Available via license: CC BY 4.0
Content may be subject to copyright.
Bioenergy-induced land-use change emissions with
sectorally fragmented policies
Leon Merfort ( leon.merfort@pik-potsdam.de )
Potsdam Institute for Climate Impact Research https://orcid.org/0000-0003-1704-6892
Nico Bauer
Potsdam Institute for Climate Impact Research (PIK), Member of the Leibniz Association, Germany
https://orcid.org/0000-0002-0211-4162
Florian Humpenöder
Potsdam Institute for Climate Impact Research https://orcid.org/0000-0003-2927-9407
David Klein
Pik-Potsdam
Jessica Streer
Potsdam Institute for Climate Impact Research https://orcid.org/0000-0002-5279-4629
Alexander Popp
Potsdam Institute for Climate Impact Research
Gunnar Luderer
Potsdam Institute for Climate Impact Research https://orcid.org/0000-0002-9057-6155
Elmar Kriegler
Potsdam Institute for Climate Impact Research https://orcid.org/0000-0002-3307-2647
Article
Keywords: land-use emission policies, climate policy, bioenergy production
DOI: https://doi.org/10.21203/rs.3.rs-404716/v1
License: This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License
Bioenergy-induced land-use change
emissions with sectorally fragmented
policies
Authors: Leon Merfort*,a, Nico Bauera, Florian Humpenödera, David Kleina, Jessica Streflera, Alexander
Poppa, Gunnar Luderera,b, Elmar Krieglera,c
Affiliations:
aPotsdam Institute for Climate Impact Research (PIK), Member of the Leibniz Association, Germany
bTechnical University of Berlin, Germany
cUniversity of Potsdam, Germany
PIK- Potsdam Institute for Climate Impact Research
Telegraphenberg A 31, 14473 Potsdam
Postal address: P.O. Box 60 12 03, 14412 Potsdam
Technical University of Berlin
Straße des 17. Juni 135
10623 Berlin
Abstract 1
We assess the impact of different land-use emission policies within a broader climate policy framework on 2
bioenergy production and associated land-use carbon emissions. We use the global Integrated Assessment 3
Model REMIND-MAgPIE integrating the energy and land-use sectors and derive alternative climate change 4
mitigation scenarios over the 21st century. If CO2 emissions are regulated consistently across sectors, land-5
use change emissions of biofuels are limited to 12 kgCO2/GJ. Without land-use emission regulations 6
applied, bioenergy-induced emissions increase substantially and the emission factor per energy unit raises 7
to levels slightly below diesel combustion (64 kg CO2/GJ). Pricing these emissions on the level of bioenergy 8
consumption diminishes bioenergy deployment and the associated CO2 emissions, while failing to reduce 9
the average emission factor. Despite effective reduction of land-use emissions, undifferentiated 10
penalization of bioenergy use substantially increases mitigation costs. If supply side policies 11
comprehensively regulate direct and indirect emissions, bioenergy can be produced much more 12
sustainably. 13
Main 14
Introduction 15
To limit global mean temperature and to achieve the Paris climate targets, society needs to bring down 16
global carbon emissions to net zero and strongly reduce non-CO2 emissions1. Future cost-effective climate 17
change mitigation strategies often rely on large-scale bioenergy deployment2. The use of biofuels provides 18
a low carbon alternative to fossil fuel-based liquids as well as the possibility to enable carbon dioxide 19
removal from the atmosphere using bioenergy with carbon capture and storage (BECCS)3. The combined 20
ability of biofuels to overcome and compensate for decarbonization bottlenecks is a major driver of its 21
future large-scale deployment4–6. However, the relevance of bioenergy as a means to climate change 22
mitigation is also controversially discussed7–9, since its production will be in competition with other land-23
use (LU) activities and thus increases the already existing pressure on land-systems10,11. For example, large-24
scale bioenergy production can threaten biosphere integrity and biogeochemical flows, might increase 25
unsustainable freshwater use and may also lead to higher food prices11–13. Bioenergy also has substantially 26
higher specific land requirements than other renewable energy sources14. There is thus a risk that the 27
growing demand and willingness to pay for bioenergy induced by strengthened climate protection 28
measures hits a land-use sector that is already under pressure today12,15. In addition, there is the threat 29
that direct and indirect land-use change16,17 (LUC/iLUC) CO2 emissions associated with bioenergy 30
production can largely offset abated emissions18,19. 31
A broad range of studies has investigated LUC and iLUC emissions induced by bioenergy production at 32
different locations. They have identified vastly different emission factors (EF) ranging from 0 to 33
100 kg CO2/GJbiofuel20–23, which might be even higher than for oil derived diesel (74 kg CO2/GJ24). This broad 34
uncertainty reflects the heterogeneity of land types23 – characterized by stored carbon content and crop 35
yield rates – and the flexibility of land-use and initial land conditions25,26. Yet, these studies do not reflect 36
the interplay between future global climate policies and the allocation of land areas for bioenergy and 37
food production. Given that stringent climate policies are projected to be the main driver for bioenergy 38
production4, however, it is crucial to link the assessment of EFs to the future transformation pathways of 39
the energy-system. Our study assesses bioenergy EFs under a range of alternative climate change 40
mitigation policies and thereby closes this gap in the literature. 41
Most studies analyzing global climate change mitigation pathways using Integrated Assessment Models 42
(IAM) consider, among others, the benchmark of an idealized climate policy framework putting a uniform 43
carbon equivalent price on greenhouse gas (GHG) emissions from all sectors, sources and countries9. This 44
is an effective policy to avoid carbon emissions from the land-use sector due to bioenergy production. 45
However, in the current real-world situation, energy- and land-use policies are regionally and sectorally 46
fragmented. While many countries already started to implement GHG emission pricing in the energy 47
sector, institutional capacity building is much less developed in the agricultural and forestry sector27. The 48
already high challenges in implementing carbon pricing in the energy system give an indication of the 49
difficulties in pricing emissions from the land-use sector. Here additional technical and in particular 50
governance challenges related to monitoring, reporting and verification (MRV) need to be overcome. It is 51
also debated whether GHG prices should be the same in the energy and land-use sector, given differences 52
in abatement cost curves and distributional impacts28–30. The implicit assumption in IAMs of institutional 53
