a partial equilibrium model of the agricultural sector
use sectors of buildings, industry, and transport, as well as the elec- with recursive dynamic optimization. It has a nonlinear objective
tricity sector. Primary energy carriers (coal, gas, oil, biomass, uranium) function for minimization of global agricultural production costs to

Nature Communications | (2024)15:8440 7

Article https://doi.org/10.1038/s41467-024-52424-0

the fulfillment of demand for agricultural products, subject to bio- MAgPIE at the same bioenergy and emissions prices that the demand
physical and socio-economic constraints MAgPIE determines agri- patterns were based on. Second, the emissions in REMIND emerging
cultural demand for 25 food categories through initial demand from pre-defined climate policy assumptions account for the GHG
trajectories obtained in a cross-country econometric regression ana- emissions from the land-use sector derived in MAgPIE under the
lysis based on population and income development, demographic emissions pricing and bioenergy use mandated by the same climate
structures, and anthropometric characteristics. The demand reacts to policy. The simultaneous equilibrium of bioenergy and emissions
income changes through iterative adjustments. The model encom- markets is established by an iteration of REMIND and MAgPIE simu-
passes various cost components, including factor requirements costs lations in which REMIND provides emissions prices and bioenergy
(capital, labor, fertilizer), irrigation costs (including investment in new demand to MAgPIE and receives land use emissions and bioenergy
infrastructure), land conversion costs, transportation costs to the prices from MAgPIE in return.
nearest market, investment costs for yield-enhancing technological The resource potential of bioenergy in REMIND is represented by
change (TC), and greenhouse gas emission tax under climate change regional bioenergy supply price curves that are updated (scaled) in the
mitigation scenarios. iteration process according to the price response of MAgPIE. The
Food demand is assessed by categorizing it into four distinct information is exchanged between REMIND and MAgPIE by region.”
product groups: animal-source foods, empty calories (including sugar,
oil, and alcohol), fruits and vegetables, and staples32. In scenarios Scenario design
where preference changes are considered, the estimation process The assumptions for these scenarios have been identified in a co-
focuses solely on caloric requirements, while exogenous assumptions production process11,13,14 with German stakeholders from policy, indus-
are utilized to determine dietary composition and account for food try, and civil society. This process involves various iterations, captures
waste. These assumptions are based on the transition towards a heal- the motivation and needs of a diverse set of stakeholders, and derives
thier and more sustainable diet (planetary health diet), as proposed by different scenarios that attempt to cater to these needs. In the stake-
the EAT-Lancet Commission20. holder process, we collected a range of transformation measures for the
MAgPIE optimally allocates land use under competing demand for five sectors power, transport, industry, buildings, agriculture, and land-
goods, forage, carbon storage, conservation, and environmental pro- use, many of which are still highly debated, e.g. concerning social
tection. Land use is broadly divided into cropland, forest, pasture, acceptance and political feasibility, but also concerning justice and
other natural land, and urban areas. The model calculates the following fairness or freedom, e.g. the balance between price signals and reg-
AFOLU greenhouse gas emissions: CO2 resulting from land use change ulatory policies. There was a broad request for an explicit accounting for
(includes change in soil and vegetation carbon stocks), CH4 resulting behavioral changes. Measures with a high level of agreement regarding
from enteric fermentation, rangeland management, and rice cultiva- their relevance and necessity were included with the same assumptions
tion, and N2O resulting from fertilization of agricultural soils. MAgPIE across all scenarios, e.g. expansion of renewable energy. Measures that
contains a fully dynamic and endogenous budget of the agricultural are more debated were included with varying assumptions in the sce-
nitrogen cycle34. Nitrogen emissions from agricultural soils are calcu- narios, e.g. the availability of carbon capture and storage (CCS), nuclear
lated based on the IPCC 2006 Tier 1 method. This method does not power, wind power, direct and indirect electrification of industrial pro-
distinguish between the different soil properties that emit N2O, but cesses, electro mobility, a phase-out of internal combustion engines and
applies a single emission factor for all soil types. fossil-based heating, precision agriculture, and a demand shift towards
The model includes a selection of policies that can be used to lower energy demand in transport and buildings as well as a dietary shift
achieve various sustainability goals. These include 1st and 2nd gen- towards less animal-based products20.
eration bioenergy, greenhouse gas emissions prices from land use The narratives were translated to scenarios that aimed to be both
change (CO2) and agricultural land uses (CH4, N2O), land use regula- globally consistent with the 1.5 °C target and meet the EU climate
tions, REDD+ measures, reforestation, environmental management target of net-zero GHG emissions in 2050 using REMIND-MAgPIE. All
protection, and agricultural trade policies. measures beyond carbon pricing were only applied to the EU in order
System Boundaries/Limitations: Economic assumptions are made to ensure comparability of scenarios, all other regions use carbon
at the level of 12 independent world regions, as in REMIND, with the pricing only. Carbon prices are regionally differentiated to represent
EU27 + UK as a stand-alone region, which is the focus of this study. different abilities to pay. To achieve their climate target, the EU is
Therefore, direct interpretation of modeling results at the EU country neither allowed to buy international emission offsets, nor import
level is not possible, and land use policy recommendations should be bioenergy to avoid inducing land-use change emissions. The interna-
understood for the EU as a whole. tional dimension is important in order to consistently present scarcity
MAgPIE models the food systems and land use sector, including all or abundance of resources such as bioenergy or fossil fuels and the
relevant market drivers and biophysical characteristics of the land. resulting prices, but also the consequences of technology diffusion.
However, some options and factors are not included in the model and
therefore cannot be analyzed in this study. These include, for example, Data availability
organic farming and various feedback effects of land management such The specific model runs and scenario data as well as plotting routines
as land degradation. The model includes the main options for mitigating for this study are archived at Zenodo under a CC-BY-4.0 license upon
GHG emissions in the land use sector (including afforestation, halting publication and is available under https://doi.org/10.5281/zenodo.
deforestation, bioenergy production, and less polluting agricultural 10552787. This includes the specific code used for this study as there
production practices), but is not comprehensive in this regard. In par- have been some changes from REMIND v2.2.
ticular, peatland and soil carbon management are not modeled in the
scenarios presented. Therefore, there may be additional emission Code availability
reduction potentials that go beyond this study. The REMIND code is available under the GNU Affero General Public
License, version 3 (AGPLv3) via GitHub (https://github.com/
Coupling of REMIND and MAgPIE remindmodel/remind/releases/tag/v2.2.0). The code used in this
The coupling of REMIND and MAgPIE was already described in Strefler study deviates slightly from version 2.2 and is therefore made avail-
et al. 20219: “REMIND and MAgPIE are soft linked to derive scenarios able at Zenodo as described above. The technical model doc-
with equilibrated bioenergy and emissions markets. In equilibrium, umentation is available under https://rse.pik-potsdam.de/doc/
bioenergy demand patterns computed by REMIND are fulfilled in remind/2.1.3/. The source code and input data of MAgPIE v.4.3.2

