Biodiesel / Biodiesel
Open Access
Numéro
OCL
Volume 26, 2019
Biodiesel / Biodiesel
Numéro d'article 39
Nombre de pages 12
Section Economy - Development
DOI https://doi.org/10.1051/ocl/2019034
Publié en ligne 22 octobre 2019

© A. Strapasson et al., Published by EDP Sciences, 2019

Licence Creative Commons
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1 Introduction

This study provides an assessment of land use changes for the major biodiesel crops, soybean, oil palm and oilseed rape OSR), from 1990 to the present. The context of the study is the compromise reached in June 2018 between the European Parliament and the EU Council. The compromise reopens the Indirect Land Use Change (ILUC) issue with a focus on the ILUC risk of feedstocks whose expansion is observed on high carbon stock land. The compromise text from Council and Parliament provides for a freeze on all high ILUC risk feedstocks at the level reached in 2019 from 2021 until 2023 and with a complete cessation by 2030. Crop based biofuels overall will be permitted by Member States at 2020 levels up to a cap of 7%. At the time of writing this article, the Commission was still expected to provide a delegated act, defining the High ILUC-risk feedstocks. They might also provide a methodology for the certification of low ILUC-risk production which could be exempted from the cap. Since this time, the Commission has finally adopted its delegated act on March 13, 2019. This article does not address the question of the certification of low ILUC-risk production.

2 Method

We start from the definition of high indirect land-use change-risk food or feed crop-based biofuels as “bio-liquids and biomass fuels produced from food or feed crops for which a significant expansion of the production area into land with high carbon stock is observed”. In addition, the Renewable Energy Directive (RED) requires that biofuels shall not be made from raw materials obtained from land with high carbon stocks such as forests and wetlands if the land had that status in January 2008.

The report assesses the land use expansion of soybean, oil palm and oil seed rape at regional, country and global levels. The assessment is prepared based on: (1) a literature review of land use change (e.g. dynamics of cropland, pasturelands, forestlands and wetlands), emissions associated with Agriculture, Forest and Other Land Use (AFOLU), oil crop productivity, and the production of vegetable oils; and (2) estimates using historical data (e.g. oil production, land area and crop yields) from FAO, USDA and EUROSAT databases. Countries which presented a net reduction in natural vegetation (e.g. forestlands, wetlands and natural grasslands), where subject to further investigation in order to identify the impacts of their respective oil crops on areas with high carbon stocks. Average values for the expansion in different biomes were taken from the literature and the IPCC (2000) data.

From the published data, we:

  • calculate the land use changes, A0, per year for soybean, oil palm and oilseed rape by country and in sub-national regions where the data permit;

  • calculate what part of this land use change occurs on high carbon stock lands (HCSL), A1. The risk ratio A1/A0 is then the measure we adopt;

  • calculate the average CO2 emissions from vegetation and soil per ha, Em, from the conversion of A1 to cropland. These carbon emission values were estimated based on IPCC (2000);

  • finally, calculate the CO2 emission intensity, Ei, in g/MJ, using equation (1), by country and then globally, where Yoil is the established oil yield on new land per year in t crop/ha and Eoil is the energy content of the oil. We allocate the emission by value (Va) and amortise over 20 years as is standard practice. We arrive at the equation (1):

(1)

Equation (1) calculates the carbon footprint for the production of vegetable oil associated with the expansion of oil crops over HCSL covered by natural vegetation per year attributed to the total land use expansion of that crop per year. Hence, the use of the term risk ratio. Table 1 shows value allocation ratios and the energy contents (Eoil) used in the calculation, taken from the INRAD-CIRAD-AFZ (2019) feed tables. Value allocation ratios (economic) are taken from Edwards et al. (2014), in order to account for the relative value of co-products compared to the oil from each supply chain. Yields (Yoil) the land areas by country (A0), and the high carbon stock land use emissions (Em) are given in Tables 2 and 3.

This work follows on from previous studies by Overmars et al. (2015) and Edwards et al. (2014). Our method is consistent with these studies and LCA practice. While also using historical data to assess the share of expansion in crop area associated with the conversion of HCSL, this report differs from Overmars et al. (2015) in its approach to calculating land use change and emissions. Rather than utilising models of global cropland expansion (IMAGE and CSAM), this report bases the assessment on literature review of expansion at the national (and in some cases subnational) level. Overmars et al. (2015) also assume a counterfactual for crops grown on abandoned land on the grounds that the land would otherwise have reverted to a more natural landscape. We take the position that the land remains as cropland. Edwards et al. (2014) also assess the historical deforestation associated with biofuel crop expansion by allocating deforestation to crop expansion rather than estimating deforestation based on production trends. This allocation may not sufficiently account for the indirect link between crop expansion and deforestation (a delay between logging activities and crop establishment) that is frequently observed (Wicke et al., 2011). This paper attempts to address this complex issue through the literature review, which includes several satellite-based analyses of land use change. The countries assessed here are Argentina, Brazil, Canada, China, Czech Republic, France, Germany, Indonesia, Malaysia, Poland, Romania, the United Kingdom and the United States, as well as the European Union (EU28) as a whole.