feasibility of LU mitigation has been criticized31, since fragmented or even completely absent LU-based 54
GHG regulatory schemes can lead to substantial emission leakage from regulated to unregulated 55
regions32,33 or sectors18,34,35 involving excess bioenergy production, both factors counteracting mitigation 56
efforts. Since LU-based regulatory schemes are hard to implement on a globally comprehensive level (i.e. 57
on the supply side), there has been the proposal to regulate bioenergy consumption directly via import 58
controls, volume caps or by attributing EFs to bioenergy – in particular biofuels (e.g. as part of EU 59
directives36). While there have been studies analyzing auxiliary policy frameworks apart from a uniform 60
carbon price37,38, none of them systematically compared the effectiveness of LU-based regulator schemes, 61
which may be weak or fragmented, with demand-side controls on bioenergy deployment in terms of EFs 62
related to bioenergy production and its land-use displacement effects. 63
By comparing the effectiveness of different bioenergy demand- and supply-side policies to reduce the EF 64
of bioenergy in climate change mitigation scenarios, we aim to answer the following research questions: 65
(1) If comprehensive LU-based regulation is not available, how effective can direct regulation of 66
bioenergy deployment be in harnessing bioenergy while limiting its adverse impacts? 67
(2) To what extent would only weak and fragmented LU regulation be able to limit the adverse 68
impacts of bioenergy deployment? 69
(3) How do different policy assumptions affect the level of carbon pricing necessary in the energy 70
sector to achieve the Paris climate target? 71
Modeling Framework 72
We apply the IAM framework REMIND-MAgPIE4,39,40 to derive climate change mitigation pathways that are 73
compatible with limiting warming to 2°C by setting a carbon budget of 1000 Gt CO2 to total energy- and 74
LU-based CO2 emissions from 2018 to 210041. Key socio-economic assumptions on population, GDP, 75
dietary choices and energy demand projections that drive the model results reflect a middle-of-the-road 76
scenario (Shared Socioeconomic Pathway 2 (SSP2))42. The coupling of the energy-system model 77
REMIND43,44 with the LU model MAgPIE45,46 allows for the analysis of feedback effects between bioenergy 78
demand, production and associated LUC emissions. By comparing these pathways with a counterfactual 79
scenario without bioenergy available for decarbonizing the energy sector, we extract the pure impact of 80
bioenergy production on total and specific LUC emissions and on the mitigation strategies of the energy 81
sector as well as on macro-economic costs. Specific emissions are thus calculated ex-post and we express 82
the emission factor EFex-post in terms of kg CO2 emitted per GJ of biofuel produced. 83
Policy Design (detailed description in methods section) 84
While all scenarios reach the same climate target by imposing a uniform price on GHGs in the energy 85
sector, they differ with respect to assumptions on bioenergy and LU-related policies (given in Table 1), 86
implying different carbon price levels. 87
There are two benchmark policy assumptions, between which the different alternative approaches unfold. 88
The scenario with a globally Uniform Carbon equivalent Price (UCP) in the energy- and LU sector marks the 89
first best policy option to comply with the climate policy target reaching a carbon price of 147 $/t CO2 in 90
2050. This policy is contrasted with a scenario where the LU sector is lacking any regulatory scheme for 91
controlling emissions (noLUreg), i.e. there is neither a price instrument on any type of land-use based GHG 92
emissions nor any widespread land-protection scheme. The 2050 carbon price here reaches a much higher 93
level of 291 $/t CO2. 94
As a first set of alternative supply-side policies we gradually explore the effect of different levels of 95
fragmentation between energy- and LU sector. This is represented by reduced GHG price levels on LU 96
sector-based emissions of 10-50% compared to the price on emissions from the energy sector 97
(LUprice10%-50%). Alongside these price-based supply-side policies we explore the efficacy of various 98
land-protection schemes, namely forest protection (protForest, protPrimforest), and the protection of 99
distinct focus areas (protBH, protCPD, protFF, and protLW). 100
Contrasted to supply-side policies that are difficult to implement at global scale, we examine how a 101
demand-side tax on bioenergy consumption can reduce LUC emissions. We analyze the effect of different 102
bioenergy tax levels, representing the uncertainty of potential bioenergy EFs (bioTax10-50, c.f. Table 1 for 103
a description of the tax levels). In addition, we explore scenarios, in which we impose a ban on bioenergy 104
imports on top of the different bioenergy tax levels (bioTaxNoImp10-50), since a considerable amount of 105
biomass can be consumed in regions other than the one where it is produced47. In particular, exports from 106
tropical regions with high carbon stocks might promote additional LUC. 107
Scenario
Policy design
Benchmark policies
UCP
Globally uniform* carbon price across energy- and land-use-sector
noLUreg
Globally uniform carbon price only within the energy sector; No regulation of land-use GHG emissions
Supply-side policies
LUprice10-50%
Globally uniform carbon price within the energy sector; Globally uniform carbon price within the land-
use sector at a level of 10-50% of the price in the energy system
protForest
Globally uniform carbon price only within the energy sector; Primary and secondary forests are protected
protPrimforest
Globally uniform carbon price only within the energy sector; Primary forests are protected
protBH
Globally uniform carbon price only within the energy sector; Biodiversity Hotspots are protected
protCPD
Globally uniform carbon price only within the energy sector; Centers of Plant Diversity are protected
protFF
Globally uniform carbon price only within the energy sector; Frontier Forests are protected
protLW
Globally uniform carbon price only within the energy sector; Last of the Wild areas are protected
Demand-side policies
bioTax10-50
Globally uniform carbon price only within the energy sector; No regulation of land-use GHG emissions;
Additionally bioenergy consumption is charged with a tax. The tax level is determined by the carbon price
multiplied with a predefined, fixed factor given in kgCO2 per GJPE of primary energy (“PE”) dry matter
biomass. E.g. bioTax20 stands for a policy charging bioenergy as if it had an EF of 20 kgCO 2/GJPE. Due to
conversion losses 10-50 kgCO 2/GJ PE correspond to 24-122 kgCO2/GJbiofuel and while the tax level is
expressed in units of primary energy, ex-post EFs are expressed in units of biofuel
for a better
comparability with other values from the literature.
