environmental science & policy 14 (2011) 248–257
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/envsci
Indirect land use change emissions related to EU biofuel
consumption: an analysis based on historical data
Koen P. Overmars *, Elke Stehfest, Jan P.M. Ros, Anne Gerdien Prins
PBL Netherlands Environmental Assessment Agency, PO Box 303, 3720 AH Bilthoven, The Netherlands1
article info
Published on line 22 January 2011
Keywords:
Indirect land use change (ILUC)
Bioenergy
Greenhouse gas emissions
Renewable energy directive
abstract
Biofuels have recently been promoted in policies as a way to reduce greenhouse gas
emissions from the transport sector. However, biofuel production in itself also induces
emissions directly as well as indirectly. This paper presents an explicit calculation of
indirect land use change (ILUC) emissions from EU biofuel consumption. The approach
includes a straightforward methodology for quantifying ILUC, based on assumptions and on
data that is readily available. The calculations show that ILUC emissions alone could shift
the CO2 balance for biofuels from reductions to more emissions relative to fossil fuels. This
calculation is largely based on historical data, which reduces uncertainty compared to
forward looking modelling approaches. However, some of the uncertainties remain. Advantage of this approach is that it can easily be reproduced, which may add to the
acceptability in the domain of policy making.
# 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
1.1.
General introduction
Biofuels have the potential to mitigate climate change by
reducing greenhouse gas (GHG) emissions compared to using
fossil fuels. Because carbon dioxide is removed from the
atmosphere by growing the feedstock used for biofuels, the
net carbon balance after consumption is practically zero.
However, in addition to the GHGs that are emitted during
consumption, GHGs are also emitted while growing and
producing biofuel. This means that the GHG balance
relative to using fossil fuels results in a reduction of less
than 100%.
In recent years, mitigation policies have been promoting
biofuels as one of the potential means to reduce GHG
emissions (EU, 2009; Pousa et al., 2007; US Congress, 2005).
In order for these policies to be effective, they must ensure
that GHG emissions are actually reduced, which requires the
introduction of sustainability criteria (CARB, 2009; EU, 2009). In
the EU, the sustainability criteria thus far have focused mainly
on the direct effects of the production chain of bioenergy
products. However, in addition to the effects from the
production chain, bioenergy may cause significant indirect
effects in other systems too (Fargione et al., 2008; Searchinger
et al., 2008). Examples of these indirect effects are indirect land
use change (ILUC) that leads to GHG emissions and biodiversity loss, and the impact on food prices that determine the
availability of food for the poor.
A general feature of the indirect effects is that they work
through systems that are larger and more complex than the
production chain of biofuel. The effects induced by the biofuel
system in these larger systems are intertwined with a variety
of other processes in such a way that they are difficult to
identify separately. Consequently, monitoring of the indirect
effects of individual biofuels directly is impossible, and
* Corresponding author. Tel.: +31 030 274 2803; fax: +31 030 274 44 79.
E-mail address: [email protected] (K.P. Overmars).
1
www.pbl.nl.
1462-9011/$ – see front matter # 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.envsci.2010.12.012
environmental science & policy 14 (2011) 248–257
249
Demand for land for the cultivation
of biomass
Conversion of
unproductive land
(unmanaged land)
Use of currently
productive land
(managed land)
Intensification
of agriculture
Conversion of
unproductive land
(unmanaged land)
Change in
consumption
Direct effects
Indirect effects
Fig. 1 – Direct land use for bioenergy and its indirect effects.
deducing the effects through modelling or other analyses is
also difficult. In addition, the indirect effects cannot be directly
attributable to producers or consumers due to the complexity
of the system.
Indirect effects must be quantified in order to be included in
policies. Generally spoken, there are two main routes that
could be taken to quantify the indirect effects of the
production and consumption of biofuels. One route is to
quantify the indirect effects based on historical data (i.e.
monitoring). The other route is to use model projections of
possible future pathways. The first route is not feasible since
the indirect effects cannot be monitored directly. However,
some other data can still be useful, for example data on
production and consumption of biofuel as well as on other
variables in the economic system and the land use system (e.g.
FAO data), which are affected by the biofuel system.
Assumptions have to be made in order to separate the
biofuel-induced effects from other effects. By adopting a set of
assumptions, for example on where ILUC will take place,
historical data can actually be used to estimate historic
indirect effects of bioenergy production. Such an approach of
combining historical data with assumptions is essentially a
simple ex-post model. This is the approach that is used in this
study.
In the second approach to quantify ILUC, often a combination of economic and environmental models is used to
determine the effects of different scenarios ex-ante in order
to foresee possible effects. Examples of such modelling
approaches are in: Al-Riffai et al. (2010), Birur et al. (2008),
Fabiosa et al. (2009) and Havlı́k et al. (in press).
1.2.
Indirect land use change
The cultivation of energy crops requires land. One option is to
use currently productive land and the other is to convert
unproductive land, which is land without any agricultural or
forestry production, i.e. nature areas (Fig. 1). The land use
change from the current use into use for biofuel crops is called
direct land use change and the effects that arise from that
direct land use change are called direct land use change
effects. With the conversion of unproductive land, the direct
land use change effects are very large compared to the indirect
effects. There is a clear, one-to-one relation between the
production of feedstock and the land use change emissions.
ILUC is absent or minimal (Fig. 1).2 With the option of
cultivating currently productive land, the original production
would have to be realised elsewhere either through the
conversion of land into agricultural use or through intensification of agriculture to increase yields. This ‘displacement’ of
agricultural production has been discussed extensively in the
literature in the last two years (e.g. Bindraban et al., 2009;
Fargione et al., 2008; Searchinger et al., 2008), and is generally
called the ILUC effect. Where Searchinger and Fargione mainly
report on the emissions of the land conversions, additional
emissions can also be induced on the land that has a more
intensive production. A third effect of using currently
productive land is a change in consumption, for example
reduced consumption due to higher land prices and therefore
higher food prices (see Fig. 1).