Nature Communications | (2024)15:8440 8

Article https://doi.org/10.1038/s41467-024-52424-0

(https://github.com/magpiemodel/magpie/releases/tag/v4.3.2) are 20. Willett, W. et al. Food in the Anthropocene: the EAT–Lancet Com-
openly available at https://zenodo.org/record/4624341. The techni- mission on healthy diets from sustainable food systems. Lancet
cal model documentation is available at https://rse.pik-potsdam.de/ 393, 447–492 (2019).
doc/magpie/4.3.2/. 21. Creutzig, F. et al. Demand-side solutions to climate change miti-
gation consistent with high levels of well-being. Nat. Clim. Change
References 12, 36–46 (2022).
1. Riahi, K. & Roberto S. Mitigation Pathways Compatible with Long- 22. Grubler, A. A low energy demand scenario for meeting the 1.5°C
term Goals. in Climate Change 2022: Mitigation of Climate Change. target and sustainable development goals. (2018).
Contribution of Working Group III to the Sixth Assessment Report of 23. Rottoli, M., Dirnaichner, A., Pietzcker, R., Schreyer, F. & Luderer, G.
the Intergovernmental Panel on Climate Change (Cambridge Uni- Alternative electrification pathways for light-duty vehicles in the
versity Press, Cambridge, UK and New York, NY, US, 2022). European transport sector. Transp. Res. Part Transp. Environ. 99,
2. European Commission. Communication on The European Green 103005 (2021).
Deal. https://ec.europa.eu/info/publications/communication- 24. Dirnaichner, A. et al. Life-cycle impacts from different dec-
european-green-deal_en (2019). arbonization pathways for the European car fleet. Environ. Res. Lett.
3. EU Commission. European Commission - Press release REPowerEU: 17, 044009 (2022).
A plan to rapidly reduce dependence on Russian fossil fuels and fast 25. Creutzig, F. et al. Demand, Services and Social Aspects of Mitiga-
forward the green transition. (2022). tion. in Climate Change 2022: Mitigation of Climate Change. Con-
4. European Union. Long-term low greenhouse gas emission devel- tribution of Working Group III to the Sixth Assessment Report of the
opment strategy of the European union and its member states. Intergovernmental Panel on Climate Change (Cambridge University
Submission to the UNFCCC on behalf of the European Union and its Press, Cambridge, UK and New York, NY, USA, 2022).
member states. (2020). 26. Baumstark, L. et al. REMIND2.1: Transformation and innovation
5. Van Soest, H. L., Den Elzen, M. G. J. & Van Vuuren, D. P. Net-zero dynamics of the energy-economic system within climate and sus-
emission targets for major emitting countries consistent with the tainability limits. Geosci. Model Dev. Discuss. 2021, 1–50 (2021).
Paris Agreement. Nat. Commun. 12, 2140 (2021). 27. Luderer, G. et al. REMIND - REgional Model of INvestments and
6. Fragkos, P. et al. Energy system transitions and low-carbon path- Development. Zenodo https://doi.org/10.5281/zenodo.
ways in Australia, Brazil, Canada, China, EU-28, India, Indonesia, 6794920 (2022).
Japan, Republic of Korea, Russia and the United States. Energy 216, 28. Dietrich, J. P. et al. MAgPIE 4 – a modular open-source framework
119385 (2021). for modeling global land systems. Geosci. Model Dev. 12, 1299–1317
7. Capros, P. et al. Energy-system modelling of the EU strategy (2019).
towards climate-neutrality. Energy Policy 134, 110960 (2019). 29. Dietrich, J. P. et al. MAgPIE - An Open Source land-use mod-
8. Clarke, L., Wei, Y.-M., De La Vega Navarro, A., Garg, A. & Hahmann, eling framework. Zenodo https://doi.org/10.5281/zenodo.
A. N. Energy Systems. in Climate Change 2022: Mitigation of Climate 4231467 (2020).
Change. Contribution of Working Group III to the Sixth Assessment 30. von Bloh, W. et al. Implementing the nitrogen cycle into the