bioTaxNoImp10-
50
Globally uniform carbon price only within the energy sector; No regulation of land-use emissions; a
Additionally a tax on bioenergy consumption of bioenergy as above; Bioenergy imports are prohibited
Table 1|Policy Design. Scenarios are divided into the two benchmark scenarios (UCP and noLUreg), scenarios with supply-side 108
policies (i.e. directly within the LU sector) and scenarios with demand-side policies (i.e. within the energy sector). *While carbon 109
prices are globally uniform from 2050 on, they differ between regions before 2050 to some extend for reasons of interregional 110
equity (see methods). 111
Ex-post emission factors 112
In the absence of restrictions on land-use emissions (noLUreg) cumulative (2020 to 2100) bioenergy-113
induced LUC emissions increase more than ten-fold from 44 Gt CO2 in the UCP case to 493 Gt CO2 (Fig. 114
1a). This is qualitatively similar but quantitatively more muted compared with Wise et al.18 Since global 115
bioenergy production only doubles to 236 EJ/yr, EFex-post on an 80 year time horizon increases from 12 kg 116
CO2/GJbiofuel to 64 kg CO2/GJbiofuel , which is only slightly smaller than the EF of diesel (Fig. 1b). 117
In between these benchmark scenarios supply-side and demand-side policies lead to different 118
consequences for bioenergy and LUC emissions. While even a small fraction of the energy systems’ carbon 119
price level applied to terrestrial GHG emissions (LUprice20%) is sufficient to reduce EFex-post to 120
33 kg CO2/GJbiofuel , a demand-side tax on bioenergy consumption is not affecting the specific average 121
emissions attributed to a unit of bioenergy, since emissions only decline as a consequence of reduced 122
demand. Interestingly prohibiting bioenergy imports is not effective at all, since the largest part of the 123
biomass is consumed domestically in most regions and a trade moratorium reduces overall production and 124
thus LUC emissions only to a small extend. 125
The impact of bioenergy on LUC emissions in the presence of land-protection schemes depends on the 126
precise areas that are removed from the land-pool available for bioenergy production and other 127
agricultural activities. While a policy protecting all forests resembles the UCP case to a large extend 128
(EFex-post = 24 CO2/GJbiofuel and emissions of only 107 Gt CO2 for protForest), removing only some focus 129
areas from the available land-pool has only very limited effect on reducing bioenergy-induced LUC 130
emissions and EFs from unregulated levels. 131
132
Fig. 1| Bioenergy-induced LUC emissions, bioenergy production and emission factors. (a) Emissions, given as the total global LUC 133
emissions and bioenergy, given as the averaged annual global production, are both evaluated for period from 2020 to 2100 and 134
shown for different policy settings. Besides the two benchmark scenarios (“bm.”), policies are grouped into “supply-side” and 135
“demand-side” policies. White bars indicate cummulative emissions in 2050 and the averaged annual bioenergy production until 136
2050, respectively. For reasons of clarity we only show a selection of policy settings, other scenarios are shown in Fig. S1 in the SI. 137
(b) Ex-post EFs are given per unit of biofuel produced for different policy settings. Reference EFs for diesel and natural gas are taken 138
from UBA24. For a comparison to EFs of N2O see Fig. S4 in the SI. 139
Spatial allocation of bioenergy production and emissions 140
Regarding the effect of distinct policies, we generally observe a considerable disconnect between the 141
spatial patterns of bioenergy production and LUC emissions. We find that – irrespective of the policy design 142
– a large fraction of LUC emissions does not originate from the land areas of bioenergy cultivation, but 143
occurs indirectly at formerly forested areas or pasture, where agricultural activity displaced by bioenergy 144
production is moved to (see Fig. 2). Those iLUC emissions as well as bioenergy plantations directly replacing 145
carbon-rich ecosystems contribute to high emissions factors (as for example in the northern regions of 146
South America for noLUreg, see Fig. 2 a,c,d). Without supply-side policies globally more than 85% of 147
additional emissions induced by bioenergy production originate from territories that together only 148
generate less than 16% of total biomass production (red and dark red wedges in Fig. 2 b). By contrast, the 149
main part of the bioenergy (more than three quarters across all policy settings) is being produced with a 150
direct emission factor of less than 37 kg CO2/GJbiofuel (half the EF of diesel, blue wedges in Fig 2.), directly 151
causing less than 10% of the total bioenergy-induced emissions if LU regulation policies are absent. 152
Therefore, by only accounting for direct LUC emissions within major bioenergy producing regions, only a 153
small fraction of attached emissions can be traced. Accordingly, the total iLUC emissions related to the 154
total bioenergy production are considerable and vary strongly with the regulatory framework. 155
This leads to two conclusions. First, the high flexibility of the LU sector in reallocating agricultural uses 156
makes iLUC emissions hard to avoid, although the absolute level depends on the underlying global LU 157
regulatory framework. Second, previous studies analyzing the direct LUC EF of bioenergy often suggested 158
that increasing bioenergy production is linked to increasing EF, implying that limiting bioenergy production 159
can also effectively limit EF (e.g. Daioglou et al23). This rests on the assumption that expanding agricultural 160
area due to bioenergy use proceeds along the lines of least marginal EF of land conversion. However, while 161
such an allocation would be optimal from a sustainability perspective it is not the allocation that emerges 162
in the land-use sector, as the EF is not the main criterion for allocating crop land by means of economic 163
choices. As a consequence of this, a demand-side bioenergy tax reducing the overall consumption of 164
energy crops does not automatically lead to sparing areas with high carbon stocks, as the allocation of 165
emissions by EFs is not affected by the tax (compare noLUreg and bioTax10, Fig. 2 b). 166
167
Fig. 2| Spatial allocation of LUC CO2 emissions and bioenergy production. Panel (a) shows a spatially disaggregated map of 168
bioenergy EFs that emerge in the absence of LU regulation (noLUreg). There are territories, where bioenergy is being produced 169
without additional LUC emissions at the place of production (bright blue areas). Here bioenergy is either being produced on 170