The mechanisms that determine the contribution of
intensification, land conversion and changes in consumption
to absorb the changes in the current situation depend on many
parameters that vary among commodities (or feedstocks),
countries and regions. Examples of these parameters are price
elasticity of consumption, availability of suitable land, trade
policies, national policies concerning the use of inputs or land
conversion, and the ability of farmers to buy inputs or invest in
the technologies required. The causal chain of effects in the
economic systems is the following. The price of the former
crop increases due to its diminished supply. This price
increase then serves as an incentive for farmers to grow
more of this commodity through intensification or expansion
and on the consumption side it causes consumers to consume
less, gives incentives to reduce losses in the food chain and
gives incentives to increase feed efficiency in the livestock
sector. Subsequently, the land use changes can lead to
changes in GHG emission and also biodiversity loss (e.g.
Searchinger et al., 2008).
1.3.
Objectives
The objective of this paper is to provide a methodology for
calculating the CO2 emissions from the ILUC for biofuel production based on historical data from national to global level
2
It is possible to imagine indirect land use changes from direct
use of nature areas since there is also a demand for nature areas
for tourism, nature protection, etc.
250
environmental science & policy 14 (2011) 248–257
and demonstrate this methodology for biofuel consumption
in the EU. We acknowledge that biofuel production generates
more indirect effects than just the effects of ILUC, and that ILUC
has more knock-on effects than just CO2 emissions. This paper
only focuses on the CO2 emissions from ILUC. As mentioned
above, in other studies agro-economic models are often applied
to estimate the direct and indirect land use implications of
biofuels. These models are complex. Often, some of the assumptions are not explicit and their results show a large variability.
Here, we aim to be as simple and transparent as possible.
2.
Methodology and data
In this paper, we quantify ILUC for cultivated biofuels
consumed in the EU in 2007.3 In this methodology section
we provide the data sources, assumptions and the calculation
methods. In some cases a range is presented to describe a
certain step of the calculation. By combining these values
systematically, the range of results gives an indication of the
uncertainty of the possible outcome.
We identify six steps in the approach.
1. EU biofuel consumption and share of EU-produced and EUimported biofuels during 2007
2. Feedstock and origin of the feedstock for the biofuels
3. Energy yields per hectare for the various feedstocks in their
region of origin taking the effect of by-products into
account
4. ILUC calculation with historical data on land expansion and
yield increase
5. Global land use conversions to agricultural land
6. CO2 emissions from land use conversions
1. EU biofuel consumption and share of EU-produced and EUimported biofuels during 2007
To determine the biofuel consumption in the European
Union, EUROSTAT data on 2007 were used (EUROSTAT, 2009).
The data also indicates that the final energy consumption in
transport is less than the production and imports of biofuels.
The EU both imports and exports biofuels. The assumption on
which of the biofuels are consumed and which of them are
exported influences the calculation because the ILUC impact
of imported biofuels may be different from EU-produced
biofuels. To define the share of EU-produced and EU-imported
biofuels in the final consumption, two opposite assumptions
were made: the first is that all imported biofuels are consumed
and EU-produced biofuels are exported, and the second is that
all EU-produced biofuels are consumed and excess imported
biofuels are exported again.
2. Feedstock and origin of the feedstock for the biofuels
In order to determine the land area necessary for the
production of the consumed biofuels, information is needed
3
Using waste products, such as food waste, is not taken into
account in this analysis. In the past, the contribution of waste to
biofuels has been low. This analysis has its focus on ILUC from
cultivated feedstocks.
about the types of feedstock used and where these are
produced. The latter is important since yields of feedstock can
vary between countries and regions. The feedstocks of EUproduced bioethanol are cereals (mainly wheat and barley)
and sugar beet, which together form the bulk of the feedstock.
Wheat and sugar beet are taken into account in the
calculations. The main feedstock of EU-produced biodiesel
is rapeseed (EU, 2008; FAO, 2008), with an estimated share of
approximately 84% (European Biomass Industry Association,
2010) to 90% (EU, 2006). Other biodiesel feedstocks are
sunflower oil, recycled oils and animal fats. Only rapeseed
is taken into account in the calculations. Country data of
imported bioethanol and biodiesel is based on data from the
World Ethanol and Biofuels Report (F.O. Licht’s, 2009). The
information on feedstocks is based on historical data.
Therefore, for this part of the calculation we do not include
different assumptions.
3. Energy yields per hectare for the various feedstocks in their
region of origin taking the effect of by-products into account
The third step completes the calculation of the net land
area necessary to produce the feedstock for the consumed
biofuels. In this step we use data on the required hectares to
produce a certain quantity of energy to convert the energy
consumption per feedstock, which is the result of steps 1 and
2, to hectares. We make use of assumptions on the use of byproducts (or co-products) to determine a minimum and
maximum net land use to produce the energy. If by-products
are not taken into account, the net land use is equal to the
actual area that the crop occupies. However, by-products can
be used as feed; therefore, part of the land can be attributed to
these by-products.
The starting point is the energy crop yields that can be
derived from monitoring data. The allocation of part of the
land use to by-products is more complicated. If these byproducts are applied as a source of energy, as is the case for
bagasse, the by-product of cane ethanol, allocation or
correction of the greenhouse gas balance is relatively
simple, and can be based on the energy content of both
products. However, by-products with high protein content
are often used as feed. Rapeseed meal and dried distillers’
grain with solubles (DDGS) from wheat are examples. In this
case, the protein-rich by-product has to be compared
somehow with the biofuel in order to determine which
part of the land use can be attributed to each product. A
substitution method is used to make this comparison. In
Europe, soy meal is an important source of protein in feed.