Table 1

Values used in calculating the CO2 emissions per MJ (Eq. (1)).

Table 2

Annual CO2 emissions from soy, palm and rapeseed expansion over land with high carbon stocks (HCSL).

Table 3

Annual GHG emissions due to soy and palm expansion over native vegetation (base 2006–2016).

3 Discussion

Table 2 shows our calculated values of the expansion ratios on HCSL and energy intensities at the national levels, and from these we determine the global values. We see a range for palm from 20–30% for the two main producer countries. For soy the range is from 0 to 28%. At the level of regions within countries, these variations are larger still and where possible, these more localised values are used to determine the national averages. The amortised emission intensities for additional production are consistent with recent studies (for example, Edwards et al., 2014; Valin et al., 2015) although our values are lower. We note that our methodology does not support a position that oilseed rape nationally or globally expands directly over high carbon stock lands.

OECD-FAO (2017) projections suggest that yield increases will outstrip the requirement for land use change to meet increased demand. Whilst we acknowledge the potential for future yield increases, our work is based on historical data rather than projections and so we have not considered the potential impacts of future yields in our calculations of the expansion ratios on HCSL. Table 3 summarises of our estimates of the GHG emissions for soy and palm expansion over native vegetation HCSL based on data covering 2006–2016.

Any calculation of this sort contains considerable uncertainties, particularly in the calculation of the expansion ratio, but also the specific prior land use. Where there is a significant range in the HCSL use change, we have used the data covering the most recent period and where available the value at the lower end to reflect the increasing efforts to reduce the conversion of HCSL seen more recently. We have not considered counterfactual land use changes. This might include using otherwise abandoned cropland, where the counterfactual would be a return to a more natural landscape. This would not be consistent with an analysis based on the historical record.

In the following sections we provide further details of the analysis of the three crops. Considerably more detail is contained in the body of the supplementary document on the background to the oil markets, carbon stocks, the land use change assessments and estimates of future production.

3.1 Soybean

We consider the four nations that supply 85% of world production: Argentina, Brazil, China, and the USA. Table 4 summaries the expansion onto high carbon stock lands in these regions.

Table 4

The expansion of high carbon stock lands for soy production.

3.1.1 Argentina

Recent land use changes in Argentina have occurred over forestlands as well as Chaco, with an almost linear decrease in the past decades. Pastureland and cropland have both expanded near linearly, directly or indirectly affecting these land use changes observed in forestlands and possibly in the “other land” category (see Appendix A). Argentina established a new forest code (Ley de Bosque Nativo) in 2007. Since then, the country was expected to reduce its deforestation rates, but the problem continued. From 2007 to 2014, 5 Mha of native vegetation were lost, with 90% of the deforestation occurring in the North Region, including the Gran Chaco area (Larraquy, 2016). The country produces soybean in different states, from North Region (e.g. States of Salta, Chaco and Formosa) to Centre Region (e.g. States of Cordoba, Santa Fe, Buenos Aires and La Pampa). Among other initiatives to reduce deforestation related to the cultivation of soybean, the Round Table on Responsible Soy – TRS (2018) sets standards for its sustainable production but has had limited success to date (Goni, 2018).

The Argentinian environmental legislation aims at reducing the deforestation rates observed in many biomes across the country, but there is a risk of non-compliance (NEPCon, 2017). There is a high deforestation rate including from soy expansion, which has partially expanded over critical wetlands (Herrera et al., 2013). Argentina increased its soy production by 10.9 Mha from 8.6 Mha in 2010 to 19.5 Mha in 2016, according to data from FAO (2018). In the same period, total cropland increased by 11.6 Mha from 28.6 Mha (2010) to 40.2 Mha (2016), whilst total pastureland increased by 8.6 Mha from 99.9 Mha (2010) to 108.5 Mha (2016). Cropland was 57% of the total agricultural land, and pastureland 43%, in 2016. At the same time, 5 Mha of deforestation occurred reducing the forest area from 31.7 Mha (2010) to 26.8 Mha (2016). Based on these data and keeping the area share in the same proportion as in 2016, as well as the share of soybean in total cropland, we to estimate that approximately 28% of the total expansion of soybean may have occurred over forestlands and Chaco. In support of this, FAO data (2018) from 2007 to 2016 shows Argentina’s soybean area expanded approximately 352 kha/year, out of which 99 kha/year occurred over natural vegetation and in line with our 28% estimate.