marginal or abandoned land or on land, where it displaces other agricultural activities. On the other hand, natural vegetation can 171
be converted to agricultural land to balance the production of agricultural goods that were displaced by bioenergy (dark red areas, 172
iLUC emissions). Other territories are classified by
ex-post
loc
, given by the ratio of bioenergy-induced emissions to bioenergy 173
production (for reference, 74 kg CO2/GJ is the EF of diesel24). In panel (b) the sizes of the pie charts reflect the total global 174
bioenergy-induced LUC emissions and the global averaged annual bioenery production, respectively, for different policy 175
assumptions (other scenarios are shown in Fig. S5, SI). The sizes of the wedges reflect the amount of emissions and bioenergy 176
production and are color-coded according to the associated EFs as described in (a). Panels (c) and (d) show the spatial distribution 177
of bioenergy production and bioenergy-induced emissions, respectively, for the example of South America. In all four panels 178
quantities are cumulated over the period between 2020 and 2100. See methods section for a description of the analysis of
ex-post
loc
179
and the SI for figures of the other policy assumptions, including maps of baseline (not bioenergy-related) emissions. 180
Components of CO2 emissions and the role of BECCS 181
Regarding the composition of total cumulated CO2 emissions, we observe vastly different allocations of 182
the carbon budget for the varying policy assumptions. While cumulated emissions induced by bioenergy 183
production up until 2100 will be offset by deploying carbon removal using BECCS and direct air capture 184
with carbon storage (DACCS) technologies (Fig. 3a), in in the absence of comprehensive LU sector emission 185
regulations a large fraction of BECCS removals will be offset by the additional LUC emissions (Fig. 3b). For 186
instance, if regulations only comprise incomplete land protection schemes or a bioenergy demand-side 187
tax, LUC emissions are relatively high in relation to the bioenergy production reflected by the high emission 188
factor (44 to over 60 kg CO2/GJbiof uel). In such a regulatory framework, carbon removal by BECCS is largely 189
used to offset these emissions. Hence, due to the heterogeneity and flexibility of the LU sector, incomplete 190
regulation of the additional emissions caused by biomass production implies that bioenergy is not 191
necessarily carbon negative, if combined with CCS (or at least only with a poor efficacy), but rather only 192
carbon neutral. Without LU mitigation, only 15% of carbon dioxide removal from BECCS remain. Before 193
2050 cumulated bioenergy-induced LUC emissions even exceed BECCS savings by far for all policy settings 194
except for a uniform carbon price (Fig. S6, SI). Note that this only compares LUC emissions with carbon 195
removal, but does not account for fossil fuel substitution. 196
In comparison to the UCP policy, we furthermore observe that cumulated energy system emissions need 197
to be reduced in scenarios with high LUC emissions to balance the total budget. The additional biomass is 198
then used to accelerate the phase out of fossil fuels, particularly oil (Fig. S7, SI). 199
If bioenergy is priced on the demand-side in the range of LUC emissions caused on the supply-side, LUC 200
emissions decrease with increasing tax level, but the reduced demand for bioenergy enforces a stronger 201
and faster electrification in comparison to both the UCP and the noLUreg scenario (Fig. S8, SI). At the same 202
time the share of emissions from the transport sector increases due to the lack of biofuels. The dwindling 203
availability of biomass even leads to higher CO2 prices (Fig. 3c) compared to the already high prices in 204
noLUreg, which makes DACCS competitive as a means to compensate for residual emissions. 205
It is also worth noting that even a comparatively small carbon price on LU-sector-based emissions 206
(LUprice10%) is sufficient to abate most of the non-bioenergy-related LUC CO2 emissions (from 235 in 207
noLUreg to 56 Gt CO2). At the same time, the required carbon price is reduced by 35% from 291 to 193 $/t 208
CO2 in 2050 (Fig. 3c). 209
210
Fig. 3| Composition emissions, BECCS efficiency and carbon prices. (a) Composition of total anthropogenic CO2 emissions, given 211
for different policy assumptions cumulated from 2020 until 2100. Black dots refer to the net totals. LUC emissions not related to 212
bioenergy production comprise CO2 LUC emission from all other agricultural activities. Bioenergy from residues is assumed to be 213
carbon neutral. For the calculation of the shares please refer to the methods section. Composition of the other scenarios are shown 214
in Fig. S2, in the SI. (b) The BECCS efficiency factor
BECCS
is an indicator of how much of the sequestered carbon is virtually removed 215
from the atmosphere if bioenergy-induced LUC emissions are subtracted. For instance,
BECCS
=15% for noLUreg implies that 216
only 15% of the CDR savings are effectively removed from the atmosphere, as the remaing 85% are offset by LUC emissions. 217
(c) Shown are energy system GHG prices in the year 2050. After a phase in period, prices are equal across regions from 2050 on. 218
Discussion and conclusion 219
In our study we analyzed the impact of bioenergy on LUC emissions in different climate policy settings, 220
comparing supply-side measures controlling land-use allocation with demand-side measures controlling 221
bioenergy use. We showed that in a scenario, where climate policy creates a large demand for bioenergy, 222
the specific emissions attributed to a unit of bioenergy produced can be lowered by supply-side emissions 223
regulations in the land use sector. Regulations of bioenergy demand reduce deployment level, but fail to 224
induce a reduction of EF, thus resulting in substantially higher overall mitigation effort and carbon prices 225
to reach climate targets. For instance, increasing the price on terrestrial carbon can decrease the EF, while 226
a bioenergy tax on consumption only reduces the total bioenergy quantity, keeping the EF virtually 227
constant at a high level. A demand-side bioenergy tax fails to steer LUC decisions towards areas with low 228
EF and is not suitable to emulate the uniform carbon price regime across all sectors. 229
In order to fully capture the impact on direct and indirect land use emissions in the absence of direct 230
regulation of land use, a demand-side bioenergy tax would need to be applied to approximately 231