In many cases, substitution of soy meal by biofuel byproducts is quite likely. Data on the actual substitution are
not available. Therefore, we assume soy meal is replaced
and the land use for soy production can be reduced. This
land can be subtracted from the direct (gross) land that is
used for cropping the biofuel. Data and more details on this
method are available in ECOFYS (2009), Eickhout et al. (2008)
and JRC/CONCAWE/EUCAR (2007). Actually this can trigger
additional effects, since the biofuel by-product will influence the feed market. Because we do not have enough
information on the actual use of the by-products and their
impact on the feed market we include two levels of
substitution which together give an indication of the
environmental science & policy 14 (2011) 248–257
uncertainty of these additional effects, including imperfect
use of the by-product.
4. ILUC calculation with historical data on land expansion and
yield increase
The assumptions on land use are the following:
a. The Renewable Energy Directive (RED) (EU, 2009) identifies
certain land use types not permitted to convert to biofuel
crops.4 In our calculation, we assume that these direct
conversions do not take place, neither within Europe nor
outside of it. Thus, no direct land use change for biofuels
occurs on these lands and therefore all land use changes on
these lands are ILUC.
b. All former production (tonnes) of the land occupied by
biofuel crops is compensated for by either land expansion
or increased yields. The change in consumption (Fig. 1) is
assumed to be zero.
c. Using set-aside land for bioenergy crops does not result in
ILUC.5 In 2007, it was permitted to grow bioenergy crops on
set-aside land. These lands would have been fallow
otherwise, so there is no production to replace and
therefore no ILUC.
The actual locations of displaced production are determined in step 5.
So, excluding consumption changes, the displaced agricultural production from the net land area will be realised
elsewhere by either yield increase (intensification) or land
expansion. This may occur directly, for example if wheat is
compensated for through a yield increase of wheat or through
land expansion for wheat, but it can also result from
displacement of other crops that can substitute one another.
Since we do not know the actual array of substitutions, we
assume that the reallocation of displaced production occurs in
the same way as increase in global agricultural production has
occurred in the past. The displaced agricultural production is
treated as an additional demand, and the way this demand is
fulfilled has the same behaviour as observed for agricultural
production in the past. We take into account trends of yield
increase and land expansion of all major crop types from the
past years.
We include two variants:
1 All displaced agricultural production from the various
feedstocks in various regions is reallocated over the world.
The fraction production increase that results from intensification and the fraction production increase that results
from expansion are determined by the world average.
2 The displaced agricultural production from a certain region is
reallocated within the region. The fraction of intensification
versus expansion is based on the regional figures.
4
Although the RED was not in place in the period under study
(i.e. 1997–2007), we use this assumption to have a methodology fit
for the discussion around ILUC within the RED.
5
Set-aside was abolished in 2008, so this is a specific situation
that no longer occurs. Potentially, this will influence the calculation of the CO2 balance for emission reduction towards more
negative.
251
Global and regional data on land use (agricultural land
expansion) and yield increase is obtained from the FAOSTAT
database on production (FAO, 2010). For the calculation of the
average land expansion the relative land expansion between
1997 and 2007 of the crop groups is used. The yield increase of
the crops groups is weighted according to the area harvested
to translate yield increase into land (i.e. a crop with a low
yield, in tonnes, and a 10% yield increase is weighted
relatively the same as a crop with a high yield and a 10%
yield increase). This assumption holds if the shares of the
different crops do not change. Another assumption that has
to be made with this is that the cropping densities do not
change. The harvested area is used in the current calculation.
Since area can be harvested more than once a year – cropping
density higher than one – harvested area is not equal to
land area. In the event the shares of the different crops and
their cropping intensities do not change, the calculation
with harvested area is valid. ILUC values are calculated with
the formula below. The result from this formula is used as
the percentage of the net land use area of the biofuel crop
that is converted somewhere else to accommodate the
displaced production that cannot be accommodated by yield
increase.
Indirect land use as percentage of direct land use by the biofuel
¼
Percentage increase in harvested area
percentage increase in yield
þ percentage increase in harvested area
#
Increase in yield is weighted over the harvested area.
In the first variant, we need an additional yield factor. Net
land use is calculated for the world region (for example, the EU
or South America) where the feedstock is cropped. Assuming
the displaced production is cropped all over the world, this
may lead to a somewhat different need for land since the
average yield of agricultural crops is higher or lower in the
region of the feedstock crop then in the world as a whole
where the displaced production is allocated. In the second
variant, the displaced production stays in the same region and
therefore no yield factor was needed.
5. Global land use conversions to agricultural land
In order to transform ILUCs into emissions it should first be
clear which former land cover has changed into the new
agricultural land use, i.e. which land use conversions have
occurred. For most countries, some information on land use for
different years is available on FAOSTAT (resources database).
Potentially, this information could be used to estimate land use
transitions caused by agriculture. However, the data has a low
level of detail on land cover, which is essential to determine the
carbon content of vegetation and soils. To determine the land
use changes that have occurred in the recent past, we use a
global model simulation of the IMAGE model (MNP, 2006), which
has been calibrated to FAOSTAT data from 1970 to 2000. IMAGE
identifies 20 land-cover types. Land use changes towards 2010
are based on scenario calculations including actual biofuel
figures until 2007 (Stehfest et al., in preparation). So, this data set
is not purely monitoring but extrapolated (by modelling) data of
1970–2000 and 2007 biofuels data. The results from IMAGE are
252
environmental science & policy 14 (2011) 248–257
maps of 1995 and 2005 land cover. Similar to the analysis in step
4 we calculated an average over all agriculture land and not
specifically over the effect caused by biofuels (since we do not
have a world without biofuels to compare with). In this
conversion analysis based on IMAGE we analyse the conversion
into arable lands as well as into cultivated grasslands. This
ensures that cases where, for example, forest is converted to
pasture and pasture to arable land, the effect of forest
conversion is also included in the average. In combination
with step 4 we also distinguish a regional and a global variant for
this calculation.