3.1.2 Brazil

Soy production occupies on 38% of Brazilian cropland. Most of the soybean production in Brazil occurs in consolidated agricultural lands, but areas of significant expansion have affected the Amazon and Cerrado. Most of its recent expansion occurred in the State of Mato Grosso (MT) and this includes areas in the so-called “Amazon deforestation arc”. Amazon deforestation is complex (Strapasson, 2014). Barona et al. (2010) assessed the role of pasture and soybean in Amazon deforestation from 2000 to 2006. They concluded that “The proximate cause of deforestation in the Legal Amazon was predominantly the expansion of pasture, and not of soybeans. However, in Mato Grosso, an increase in soybeans occurred in regions previously used for pasture, which may have displaced pastures further north into the forested areas, causing indirect deforestation”.

The deforestation of Cerrado is a major issue. The expansion of soy in the MATOPIBA region was approximately 62% over forestlands (mainly Cerrado). In all the other states it occurred at about 11% (including Amazon, Cerrado and Atlantic Forest). MATOPIBA accounted for approximately 12% of total soy production in Brazil (2016/2017 crop season), whereas all the other states combined provided 88%. Based on a weighted average of the soybean production in these regions, we estimate that around 17% of the Brazilian soy area has expanded over forestlands in recent years.

From 2007 to 2016, Brazil’s soybean area expanded approximately 1259 kha y−1 (FAO, 2018), out of which 214 kha y−1 occurred over natural vegetation in correspondence with our 17% estimate. Soybean area has approximately the carbon stock for average cropland. The expansion occurred over areas with an average carbon stock similar to tropical forestland. The soy expansion over native vegetation has emitted about 88 MtCO2eq y−1, including changes in soil carbon.

3.1.3 China

Forestlands have been linearly increasing in China in the past decades, mainly driven by planted forests. Pastureland has been almost stable, as have soybean and oilseed rape areas. China is a major importer of soybean. Total cropland has oscillated along a downward trend in the past decades. Soybean production in China has varied substantially in the past decades. Using OECD-FAO (2017) forecast of increasing linear crop yields we may nevertheless see an increase in land for soy. The current dynamics do not correlate with the movement of soy onto HCSL.

3.1.4 USA

Forestlands have increased slightly and croplands decreased over the last few decades. Soybean area has been substantially increasing while total cropland has decreased. The OECD-FAO (2017) project a near linear expansion of soybean production in the US as the potential for yield increases is good. The total required land may not be significantly affected in the coming decade. In terms of emissions, forestlands are a substantial carbon sink. As with China we cannot correlate the change in soy production with expansion on high carbon stock lands.

3.2 Oil palm

Malaysia and Indonesia produce 84% of global palm oil. In this work these two nations stand as representative of the global production.

3.2.1 Indonesia

The majority of Indonesian oil palm plantations by area are located in Sumatra, followed by Kalimantan. Kalimantan has experienced particularly rapid growth in oil palm plantations since the year 2000 (Gunarso et al., 2013; Austin et al., 2017). Estimates of the contribution of oil palm to deforestation are challenging and variable (Wicke et al., 2011). In Indonesia, the primary direct driver of deforestation has been logging (often illegal) rather than the expansion of oil palm plantations. However, oil palm plantations have frequently replaced previously forested land after logging and therefore are associated with the carbon loss from this land use change (Gunarso et al., 2013). Several studies have quantified the proportion of growth in oil palm area linked to deforestation (Tab. 5). Based on FAO data (2018), the expansion of oil palm in Indonesia was approximately 477 kha y−1 between 2007 and 2015.

The expansion of oil palm over forestlands in Indonesia has ranged from approximately 18 to 63% in recent years. We conservatively take a value of 30%. Of this conversion, Austin et al. (2017) report that the majority (94.9%) occurred in secondary forests and 5.1% in primary forests. In the body of the report, we give further details for Sumatra and Kalimantan.

Conversion of peatlands, often through the use of fire (although no longer legal), has contributed to considerable loss of carbon stocks (Gunarso et al., 2013). The estimated area of industrial oil palm plantations on peatlands in Indonesia increased from 19 kha in 1990 to 1311 kha in 2010 (Miettinen et al., 2012). Austin et al. (2017) estimate an expansion of 305 kha on peatlands between 1995–2000 and 619 kha between 2010–2015 (20% proportion for both periods). From Austin et al. (2017) and Gunarso et al. (2013), we estimate that for Sumatra, Kalimantan and Papua, the share of total expansion over native vegetation was approximately 50% and occurred in areas analogous to “tropical forest” and the other 50% in areas analogous to “wetland” (peatland). Carlson et al. (2013) projected that the land use change from the expansion of oil palm from 2010–2020 will conservatively result in emissions for which 56% are from the clearing of forests and 25% from the conversion of peatlands.

Table 5

Expansion over high carbon stock lands in Indonesia.