25 kg CO2/GJPE at the primary energy level (61 kg CO2/GJbiofuel with 41% conversion efficiency), in order to 232
value fossil and terrestrial emissions equally on a 80-year time horizon. Computed on a shorter time 233
horizon, the EF to be applied is substantially higher (200 kg CO2/GJbiofuel if evaluated until 2050, see Fig. 234
S3, SI), since most emissions occur upfront, while by far the largest part of the bioenergy is being produced 235
after 2050. This EF for biofuels is even higher than values identified in dedicated studies, which are mostly 236
found to be less than 100 kg CO2/GJPE on a 30-year time horizon21 for biofuels from thermo-chemical 237
conversion. 238
On the supply side, however, implementing a land-use emission carbon price of only 20% of the price in 239
the energy-system (33$/t CO2 in 2050) eliminates almost half the bioenergy-induced LUC emissions. A full 240
forest protection scheme is also effective, while conserving only specific land-types is not sufficient due to 241
the flexibility to move agricultural activities35. In combination with the ineffectiveness of demand-side 242
measures to reduce the EF of bioenergy, this is a strong argument for a direct regulation of all LU-sector-243
based emission fluxes as a means to mitigate bioenergy-related LUC emissions. This is in particular 244
compelling, since GHG emissions associated with overall food production need to be reduced substantially 245
to achieve the 1.5° or well below 2° target48. It is important to note that the focus of this study was purely 246
on CO2 LUC emissions. To assess the whole value of bioenergy for climate change mitigation strategies 247
(including also e.g. N2O emissions from fertilization of bioenergy crops49), other adverse side effects (such 248
as unsustainable freshwater use or higher food prices) but also benefits from fossil fuel substitution need 249
to be considered as well. 250
Our study confirms that LUC emission pricing is an effective and efficient instrument to regulate LUC 251
emissions even under large-scale deployment of bioenergy. However, we also show that those policies 252
cannot be emulated by demand-side regulation of bioenergy use, raising the question to what extent the 253
land-use sector can be effectively regulated to make large-scale bioenergy use sustainable. The literature 254
points to numerous challenges for regulating land use, ranging from MRV to the need for huge institutional 255
capacity50–52. Moreover, the distributional implications of regulating land-use emissions affect land tenure 256
and livelihoods, raising strong equity and political economy concerns28,29. Hence, the policy challenge is to 257
either comprehensively regulate the LU sector and produce biomass at scale or reduce bioenergy demand. 258
The main driving force behind this challenge is the huge demand for non-electric energy, particularly 259
transport fuels. Thus, broad and deep electrification of end-uses would lower the pressure on the land-260
system and bypass the regulatory gaps in the land-use sector53. 261
Methods 262
General. To assess the impact of bioenergy and to fully cover feedback effects between LUC CO2 emissions 263
and bioenergy-demand, we use the coupled integrated assessment modeling framework REMIND-264
MAgPIE. 265
REMIND 2.1.2 is an open source global multi-regional Ramsey-type general equilibrium model of economic 266
growth with a detailed representation of the energy sector, hard-coupled to the macro-economic core43,44. 267
Using optimization methods, it finds a market equilibrium while maximizing intertemporal global welfare. 268
Via different conversion routes REMIND represents the supply, trade and conversion of biomass 269
feedstocks along the value chain to final energy carriers along with relevant GHG emissions and removals. 270
Therefore, the REMIND model values the energy and the carbon content of biomass feedstocks given the 271
market conditions and the regulatory framework. In climate change mitigation scenarios most of the 272
biomass is converted into bio-liquids. 273
MAgPIE v4.2.1 is an open source global multi-regional partial equilibrium model of the land-use sector that 274
models land-use dynamics spatially explicitly using recursive dynamic optimization45,46. The model covers 275
two types of modern (2nd generation) bioenergy production, namely grassy and woody biomass. Since 276
irrigation of bioenergy crops leads to unsustainable freshwater use11, we only allowed for rain fed 277
production. 278
Both models are soft-coupled, balancing prices and quantities of bioenergy feedstocks and GHGs4. The 279
main policy instrument to meet a given climate target is a pricing of GHG emissions. GHG prices that are 280
by default applied to all types of GHGs from all sectors and sources are derived in REMIND and passed to 281
MAgPIE so as to meet the predefined GHG budget in 2100 of total energy- and LU-sector based CO2 282
emissions. All scenarios in this study are derived with middle of the road assumptions on socioeconomic 283
drivers (SSP2) and meet a global CO2 emissions budget of 1000 Gt CO 2 to total energy- and LU-based CO 2 284
emissions from 2018 to 2100, allowing for a temporary overshoot. This budget is derived by subtracting 285
100 Gt CO2 emissions due to earth system feedback from the remaining carbon budget of 1170 given in 286
Rogelj et al.41 (67th percentile for the 2°C target), arriving at 1070 Gt CO2. As safety margin, this value is 287
rounded down to 1000 Gt CO2. 288
Carbon prices. In the UCP scenario all types of GHG emissions from the energy and the LU sector are 289
charged with a uniform carbon equivalent price GHG(,) in [$/t CO] that is increasing with time . Prices 290
can differ between modeling regions before 2050 for reasons of inter-regional equity, but they will 291
eventually converge to a globally harmonized prices until 205039. In scenarios with a partial LU price 292
(LUprice10%-50%) the price on GHG emissions in the LU sector is reduced for every time step and every 293
modeling region to the corresponding fraction of the respective price level on energy system related GHG 294
emissions (e.g. to 10%). 295
In order to be consistent with the narrative of a largely unregulated Agriculture, Forestry and Other Land 296
Use sector (AFOLU), we assumed that in scenarios without a price on CO2 emissions from LUC (noLUreg, 297
protForest, protPrimforest, protBH, protCPD, protFF, protLW and bioTax scenarios) also non-CO2 GHGs are 298
exempted from the GHG price in the LU sector. This has the side-effect that these scenarios also involve 299