6. CO2 emissions from land use conversions
Land contains carbon stored in vegetation and soil. The
amount of carbon depends on the type of vegetation and the
type of soil. In general, agricultural land contains less carbon
than natural land. We assume that the carbon in the
vegetation is emitted into the air as CO2 at the moment of
the conversion. This is an important assumption, which may
not necessarily be true, as removed biomass can be used for
energy purposes or stored in other ways, for example in timber
products. The carbon content in soil slowly changes to a new
equilibrium after removing the natural vegetation. This
equilibrium can be reached after several decades. While the
emissions vary over time; for many biofuel emission calculations these ILUC emissions are calculated as a yearly average
over periods of 20–50 years. Possibly, the carbon effect can
even be the other way around. For example, if energy crops are
grown on degraded land with almost no vegetation and low
soil carbon content, then a net uptake of CO2 might be the
result.
In this final step, the land conversions from step 5 are
converted to CO2 emissions. The data on the carbon stored in
the vegetation and the soils is based on Sabine et al. (2004);
WBGU (1988) in IPCC (2001); and Carter and Scholes (2000), De
Fries et al. (1999), Roy et al. (2001) in IPCC (2001). This data on
global carbon includes information on pools in various biomes
and the land area of the biomes. This is transformed into
average carbon content per hectare per biome. The minimum
and maximum values from the three sources are taken to
construct a range. The loss due to conversion to arable land is
calculated as the difference between these values and the
minimum carbon content of arable land.
Emissions from land use intensification, which is the main
contributor to increased agricultural production in the world,
are not included in this study. Intensification of agricultural
production can add to the emissions of GHGs in agriculture
through higher emissions from fertilizer use or machinery use
(Smith et al., 1997). We do acknowledge that these emissions
can play a significant role, however, in this study we focus on
the effects from land expansion and only include the effect of
intensification on the total agricultural area.
feedstock and origin of the feedstock. Combining this with
energy yields per hectare (TJ/ha) for the various feedstocks and
the by-products that are produced leads to the net amount of
hectares needed for growing these crops. By making assumptions on how the production of this occupied land is produced
elsewhere, we determined the percentage of ILUC associated
with the feedstock production. Subsequently, we use modelled data on the actual land use conversions to agricultural
land that took place (1995–2005). By using data on CO2
emissions from the different land use conversions we arrive
at CO2 emissions from ILUC caused by biofuel consumption
(tonne CO2/year). By dividing this through the energy consumption we arrive at emissions in gCO2/MJ.
Energy from a certain feedstock and country ðTJÞ
net land use ðha=TJÞ ILUC ð%Þ yield factor ðÞ
conversion emissions ðtonne CO2 =ðha yearÞÞ
¼ emission ðtonne CO2 =yearÞ
(2)
This is summed for all feedstock country combinations
occurring in our datasets.
In order to assess the range of possible outcomes of this
calculation, we combine the highest and lowest figures – in all
possible combinations – for those steps where we included a
range. This gives a good estimate of the range of possible
outcomes, though not of their likelihood. The steps that have a
high and low variant are: imports (step 1), net land use (step 3),
ILUC (steps 4 and 5) and conversion emissions (step 6). This
analysis results in 16 possible outcomes which together span
the potential outcome space. We do provide an average of the
ILUC emissions.
3.
Results
This section starts with the results of the six separate steps.
Subsequently, we describe the result of the final calculation of
the emissions from historical biofuel consumption in the EU.
1. EU biofuel consumption and share of EU-produced and EUimported biofuels during 2007
Final energy consumption of road transport in the EU27
was 49.714 TJ biogasoline and 279.488 TJ biodiesel. Under the
assumption that all imported biofuels are consumed and that
the excess of EU-produced biofuels is exported, the EU
bioethanol consumption consists for 13% of imported
bioethanol; and EU biodiesel consumption consists for 17%
of imported biodiesel. Under the assumption that all EUproduced biofuels are consumed and that the excess imported
biofuels are exported the EU bioethanol consumption includes
2% imported bioethanol and the EU biodiesel consumption
10% imported biodiesel (EUROSTAT, 2009).
Overall analysis
In summary, the set up of the overall calculation is as follows.
We can identify which part of the energy consumption
originates from which crop – measured in terajoules (TJ) –
by using the figures on the total EU biofuel consumption, the
share of EU-produced and EU-imported biofuels, and the
2. Feedstock and origin of the feedstock for the biofuels
Table 1 displays the main exporters of biofuels to the EU. In
the calculation, all smaller exporters are ignored and the
exporters presented are considered to provide 100% of the
import to the EU, for example, where Brazil and Pakistan
provide 80% and 13% of the bioethanol imports in reality,
253
environmental science & policy 14 (2011) 248–257
Table 1 – Sources of biofuel imports to the EU.
Biofuel
Exporter to EU
Feedstock
Share
Bioethanol
Brazil
Pakistan
Sugar cane
Sugar cane
80%
13%
Biodiesel
USA
Argentina
Indonesia
Malaysia
Soy
Soy
Palm
Palm
82%
4%
10%
2%
Authors’ calculations based on F.O. Licht’s, 2009.
Table 2 – Origin of feedstock (percentage).
Feedstock
Bioethanol
Wheat
EU set-aside land
EU non set-aside land
Sugar beet
EU set-aside land
EU non set-aside land
Sugar cane
Brazil
Pakistan
Total
Biodiesel
Rapeseed
EU set-aside land
EU non set-aside land
Soy
United states of America
Argentina
Palm oil
Indonesia
Malaysia
Total
Import
assumption 1
Import
assumption 2
13.0
38.9
14.8
44.3
4.9
29.7
5.6
33.7
11.6
1.9
100
1.4
0.2
100
23.0
60.3
24.7
64.9
13.9
0.8
8.7
0.5
1.6
0.4
100
1.0
0.2
100
respectively, we recalculated this to be 86% and 14%. For
biodiesel, the countries included account for 98% of the
imports; for bioethanol, it is 93%.