3.2.2 Malaysia

Forestlands in Malaysia have been relatively stable since 2010. On the other hand, oil palm has expanded significantly onto croplands and into the land use category “other” (see Appendix A). Some of these “other” land use classes may have originated on previously forested areas (Gunarso et al., 2013). A decline in permanent cropland has been observed since 1990 (Wicke et al., 2011). There is also evidence of an increase in clearing of land by 150% in Malaysia in the last decade (Geographical Magazine, 2018). The national harvested plantation area reached an estimated 4.86 Mha in 2015 (FAO, 2018). Estimating this impact is challenging due to the often indirect relationship between deforestation and the establishment of new oil palm plantations (via prior logging). However, oil palm establishment has more directly driven deforestation in Malaysia than in Indonesia (Gunarso et al., 2013). This applies for the regions of Sabah and Sarawak (Gaveau et al., 2016), while deforestation in the Malaysia Peninsular has been more driven by the expansion of rubber and other crops (Gunarso et al., 2013).

Estimates of deforestation associated with palm oil production in Malaysia vary and are given in Table 6. In summary, the expansion of oil palm over forestlands in Malaysia has ranged from approximately 17 to 39% in recent years. We take a conservative estimate of 20% but note that it may be an underestimate due to the uncertainties related the definition of “other land”, which are not considered as forestlands, but which may include peatlands and bare soils. On the other hand, the Malaysian government reports that the emissions associated with land use, land use change and forestry sector have roughly stabilised since 1995 (Ministry of Natural Resources and Environment Malaysia, 2015).

Table 6

Expansion over high carbon stock lands in Malaysia.

3.3 Oilseed rape

The production of rapeseed oil also has a concentrated market, with Canada, China and EU28 dominating the global production with about 78% of the global production. Including India and the United States, this total share increases to 90%. It is third most important vegetable oil crop worldwide, behind oil palm and soybean with global production expanding by 40% between 2006 and 2016. All the assessed countries presented net afforestation/reforestation in the past decade, apart from Canada, which presented a small net deforestation area, based on data from FAO (2018). Thus, as shown in Table 7, there is no apparent correlation between the recent expansion of oilseed rape and forest dynamics.

Table 7

Oilseed rape producing countries and risks associated with high carbon stock lands.

4 Conclusion

According to the analysis of the historical data and using the high ILUC-risk definition as it stands, we consider that the emissions associated with palm and soy oil production are likely to be significant, however we acknowledge that the actual emissions are both highly uncertain and spatially variable depending on the assessed country and sub-region. For oil palm we take Indonesia and Malaysia as proxy for the global position. We calculate an average expansion of 29% over high carbon stock land. For soy we calculate a global average of 19% expansion over high carbon stock land. We calculate the global average greenhouse gas emissions intensities based on the ILUC-risks as 56 gCO2eq/MJ for soy oil and 108 gCO2eq/MJ for palm oil. Future projections (OECD-FAO, 2017) suggest these numbers could drop significantly. We do not find evidence for high ILUC-risk expansion of oilseed rape, given that there is no apparent correlation between the recent expansion of oilseed rape and forest dynamics.

Conflicts of interest

This article is part of a research that was commissioned by Terres Univia (France) on behalf of the organisations of the French vegetable oil and protein sector to LCAworks Limited (United Kingdom). The funding source had no interference in the preparation of this document.

The authors have no competing interests to declare. The opinions and conclusions expressed in this study are those of the authors alone and do not necessarily reflect the views of their respective institutions of affiliation or the opinions of its funding source.

Appendix

Appendix A

Glossary of land use classifications.

References

Cite this article as: Strapasson A, Falcão J, Rossberg T, Buss G, Woods J, Peterson S. 2019. Land Use Change and the European Biofuels Policy: The expansion of oilseed feedstocks on lands with high carbon stocks. OCL 26: 39.

All Tables

Table 1

Values used in calculating the CO2 emissions per MJ (Eq. (1)).

Table 2

Annual CO2 emissions from soy, palm and rapeseed expansion over land with high carbon stocks (HCSL).

Table 3

Annual GHG emissions due to soy and palm expansion over native vegetation (base 2006–2016).

Table 4

The expansion of high carbon stock lands for soy production.

Table 5

Expansion over high carbon stock lands in Indonesia.

Table 6

Expansion over high carbon stock lands in Malaysia.

Table 7

Oilseed rape producing countries and risks associated with high carbon stock lands.

Appendix A

Glossary of land use classifications.

Les statistiques affichées correspondent au cumul d'une part des vues des résumés de l'article et d'autre part des vues et téléchargements de l'article plein-texte (PDF, Full-HTML, ePub... selon les formats disponibles) sur la platefome Vision4Press.

Les statistiques sont disponibles avec un délai de 48 à 96 heures et sont mises à jour quotidiennement en semaine.

Le chargement des statistiques peut être long.