substantially higher non-CO2 GHG emissions from agricultural activities compared to scenarios with a 300
carbon price, in particular CH4 and N2O. As a result, radiative forcing levels and resulting global mean 301
temperature responses can differ between scenarios, even though cumulative CO2 emissions coincide. 302
However, since agricultural CH4 emissions are not related to bioenergy production and N2O emissions 303
from grassy bioenergy production are negligible compared to LUC CO2 emissions (see Fig. S4, SI, for a 304
comparison), we omit the effect of non-CO2 GHG emissions for assessing the impact of bioenergy. 305
Nevertheless, differences between scenarios in global mean temperature in 2100 as a result of varying LU-306
related CH4 and N2O emissions (derived with MAGICC 654), are only in the range of less than 0.2 K (see Fig. 307
S16, SI). 308
Land protection. In scenarios with explicit land-protection schemes (protForest, protPrimforest, protBH, 309
protCPD, protFF, protLW) we removed the respective areas from the land-pool that is potentially available 310
for any agricultural activities. In Fig. S9 - Fig. S14 in the SI, the protected areas are depicted. protForest is 311
a scenario in which all primary and secondary forests are protected, which is a total area of 3683 Mha, 312
while in protPrimforest only primary forests with a total area of 1339 Mha are removed from the land-313
pool. The other land-protection policies only affect some focus areas. In protBH Biodiversity Hotspots, in 314
protCPD Centers of Plant Diversity, in protFF Frontier Forests and in protLW Last Wild areas are protected55. 315
These focus areas cover areas of 909, 651, 1084 and 3635 Mha respectively. 316
Additionally, in all scenarios specific land areas are protected or dedicated for afforestation according to 317
the Nationally Determined Contributions (NDC) targets of the nations that are participating in the Paris 318
climate agreement. 319
Bioenergy tax. As explained above, the default policy assumption regarding the pricing of emissions is a 320
uniform carbon price on both energy- and LU-based GHG emissions. Emissions related to bioenergy-321
production are thus already penalized directly within the LU sector, which is why the energy system by 322
default treats bioenergy as a carbon neutral energy carrier. In the scenarios with demand-side policies 323
(bioTax) we assign an ex-ante emission factor (ex-ante) to bioenergy that should reflect potential 324
bioenergy-related GHG emissions on a global average. ex-ante represents emissions on a global average 325
and is equal for each economic region and time step . It directly transforms into a bioenergy tax bio(,) 326
via the price on GHGs GHG(,) 327
bio(,)=ex-ante GHG(,) [$/GJPE] 328
which is applied to every unit of dry matter biomass, i.e. at the level of primary energy (PE). Since the 329
literature and the results of the present study indicate that specific emissions attributed to a unit of 330
bioenergy are highly uncertain even on a global average, we explore the effect of different values of 331
ex-ante ranging from 10 to 50 kg CO2/GJPE, which translates to 24 to 122 kg CO2/GJbiofuel (41% energy 332
conversion efficiency). It is worth noting that ex-ante is in general not equal to the actual emissions that 333
are eventually attributed to bioenergy and which are derived ex-post from our scenarios (ex-post). 334
Please also note that in most other publications applying the REMIND model bioenergy is actually charged 335
with a “sustainability tax” that reduces the demand for bioenergy irrespective of the policy design to 336
reflect uncovered externalities, such as unsustainable water usage, food price increase, the loss of 337
biodiversity and nitrogen losses to the environment39. In the present study, however, we deactivated this 338
tax, since we wanted to assess the impact of bioenergy given a certain policy assumption in an otherwise 339
uncontrolled market. 340
Ex-post emission factor. Due to iLUC induced by bioenergy production it is intrinsically impossible to 341
disentangle LUC CO2 emissions related to bioenergy production from LUC emissions that result from other 342
agricultural activities such as an expansion of crop land or pasture. For each policy setting we therefore 343
first derive a counterfactual scenario that depicts a world, in which purpose grown bioenergy production 344
is not allowed (bioOff) – a similar approach has been applied for example in Daioglou et al.23 and Pehl et 345
al.19 By comparing the actual policy run, in which bioenergy production is activated (bioOn), with this 346
counterfactual scenario, we can reveal the effect that bioenergy has on the coupled energy-LU-system for 347
a given policy assumption . The ex-post emission factor comprising all LUC CO2 emissions attributed to 348
bioenergy production is then given by 349
ex-post() = bioOn() bioOff()
() [kg CO/GJ] 350
where bioOn and bioOff are the total LUC emissions that emerge over the period from 2020 to 2100 for 351
the scenario with bioenergy on and off, respectively. is the total amount of purpose grown 352
lignocellulosic biomass produced globally over the same period. Please note that while ex-post is usually 353
expressed in terms of CO2 emissions per unit of biofuel, to make it comparable with fossil fuels, the 354
thermo-chemical conversion to liquid fuels is subject to substantial conversion losses (the energy 355
conversion efficiency for second generation biofuels (Fischer-Tropsch diesel) is only 41%). Other energy 356
carriers derived from biomass, in particular electricity or hydrogen, exhibit different emission factors due 357
to different energy conversion efficiencies. 358
EFs are also evaluated spatially disaggregated. For our study the LU model MAgPIE was applied using 1000 359
distinct simulation units MAgPIE revealing individual patterns of agricultural activities. Each simulation unit 360
represents a cluster of aggregated 0.5-degree resolution grid cells with similar properties45,56 (see Fig. S15, 361
SI) and for each of them an EF can be calculated individually: 362
ex-post
loc
,,MAgPIE= bioOn,MAgPIE bioOff,MAgPIE
bioOn,MAgPIE [kg CO/GJ] 363
There are clusters of grid cells without bioenergy production (ex-post
loc = ), and others, for which the 364
difference in emissions to the counterfactual scenario is zero or even marginally negative, i.e. a simulation 365
unit MAgPIE with equal or less emissions than in the scenario without bioenergy. Here the ex-post
local is set 366
to zero. 367
Please note that, since the EFs are given as the ratio of emissions and bioenergy production, there is no 368