A second element of the feedstock analysis is the share of
the former set-aside area that is used for the production of
biofuels. Since this land had no production before biofuels
were cropped, this feedstock production does not induce
indirect land use change. In 2007, the percentage of EU
produced biofuel feedstock on set-aside land was 25% for
wheat, 14% for sugar beet and 28% for rapeseed (EU, 2008).
Combining this information with Table 1 results in the figures
in Table 2. Since 2009 the set-aside policy is abolished. For the
calculation of current biofuel emission this means that all
feedstocks in Europe should be treated similarly. This will
cause the calculated emissions to increase.
3. Energy yields per hectare for the various feedstocks in their
region of origin taking the effect of by-products into account
Table 3 presents the results of the calculations of energy
yields per hectare for the various feedstocks in their region of
origin. Note that inclusion or exclusion of by-products creates
large differences (up to a factor five in case of soy).
In the calculation, we do not use the two extremes of Table
3, but a combination of both. For rapeseed, wheat and sugar
beet, we assume that only 50% (lower assumption) or 90%
(higher assumption) of the by-products is effectively used. The
inefficiency of 50% or respectively 10% is considered to stem
from unused by-products, inefficient use of by-products and a
possible consumption effect that occurs due to higher levels of
feed supply. For soy, we assumed variants of 90% and 100%
efficiency because soy is mainly cultivated for the soy meal
production. For palm oil and sugar cane, we assumed a 0% and
a 100% variant. The by-products for these feedstocks are a
relatively small part of the production; consequently, results
are hardly influenced by these contrasting assumptions.
4. ILUC calculation with historical data on land expansion and
yield increase
The model variant in which all displaced agricultural
production is reallocated over the world produces just one
global figure for ILUC, which is 32%. So, for each hectare of
biofuel on land that is currently agricultural somewhere in the
world, 0.32 hectares are converted to arable land. This
assumes that agricultural production increase is developing
according to the trend of the past ten years.
Table 4 presents the results for the regional approach of the
ILUC calculation. In order to calculate the ILUC in the regional
approach the same formula (1) was used for each region. A few
exceptions were made on this method. First, set-aside area did
not have production prior to the cultivation of biofuels, so no
production is substituted on set-aside area and subsequently
the ILUC is zero. Second, the FAO data for North America
indicated that total agricultural production had increased over
the period with less land. Therefore, we assumed that no extra
land was converted in North America. Here the question
remains as to whether or not production was moved to
Table 3 – Average land demands per unit of energy accounting for the use of by-products.
ha/TJ
ha/TJ
GJ/ha
GJ/ha
Excl. by-products
Incl. by-products
Excl. by-products
Incl. by-products
Bioethanol
Cereals (EU-produced)
Sugar beet (EU-produced)
Sugar cane (import)
14.7
11.3
7.2
4.3
5.9
6.9
68
88
139
232
170
144
Biodiesel
Rapeseed (EU-produced)
Soy (import)
Palm oil (import)
17.7
54.1
5.9
6.8
10.8
5.7
56
18
168
147
92
176
254
environmental science & policy 14 (2011) 248–257
Table 4 – Regional ILUC factors as percentage of direct net
land use.
Table 6 – Emissions from soil and biomass from
conversions to arable land (tonne CO2/ha).
Biofuel feedstock
Original land cover
ILUC
Bioethanol
Wheat
EU set-aside land
EU non set-aside land
Sugar beet
EU set-aside land
EU non set-aside land
Sugar cane
Brazil
Pakistan
0%a
32%
a
0%
32%
49%
19%
Biodiesel
Rapeseed
EU set-aside land
EU non set-aside land
Soy
US
Argentina
Palm oil
Indonesia
Malaysia
0%b
49%
48%
48%
No production is substituted on set-aside area.
b
In the period under analysis agricultural production in the US is
increasing with decreasing agricultural land use.
somewhere else. The regional approach assumes that this is
not the case. Another exception is Europe. Total production in
Europe has decreased, which is evidence that replaced
production could not have been produced in Europe. Therefore, the assumption of regional replacement was not valid for
Europe. Here we used the figure for the world outside Europe,
which is nearly the same as the world average: 32%.
5. Global land use conversions to agricultural land
Table 5 displays the land use conversions in the world in
total, as calculated by the IMAGE model. For the regional
approach, this was further subdivided per world region.
Table 5 – Original land-cover types of land that was
converted to arable and intensive grassland from 1995 to
2005 (as percentage of the new arable land and intensive
grassland).
Regrowth forest (abandoning)
Regrowth forest (timber)
Tundra
Wooded tundra
Boreal forest
Cool conifer forest
Temperate mixed forest
Temperate deciduous forest
Warm mixed forest
Natural grassland/steppe
Hot desert
Scrubland
Savanna
Tropical woodland
Tropical forest
Total
Forests
Tropical
Temperate
Boreal
Grassland
Tropical savanna
Non tropical
Tundra
Deserts
Min.
Max.
591
260
254
1862
1114
1196
133
110
190
77
263
592
656
12
0%a
32%
a
Original land-cover type
Emission (tonne CO2/
ha)
Land area (%)
2.1
13.5
0.1
0.2
4.5
3.5
5.5
5.1
8.8
9.2
0.8
7.7
21.5
9.6
7.8
100
Table 7 – ILUC emissions and carbon breakeven points
(summary of 16 calculations with various assumptions).
Avg.