Report of the Intergovernmental Panel on Climate Change (Cam- dynamic global vegetation, hydrology, and crop growth model
bridge University Press, Cambridge, UK and New York, NY, LPJmL (version 5.0). Geosci. Model Dev. 11, 2789–2812 (2018).
USA, 2022). 31. Dietrich, J. P., Popp, A. & Lotze-Campen, H. Reducing the loss of
9. Strefler, J. et al. Carbon dioxide removal technologies are not born information and gaining accuracy with clustering methods in a
equal. Environ. Res. Lett. 16, 074021 (2021). global land-use model. Ecol. Model. 263, 233–243 (2013).
10. Luderer, G. et al. Impact of declining renewable energy costs on 32. Bodirsky, B. L. et al. The ongoing nutrition transition thwarts long-
electrification in low-emission scenarios. Nat. Energy 7, 32–42 (2022). term targets for food security, public health and environmental
11. Schmid, E. & Knopf, B. Ambitious mitigation scenarios for Germany: protection. Sci. Rep. 10, 19778 (2020).
A participatory approach. Energy Policy 51, 662–672 (2012). 33. Dietrich, J. P., Schmitz, C., Lotze-Campen, H., Popp, A. & Müller, C.
12. Rodrigues, R. et al. Narrative-driven alternative roads to achieve mid- Forecasting technological change in agriculture—An endogenous
century CO2 net neutrality in Europe. Energy 239, 121908 (2022). implementation in a global land use model. Technol. Forecast. Soc.
13. Brouwer, H., Woodhill, J., Hemmati, M., Verhoosel, K. & van Vugt, S. Change 81, 236–249 (2014).
The MSP Guide: How to Design and Facilitate Multi-Stakeholder 34. Bodirsky, B. L. et al. N2O emissions from the global agricultural
Partnerships. (Practical Action Publishing, 2016). nitrogen cycle – current state and future scenarios. Biogeosciences
14. Fischer, K. et al. Climate Dialogue – Process-Management as a 9, 4169–4197 (2012).
Planning Tool. Process Optimisation, Mobilisation, and Commu-
nication in Local Climate Action. (2014). Acknowledgements
15. Pietzcker, R. C., Osorio, S. & Rodrigues, R. Tightening EU ETS targets The research leading to these results has received funding from the
in line with the European Green Deal: Impacts on the decarbonization German Federal Ministry of Education and Research (BMBF, grant
of the EU power sector. Appl. Energy 293, 116914 (2021). numbers 01LA1809A and 01LA1809B, DIPOL project, J.S., L.M., M.S.,
16. Sitarz, J., Pahle, M., Osorio, S., Luderer, G. & Pietzcker, R. EU carbon N.B., E.K., D.T.). This work was supported by the European Union’s Hor-
prices signal high policy credibility and farsighted actors. Nat. izon 2020 research and innovation program under grant number
Energy https://doi.org/10.1038/s41560-024-01505-x (2024). 821124561 (NAVIGATE project, F.H., E.K.) and by the BMBF under grant
17. Büchs, M., Bardsley, N. & Duwe, S. Who bears the brunt? Distribu- number 03SFK5A (ARIADNE project, R.P., M.P., R.R., M.R., G.L.).
tional effects of climate change mitigation policies. Crit. Soc. Policy
31, 285–307 (2011). Author contributions
18. Grainger, C. A. & Kolstad, C. D. Who pays a price on carbon? J.S. designed the research together with L.M., N.B., M.S., and E.K.; J.S.,
Environ. Resour. Econ. 46, 359–376 (2010). L.M., N.B., D.K., G.L., R.P., M.P., R.R., M.R. and E.K. contributed to
19. Feindt, S., Kornek, U., Labeaga, J. M., Sterner, T. & Ward, H. developing the energy–economy modelling. M.S., F.H. and A.P. con-
Understanding regressivity: Challenges and opportunities of Eur- tributed to the land-use modelling. L.M. and M.S. performed scenario
opean carbon pricing. Energy Econ. 103, 105550 (2021). modelling. J.S., L.M. and M.S. performed data analysis and created the