information on the total volume of each of these quantities in the different areas in Fig. 2a. The spatial 369
allocation of LUC emission and bioenergy production quantities is depicted in section “Spatial land-use 370
characteristics” of the SI. 371
It is also important to again highlight that EFs result from a comparison with a counterfactual scenario, in 372
which bioenergy is not used. This approach can lead to a situation, in which additional LUC emissions to 373
the counterfactual scenario are rather small, while baseline LUC emissions from the counterfactual 374
scenario (in the same simulation unit) are already substantial. Bioenergy production is then associated 375
with a relatively small ex-post
loc , even though the actually occurring emissions are large. However, since 376
these emissions also emerge in the baseline/counterfactual scenario, in this study they are not attributed 377
to bioenergy. 378
The BECCS efficiency factor. We defined the efficiency of the CDR potential of BECCS by 379
BECCS =1bioOn bioOff
BECCS, bioOn x 100%, 380
where BECCS, bioOn are all negative emissions associated with BECCS from purpose grown biomass. A 381
scenario without bioenergy-induced LUC would thus imply an efficiency of 100%, while in a scenario, in 382
which bioenergy-related emissions are equal to the CDR saving via BECCS, the efficiency is 0%. 383
Since bioenergy from residues is allowed in the counterfactual scenarios, we excluded the BECCS emission 384
savings related to residues from the calculation of BECCS. Please note that this efficiency factor is derived 385
to relate bioenergy-induced LUC emissions to the CDR potential of BECCS. It does, however, not cover 386
other benefits of bioenergy to the energy system, particularly the benefits of substituting fossil fuels by 387
biofuels. On the other hand bioenergy related emissions do not cover all negative effects, as described in 388
the paragraph on the bioenergy tax. 389
390
Acknowledgements 391
The research leading to these results has received funding from the German Federal Ministry of 392
Education and Research (BMBF, grant number 01LA1809A, DIPOL project). 393
Author contributions 394
Leon Merfort performed model experiments, analyzed the scenarios, produced the figures, and lead the 395
writing of the manuscript. Leon Merfort, Nico Bauer, Florian Humpenöder, Gunnar Luderer and Elmar 396
Kriegler designed the study, the scenarios and the analysis. All authors contributed to the development of 397
the models, the presented ideas, and to the text. 398
Competing interests 399
The authors declare no competing interests. 400
Code and data availability 401
REMIND is open source and available on GitHub. The model version used in this study is 2.1.2, which can 402
be downloaded at https://github.com/remindmodel/remind/releases/tag/v2.1.2. 403
MAgPIE is open source and available on GitHub. The model version used in this study is 4.2.1, which can 404
be downloaded at https://github.com/magpiemodel/magpie/releases/tag/v4.2.1. Documentation can be 405
found at https://rse.pik-potsdam.de/doc/magpie/4.2.1/. 406
The results of the scenarios shown in this paper will be archived at Zenodo upon publication of this 407
paper. 408
409
References 410
1. Allen, M. et al. Summary for Policymakers. In: Global Warming of 1.5 °C an IPCC special report. (2018). 411
2. Bauer, N. et al. Global energy sector emission reductions and bioenergy use: overview of the 412
bioenergy demand phase of the EMF-33 model comparison. Climatic Change (2018) 413
doi:10.1007/s10584-018-2226-y. 414
3. Klein, D. et al. The value of bioenergy in low stabilization scenarios: an assessment using REMIND-415
MAgPIE. Climatic Change 123, 705–718 (2014). 416
4. Bauer, N. et al. Bio-energy and CO2 emission reductions: an integrated land-use and energy sector 417
perspective. Climatic Change (2020) doi:10.1007/s10584-020-02895-z. 418
5. Rose, S. K. et al. Bioenergy in energy transformation and climate management. Climatic Change 123, 419
477–493 (2014). 420
6. Krey, V., Luderer, G., Clarke, L. & Kriegler, E. Getting from here to there – energy technology 421
transformation pathways in the EMF27 scenarios. Climatic Change 123, 369–382 (2014). 422
7. Jia, G. et al. Land–climate interactions. in Climate Change and Land: an IPCC special report on climate 423
change, desertification, land degradation, sustainable land management, food security, and 424
greenhouse gas fluxes in terrestrial ecosystems (eds. Shukla, P. R., Skea, J., Calvo Buendia, E. & 425
Masson-Delmotte, V.) (In press). 426
8. IPCC. Special Report on climate change, desertification, land degradation, sustainable land 427
management, food security, and greenhouse gas fluxes in terrestrial ecosystems (SRCCL). (2019). 428
9. Creutzig, F. et al. Reconciling top-down and bottom-up modelling on future bioenergy deployment. 429
Nature Clim Change 2, 320–327 (2012). 430
10. Popp, A. et al. The economic potential of bioenergy for climate change mitigation with special 431
attention given to implications for the land system. Environ. Res. Lett. 6, 034017 (2011). 432
11. Humpenöder, F. et al. Large-scale bioenergy production: how to resolve sustainability trade-offs? 433
Environ. Res. Lett. 13, 024011 (2018). 434
12. Heck, V., Gerten, D., Lucht, W. & Popp, A. Biomass-based negative emissions difficult to reconcile 435
with planetary boundaries. Nature Clim Change 8, 151–155 (2018). 436
13. Hof, C. et al. Bioenergy cropland expansion may offset positive effects of climate change 437
mitigation for global vertebrate diversity. PNAS 115, 13294–13299 (2018). 438
14. Luderer, G. et al. Environmental co-benefits and adverse side-effects of alternative power sector 439
decarbonization strategies. Nat Commun 10, 5229 (2019). 440
15. Gerten, D. et al. Feeding ten billion people is possible within four terrestrial planetary 441
boundaries. Nature Sustainability 3, 200–208 (2020). 442
16. Verstegen, J. A. et al. What can and can’t we say about indirect land-use change in Brazil using an 443
integrated economic – land-use change model? GCB Bioenergy 8, 561–578 (2016). 444
17. Arima, E. Y., Richards, P., Walker, R. & Caldas, M. M. Statistical confirmation of indirect land use 445
change in the Brazilian Amazon. Environ. Res. Lett. 6, 024010 (2011). 446
18. Wise, M. et al. Implications of Limiting CO2 Concentrations for Land Use and Energy. Science 324, 447
1183–1186 (2009). 448
19. Pehl, M. et al. Understanding future emissions from low-carbon power systems by integration of 449
life-cycle assessment and integrated energy modelling. Nature Energy 2, 939–945 (2017). 450