ILUC emissions (g CO2/MJ)
Bioethanol 20 years
Biodiesel 20 years
Bioethanol 50 years
Biodiesel 50 years
Carbon breakeven point: ‘‘payback
Bioethanol
Biodiesel
79
97
32
39
time’’ (years)
35
50
Min.
Max.
26
30
10
12
154
204
61
82
12
15
70
106
6. CO2 emissions from land use conversions
The minimum and maximum carbon pools according to
the literature used in the various biomes were translated
into CO2 equivalents. The difference of these values
and the carbon content of arable land are reported in
Table 6.
Overall analysis
Overall, the analysis has produced 16 calculations with
possible outcomes of the ILUC emissions of 2007 biofuel use
in the EU. A summary of these calculations is given in Table 7.
The first part of this table shows that the ILUC emissions are
substantial compared to the emissions of a traditional fossil
fuel, which is approximately 84 g CO2 equiv/MJ: for bioethanol
ILUC emissions are 26–154 g CO2/MJ and for biodiesel 30–
204 g CO2/MJ when the conversion emissions, which are
produced mostly during conversion itself, are spread over
20 years. These figures are 2.5 times smaller when spread over
50 years. A calculation that does not need a range of time to
spread the emissions out is the carbon breakeven point, or
payback time. This is the actual time it takes the biofuel
system to actually start reducing carbon emissions. On
average, these terms are 35 years for bioethanol and 50 years
for biodiesel.
The differences in the results (Fig. 2) are caused by the
ranges in assumptions and data that were used in several
steps of the calculation. The largest differences were caused
by the assumptions on conversion emissions (tonne
CO2/ha*year, step 6)) and on net land use (ha/TJ, step 3) of
produced biofuel. The remaining variation is due to
environmental science & policy 14 (2011) 248–257
255
ILUC assumptions – Regional
Conversion emissions high
All imports consumed in EU
Net land use high
Net imports consumed in EU
ILUC assumptions – World
Net land use low
All imports consumed in EU
Net imports consumed in EU
Conversion emissions low
Net land use high
Net land use low
0
40
80
120
160
Emissions (g CO 2 / MJ)
Fig. 2 – Variability in ILUC emissions resulting from the different assumptions.
imports assumptions (step 1) and ILUC assumptions (steps 4
and 5).
as shown above we would arrive at somewhat lower averages,
which are close to the IFPRI study.
4.2.
4.
Discussion and conclusions
4.1.
Uncertainties
As illustrated in the results section, the carbon emission from
land use conversion is the factor that causes most variability
in the results. In modelling studies with similar objectives, the
uncertainty of conversion emissions is often not included (e.g.
Burney et al., 2010; Searchinger et al., 2008). Searchinger et al.
(2008) use an average conversion emission of 351 tonne CO2 equiv/ha for all land conversions in the world. Using this figure
in this analysis would obviously result in less variable results:
ILUC emissions for bioethanol would be 34–56 g CO2/MJ and for
biodiesel 39–76 g CO2/MJ (assuming an annualization period of
20 years). These figures are within the range of the calculations
in this study. On average they are lower than the average of
our calculations.
The second largest source of variability lies in the
assumptions on the use of by-products that are included in
the calculation of the net land use of biofuel cropping.
The figures presented in the results section are estimates
of the authors. This uncertainty can be reduced by
collecting empirical data on the use of by-products of biofuel
feedstocks.
The different assumptions regarding imports and the
regional or global approach towards the location of the ILUC
effects add little to the variability caused by the other two
elements (Fig. 2). The latter is mainly caused by the
assumption that European production was not replaced
within Europe. Therefore, the majority of the biofuel production was treated similarly in both approaches.
Compared to modelling studies, as for example reported by
JRC (JRC-IE, 2010), the values have a similar range. The JRC
study reports feedstock specific values for biodiesel ranging
from 14 to 337 g CO2/MJ and for bioethanol of 19 to 151 g CO2/
MJ. Figures from an IFPRI study (Al-Riffai et al., 2010) fall
completely in the feedstock specific ranges of this study,
except for sugar cane in Brazil where we found higher
emissions. Using the conversion emission used by Searchinger
Methodology
Accuracy and reproducibility are of great importance in order
for the methods to be adopted by policy makers. It is important
to have the most recent information for accurate monitoring.
However, fluctuations between years can be substantial; in
which case a trend based on several years can be more useful
for policy making. The time lag is longer using a longer period
to determine the trend. In the present study, we used a period
of ten years: 1997–2007. This is a relatively short period in
terms of development in agriculture. However, this is quite a
long period in the current developments of biofuel production.
So far, the biofuel production is relatively small compared to
overall production and production increase, which justifies
the time period of ten years.
The approach in this study is based on the assumption that
agricultural production and land use is a regional or global
system. Direct land use for cultivating feedstocks was
calculated in a regional to country specific way, but we
assume that the ILUC will spread regionally or globally. This
may not be the case under all circumstances. Some countries
will claim that they can control the indirect effects of land use
change and that policies should acknowledge that. We think
that for many commodities, the assumption of a regional or
global market will hold.
The way biofuels are accommodated (by intensification or
expansion) is derived from the aggregation over all agricultural crops. This is different from analysing the difference
between a scenario with and a scenario without biofuels. In
that case the effect of biofuels would be judged by taking in use
the extra amount of hectares that may be less productive and
cause biofuels to be more negatively evaluated. In this study
expansion and intensification due to biofuels is treated similar
as for food and feed.
For the interpretation of the results, it should be clearly
noted that this analysis focuses only on indirect effects of land
use change and particularly on land expansion. Emission
effects from land use intensification are not included. Other
indirect effects were assumed to be less relevant for this
approach because this analysis is a historical analysis and not
256
environmental science & policy 14 (2011) 248–257
a predictive approach such as many of the modelling
approaches. Examples of additional effects that some forward
looking modelling approaches incorporate are: changes in
consumption due to increased demand for agricultural
products and a difference in average yields between the case
with and the case without biofuels.