Nature Communications | (2024)15:8440 9

Article https://doi.org/10.1038/s41467-024-52424-0

figures. D.T. conceptualized the stakeholder dialogue and facilitated the Reprints and permissions information is available at
co-creation process. J.S. wrote the paper with input and feedback from http://www.nature.com/reprints
L.M., N.B., M.S., D.T., F.H., D.K., G.L., M.P., R.P., A.P., R.R., M.R., E.K.
Publisher’s note Springer Nature remains neutral with regard to jur-
Funding isdictional claims in published maps and institutional affiliations.
Open Access funding enabled and organized by Projekt DEAL.
Open Access This article is licensed under a Creative Commons
Competing interests Attribution 4.0 International License, which permits use, sharing,
The authors declare no competing interests. adaptation, distribution and reproduction in any medium or format, as
long as you give appropriate credit to the original author(s) and the
Additional information source, provide a link to the Creative Commons licence, and indicate if
Supplementary information The online version contains changes were made. The images or other third party material in this
supplementary material available at article are included in the article’s Creative Commons licence, unless
https://doi.org/10.1038/s41467-024-52424-0. indicated otherwise in a credit line to the material. If material is not
included in the article’s Creative Commons licence and your intended
Correspondence and requests for materials should be addressed to use is not permitted by statutory regulation or exceeds the permitted
Jessica Strefler. use, you will need to obtain permission directly from the copyright
holder. To view a copy of this licence, visit http://creativecommons.org/
Peer review information Nature Communications thanks the anon- licenses/by/4.0/.
ymous, reviewer(s) for their contribution to the peer review of this work.
A peer review file is available. © The Author(s) 2024

Nature Communications | (2024)15:8440 10