20. Daioglou, V. et al. Progress and barriers in understanding and preventing indirect land-use 451
change. Biofuels, Bioproducts and Biorefining 14, 924–934 (2020). 452
21. Warner, E., Zhang, Y., Inman, D. & Heath, G. Challenges in the estimation of greenhouse gas 453
emissions from biofuel-induced global land-use change. Biofuels, Bioproducts and Biorefining 8, 114–454
125 (2014). 455
22. Wicke, B., Verweij, P., Meijl, H. van, Vuuren, D. P. van & Faaij, A. P. Indirect land use change: 456
review of existing models and strategies for mitigation. Biofuels 3, 87–100 (2012). 457
23. Daioglou, V. et al. Greenhouse gas emission curves for advanced biofuel supply chains. Nature 458
Clim Change 7, 920–924 (2017). 459
24. Redaktionsassistenz 2, U. B. A. CO2-Emissionsfaktoren für fossile Brennstoffe. 460
(Umweltbundesamt, 2016). 461
25. Wicke, B., Dornburg, V., Junginger, M. & Faaij, A. Different palm oil production systems for 462
energy purposes and their greenhouse gas implications. Biomass and Bioenergy 32, 1322–1337 463
(2008). 464
26. Creutzig, F. et al. Bioenergy and climate change mitigation: an assessment. GCB Bioenergy 7, 465
916–944 (2015). 466
27. World Bank. State and Trends of Carbon Pricing 2020. 467
https://openknowledge.worldbank.org/handle/10986/33809 (2020). 468
28. Stevanović, M. et al. Mitigation Strategies for Greenhouse Gas Emissions from Agriculture and 469
Land-Use Change: Consequences for Food Prices. Environ. Sci. Technol. 51, 365–374 (2017). 470
29. Hasegawa, T. et al. Risk of increased food insecurity under stringent global climate change 471
mitigation policy. Nature Climate Change 8, 699–703 (2018). 472
30. Fujimori, S. et al. A multi-model assessment of food security implications of climate change 473
mitigation. Nature Sustainability 2, 386–396 (2019). 474
31. Creutzig, F. Economic and ecological views on climate change mitigation with bioenergy and 475
negative emissions. GCB Bioenergy 8, 4–10 (2016). 476
32. Otto, S. A. C. et al. Impact of fragmented emission reduction regimes on the energy market and 477
on CO2 emissions related to land use: A case study with China and the European Union as first 478
movers. Technological Forecasting and Social Change 90, Part A, 220–229 (2015). 479
33. González-Eguino, M., Capellán-Pérez, I., Arto, I., Ansuategi, A. & Markandya, A. Industrial and 480
terrestrial carbon leakage under climate policy fragmentation. Climate Policy 17, S148–S169 (2017). 481
34. Reilly, J. et al. Using Land To Mitigate Climate Change: Hitting the Target, Recognizing the Trade-482
offs. Environ. Sci. Technol. 46, 5672–5679 (2012). 483
35. Popp, A. et al. Land-use protection for climate change mitigation. Nature Climate Change 4, 484
1095–1098 (2014). 485
36. Directive (EU) 2018/2001 of the European Parliament and of the Council of 11 December 2018 on 486
the promotion of the use of energy from renewable sources (Text with EEA relevance.). 328 vol. OJ L 487
(2018). 488
37. Calvin, K. et al. Trade-offs of different land and bioenergy policies on the path to achieving 489
climate targets. Climatic Change 123, 691–704 (2014). 490
38. Calvin, K. et al. Near-term limits to mitigation: Challenges arising from contrary mitigation effects 491
from indirect land-use change and sulfur emissions. Energy Economics 42, 233–239 (2014). 492
39. Kriegler, E. et al. Fossil-fueled development (SSP5): An energy and resource intensive scenario 493
for the 21st century. Global Environmental Change 42, 297–315 (2017). 494
40. Bauer, N. et al. Quantification of an efficiency–sovereignty trade-off in climate policy. Nature 495
588, 261–266 (2020). 496
41. Rogelj, J. et al. Mitigation Pathways Compatible with 1.5°C in the Context of Sustainable 497
Development. 82. 498
42. Riahi, K. et al. The Shared Socioeconomic Pathways and their energy, land use, and greenhouse 499
gas emissions implications: An overview. Global Environmental Change 42, 153–168 (2017). 500
43. Baumstark, L., Luderer, G. & Kriegler, E. REMIND2.1: Transformation and innovation dynamics of 501
the energy-economic system within climate and sustainability limit. (in preparation). 502
44. Luderer, G. et al. REMIND - REgional Model of INvestments and Development. (Zenodo, 2020). 503
doi:10.5281/zenodo.4091409. 504
45. Dietrich, J. P. et al. MAgPIE 4 – a modular open-source framework for modeling global land 505
systems. Geoscientific Model Development 12, 1299–1317 (2019). 506
46. Dietrich, Jan Philipp et al. MAgPIE - An Open Source land-use modeling framework. (Zenodo, 507
2020). doi:10.5281/ZENODO.1418752. 508
47. Daioglou, V. et al. Implications of climate change mitigation strategies on international bioenergy 509
trade. Climatic Change (2020) doi:10.1007/s10584-020-02877-1. 510
48. Clark, M. A. et al. Global food system emissions could preclude achieving the 1.5° and 2°C 511
climate change targets. Science 370, 705–708 (2020). 512
49. Popp, A. et al. On sustainability of bioenergy production: Integrating co-emissions from 513
agricultural intensification. Biomass and Bioenergy 35, 4770–4780 (2011). 514
50. Ochieng, R. M., Visseren-Hamakers, I. J., Arts, B., Brockhaus, M. & Herold, M. Institutional 515
effectiveness of REDD+ MRV: Countries progress in implementing technical guidelines and good 516
governance requirements. Environmental Science & Policy 61, 42–52 (2016). 517
51. Wang, X. et al. Taking account of governance: Implications for land-use dynamics, food prices, 518
and trade patterns. Ecological Economics 122, 12–24 (2016). 519
52. Waller, L. et al. Contested framings of greenhouse gas removal and its feasibility: Social and 520
political dimensions. WIREs Climate Change 11, e649 (2020). 521
53. Luderer, G. et al. Accelerated electrification based on cheap renewables facilitates reaching Paris 522
Climate targets. Nat Energy 20 (in review). 523
54. Meinshausen, M., Wigley, T. M. L. & Raper, S. C. B. Emulating atmosphere-ocean and carbon 524
cycle models with a simpler model, MAGICC6 – Part 2: Applications. Atmospheric Chemistry and 525
Physics 11, 1457–1471 (2011). 526
55. Kreidenweis, U. et al. Pasture intensification is insufficient to relieve pressure on conservation 527
priority areas in open agricultural markets. Glob Change Biol 24, 3199–3213 (2018). 528
56. Dietrich, J. P., Popp, A. & Lotze-Campen, H. Reducing the loss of information and gaining 529
accuracy with clustering methods in a global land-use model. Ecological Modelling 263, 233–243 530
(2013). 531
57. Pachauri, R. K. et al. Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II 532
and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. (IPCC, 533
2014). 534
535