In step 4 the change in consumption (Fig. 1) is assumed to
be zero. This is a rather strong assumption. Increasing prices
due to biofuels might induce a reduction in consumption of
other products. The consumption effect is estimated to be in
the range of 0–50% of the produced biofuels (IIASA, 2009; JRCIE, 2010). However, there are strong arguments to not attribute
consumption effects to a more favourable (i.e. smaller) ILUC
since this would mean that commodities are diverted from
human use. Few policymakers are likely to argue for reduced
food consumption, especially because it will affect those that
are already malnourished, as a desirable way of reducing
greenhouse gases (Searchinger, 2010).
On intensification we assume that the share between
expansion and intensification under a biofuels scenario is the
same as without biofuels. In general, most other studies would
see a slightly higher contribution from expansion to the
production of biofuels, than in the counterfactual case, as the
long-term yield improvements, which are rather linked to
autonomous technological progress than to price effects, are
not changed when introducing biofuels. In this study, the
changes in yields are incorporated in the observed data. These
yield changes are as attributed to the biofuels in the same way
as to the other commodities.
Another effect is higher energy consumption due to
increased supply of energy by the biofuels. This is not
accounted for in this study.
ally, the production of biofuels can have additional negative
social and ecological effects that are not presented in this
study. However, the arguments to promote biofuels are more
extensive than emission reductions only, for example energy
security. It is a policy decisions how to weigh these
arguments.
Policy questions that arise from this kind of analyses are,
for example, if biofuels are the best way to mitigate GHG
emissions from transport and, if not, what possible alternative
ways to mitigate those emissions are. Others perceive the
developments in first generation biofuels as an essential step
towards second-generation biofuels that may have a better
GHG balance. This may also be an argument to continue the
promotion of biofuels even though the current contribution to
GHG reductions is uncertain.
Policies that are strongly related to the policies aiming at
promoting biofuels are the LULUCF (Land Use, Land use
Change And Forestry) accounting rules under the Kyoto
protocol (Searchinger et al., 2009) and REDD (Miles and Kapos,
2008), which both aim to reduce emissions from land use
change. LULUCF accounting rules regulate part of land use
change emissions for ANNEX 1 countries and REDD may
become part of international agreements in the future. It is
evident that accounting for all land use emissions in all
countries accurately could ensure a much more logical
consideration to cultivate biofuel crops, since emissions from
land use changes should be accounted for. However, biofuel
policies as implemented currently are mainly unilateral
arrangements and have the risk that the land use change
leaks towards countries that have fewer obligations to reduce
emissions.
references
4.3.
Monitoring versus modelling in a policy perspective
ILUC and its resulting emissions are an important issue in
discussions on the sustainability of biofuels. Discussions focus
on how to quantify the effects of ILUC and if this should be
incorporated in sustainability criteria; and if so, how. The two
main options for quantification of ILUC effects are monitoring
and modelling. The most important difference is that most
modelling approaches aim to predict future land use under a
baseline and a certain policy scenario and a monitoring
approach is based on historical data. The approach in this
study is an approach in between modelling and monitoring.
The approach is partly based on data from past events, for
example on actual biofuels consumed, land expansion area
and yields, for the remaining part assumptions were made. For
the parts based on historical data this approach has less
uncertainty compared to modelling approaches. We conclude
therefore that in this aspect, the approach has fewer
uncertainties and may be easier to be adopted by policy
makers. Another advantage of this approach is that it can
easily be reproduced, which may add to the acceptability of
this method and its results.
4.4.
Policy perspectives
This analysis shows that based on ILUC effects alone there is
evidence that biofuels do not mitigate emissions. Addition-
Al-Riffai, P., Dimaranan, B., Laborde, D., 2010. Global Trade and
Environmental Impact Study of the EU Biofuels Mandate.
IFPRI.
Bindraban, P.S., Bulte, E.H., Conijn, S.G., 2009. Can large-scale
biofuels production be sustainable by 2020? Agricultural
Systems 101, 197–199.
Birur, D., Hertel, T., Tyner, W., 2008. Impact of Biofuel
Production on World Agricultural Markets: A Computable
General Equilibrium Analysis. GTAP Working Paper
No. 53.
Burney, J.A., Davis, S.J., Lobell, D.B., 2010. Greenhouse gas
mitigation by agricultural intensification. Proceedings of the
National Academy of Sciences.
CARB, 2009. Proposed Regulation to Implement the Low Carbon
Fuel Standard Volume I Staff Report: Initial Statement of
Reasons. California Environmental Protection Agency, Air
Resources Board, Sacramento, California, p. 374.
Carter, A.J., Scholes, R.J., 2000. Spatial Global Database of Soil
Properties. IGBP Global Soil Data Task CD-ROM. International
Geosphere-Biosphere Programme (IGBP) Data Information
Systems, Toulouse.
De Fries, R.S., Field, C.B., Fung, I., Collatz, G.J., Bounoua, L., 1999.
Combining satellite data and biogeochemical models to
estimate global effects of human-induced land cover change
on carbon emissions and primary productivity. Global
Biogeochemical Cycles 13, 803–815.
ECOFYS, 2009. Summary of Approaches to Account for and to
Monitor Indirect Impacts of Biofuel Production. , p. 76.
environmental science & policy 14 (2011) 248–257
Eickhout, B., Born, G.J.v.d., Notenboom, J., Oorschot, M.v., Ros,
J.P.M., Vuuren, D.P.v., Wetshoek, H.J., 2008. Local and
Global Consequences of the EU Renewable Directive
for Biofuels. MNP (Netherlands Environmental Assessment
Agency), Bilthoven, p. 68.
EU, 2006. Biofuels in the European Union. A Vision for 2030 and
Beyond. Final Report of the Biofuels Research Advisory
Council. .
EU, 2008. Bioenergy and the CAP. , http://ec.europa.eu/
agriculture/bioenergy/index_en.htm#cap.
EU, 2009. Directive 2009/28/EC. Official Journal L 140/16–62.
European Biomass Industry Association, 2010. http://
p9719.typo3server.info/214.0.html.
EUROSTAT, 2009. Supply, Transformation, Consumption –
Renewables (Biofuels) – Annual Data. EUROSTAT.
F.O. Licht’s, 2009. World Ethanol and Biofuels Report. .
Fabiosa, J.F., Beghin, J.C., Dong, F., Elobeid, A., Tokgoz, S., Yu, T.H., 2009. Land Allocation Effects of the Global Ethanol Surge:
Predictions from the International FAPRI Model, Working
Paper 09-WP 488. Center for Agricultural and Rural
Development, Iowa State University, Ames, Iowa.
FAO, 2008. The State of Food and Agriculture 2008 Biofuels:
Prospects, Risks and Opportunities. FAO, Rome, p. 128.
FAO, 2010. Production-crops. FAO, In: http://faostat.fao.org/site/
567/default.aspx#ancor.
Fargione, J., Hill, J., Tilman, D., Polasky, S., Hawthorne, P., 2008.
Land clearing and the biofuel carbon debt. Science 319,
1235–1238.
Havlı́k, P., Schneider, U.A., Schmid, E., Böttcher, B., Fritz, S.,
Skalský, S., Aoki, K., Cara, S.D., Kindermann, G., Kraxner, F.,
Leduc, S., McCallum, I., Mosnier, A., Sauer, T., Obersteiner,
M. Global land use implications of first and second
generation biofuel targets. Energy Policy, in press,
doi:10.1016/j.enpol.2010.03.030.
IIASA, 2009. Biofuels and food security implications of an
accelerated biofuels production. In: OFID (Eds.), OFID
Pamphlet Series. The OPEC Fund for International
Development (OFID), Vienna.
IPCC, 2001. In: Houghton, J.T., Ding, Y., Griggs, D.J., Noguer,
M., Linden, P.J.v.d., Dai, X., Maskell, K., Johnson, C.A.
(Eds.), Climate Change 2001: The Scientific Basis.
Contribution of Working Group I to the Third Assessment
Report of the Intergovernmental Panel on Climate
Change. Cambridge University Press, Cambridge/New York,
p. 881.
JRC/CONCAWE/EUCAR, 2007. Well-to-wheel Analysis of Future
Automotive Fuels and Powertrains in the European Context
(+ Annex).
257
JRC-IE, 2010. Indirect Land Use Change from Increased Biofuels
Demand: Comparison of Models and Results for Marginal
Biofuels Production from Different Feedstocks. European
Commission, Joint Research Centre, Institute for Energy,
Ispra, Italy.
Miles, L., Kapos, V., 2008. Reducing greenhouse gas emissions
from deforestation and forest degradation: global land use
implications. Science 320, 1454–1455.
MNP, 2006. Integrated Modelling of Global Environmental
Change. An Overview of IMAGE 2.4. Netherlands
Environmental Assessment Agency, Bilthoven.
Pousa, G.P.A.G., Santos, A.L.F., Suarez, P.A.Z., 2007. History and
policy of biodiesel in Brazil. Energy Policy 35, 5393–5398.
Roy, J., B. Saugier & H. Mooney (2001), Terrestrial Global
Productivity: Past, Present and Future. San Diego: Academic
Press.
Sabine, C.L., Heimann, M., Artaxo, P., Bakker, D.C.E., Chen, C.-T.A.,
Field, C.B., Gruber, N., Le Quere, C., Prinn, R.G., Richey, J.E.,
Romero, P., Sathaye, J.A., Valentini, R., 2004. Current status of
past trends of the global carbon cycle. In: Field, C., Raupach, M.
(Eds.), Global Carbon Cycle, Integrating Humans, Climate and
the Natural World. Island Press, Washington DC, pp. 17–44.
Searchinger, T., Heimlich, R., Houghton, R.A., Dong, F., Elobeid,
A., Fabiosa, J., Tokgoz, S., Hayes, D., Yu, T.H., 2008. Use of
U.S. croplands for biofuels increases greenhouse gases
through emissions from land use change. Science 319,
1238–1240.
Searchinger, T.D., Hamburg, S.P., Melillo, J., Chameides, W.,
Havlik, P., Kammen, D.M., Likens, G.E., Lubowski, R.N.,
Obersteiner, M., Oppenheimer, M., Philip Robertson, G.,
Schlesinger, W.H., David Tilman, G., 2009. Fixing a critical
climate accounting error. Science 326, 527–528.
Searchinger, T.D., 2010. Biofuels and the need for additional
carbon. Environmental Research Letters 5.
Smith, K.A., McTaggart, I.P., Tsuruta, H., 1997. Emissions of N2O
and NO associated with nitrogen fertilization in intensive
agriculture, and the potential for mitigation. Soil Use and
Management 13, 296–304.
Stehfest, E., Woltjer, G., Tabeau, A., Prins, A.G., Banse, M.,
Eickhout, B., Van Meijl, H. Land use and greenhouse gas
effects of European and global biofuel mandates, to be
submitted to Biomass and Bioenergy.
US Congress, 2005. Energy Policy Act of 2005. 119 Stat. 594. .
WBGU, 1988. Die Anrechnung biolischer Quellen und Senken in
Kyoto-Protokoll: Fortschritt oder Rückschlang für den
globalen Umweltschutz Sondergutachten.
Wissenschaftlicher Beirat der Bundesregierung Globale
Umweltveränerungen, Bremerhaven, Germany, p. 76.