EUROPEAN COMMISSION
Brussels, 13.3.2019
COM(2019) 142 final
ANNEXES
to the
REPORT FROM THE COMMISSION TO THE EUROPEAN PARLIAMENT, THE COUNCIL, THE EUROPEAN ECONOMIC AND SOCIAL COMMITTEE AND THE COMMITTEE OF THE REGIONS
on the status of production expansion of relevant food and feed crops worldwide
ANNEX 1
review of the literature on expansion of crops onto high carbon land
Scope
This review undertaken by the Commission’s Joint Research Centre gives an overview and summarizes the most relevant results of the scientific literature on the expansion of production areas of agricultural commodities into high carbon-stock land, as defined in RED II.
Soybean
There is only one peer-reviewed study that estimates deforestation caused by soybean on a global scale covering a time-frame that includes deforestation after 2008. [Henders et al. 2015] started with GIS-based measurements of year-by-year deforestation in all tropical regions, and allocated it to different drivers, including soy and palm oil expansion, according to a comprehensive review of the regional literature (the review is detailed in their Supplementary Information). However, their data only cover the period 2000-2011.
JRC estimate of percentage deforestation in Brazilian soy expansion
|
|
Amazon
|
Cerrado
|
rest of Brazil
|
% of Brazilian soy expansion 2008-17
|
11%
|
46%
|
44%
|
% of expansion on forest
|
5%
|
14%
|
3%
|
BRAZIL WEIGHTED AVERAGE of expansion on forest
|
8.2%
|
Given the lack of studies providing recent data on a global scale, data were combined from Brazil, other South American countries and the rest of the world. For Brazil, data on soy expansion since 2008 was taken from the Brazilian IBGE-SIDRA database and combined with data on expansion into forest areas in the Cerrado [Gibbs et al. 2015], averaging for the period 2009-13 in the Amazon [Richards et al]
and the rest of Brazil [Agroicone 2018]. It resulted in a weighted average of expansion into forests of 10.4%: This was combined with the numbers from Argentina, Paraguay, Uruguay and Bolivia and the rest of the world, as follows:
JRC estimate of Latin America average percentage of soy expansion onto forest
|
2008-2017
|
Brazil
|
Argentina
|
Paraguay
|
|
Bolivia
|
% of Latin American soy expansion
|
67%
|
19%
|
7%
|
5%
|
2%
|
% onto forest
|
8.2%
|
9%
|
57%
|
1%
|
60%
|
Latin America Average % onto forest
|
14%
|
ESTIMATE OF WORLD AVERAGE % OF SOY EXPANSION ONTO FOREST
|
Fraction of world soy expansion in Latin America
|
53%
|
Assumed % expansion onto forest in rest of the world
|
2%
|
World average fraction of soy expansion onto forest
|
8%
|
For other Latin American countries, the only quantitative data found is [Graesser et al. 2015], who measured the expansion of all arable crops onto forest. For the rest of the world, where the greatest soy expansions since 2008 have been observed, i.e. India, Ukraine, Russia and Canada, little evidence for soy cultivation causing direct deforestation could be found. Therefore, a low share of 2% expansion onto forests was assumed for the rest of the world. As a result, the world average fraction of soy expansion was estimated at 8%.
Comparison with other recent reviews
Most of the data on deforestation by soy pre-dates the Brazilian soy moratorium of 2008, and are therefore not relevant to the present estimate.
A review commissioned by Transport and Environment [Malins 2018] contains a careful review of regional data on soy expansion and deforestation concluding that at least 7% of global soy expansion since 2008 was on forest. However, different years were used for the soy expansion fractions and data and results from [Agricone 2018] and [Richards et al 2017] were not used.
A review commissioned by Sofiproteol [LCAworks 2018] also includes a review of the regional literature on deforestation by soy in the world from 2006-2016. It concludes that 19% of global soy expansion has been onto forest. However, the source of their assumption concering the expansion onto forest in “rest of Brazil” is unclear, and they have sometimes elided “natural land” with forest. Furthermore, when calculating averages, they weight the regional soy data by the total regional production of soy rather than the area of its expansion. Therefore, the number of 19% cannot be considered to be very robust.
Agroicone prepared a document for the Commission which cites unpublished 2018 work by Agrosatelite showing a huge reduction in the fraction of forest in soy expansion in the Cerrado (especially in the Matipoba part) in 2014-17, from 23% in 2007-14 to 8% in 2014-17.
Palm oil
Using sampling of palm oil plantations in satellite data, [Vijay et al. 2016] estimated the fraction of palm oil expansion onto forest from 1989 to 2013, and reported results by country. When setting those national averages in relation to the increases in national harvested area of palm oil in 2008 to 2016, the study derived that, globally, 45% of palm oil expansion was onto land that was forest in 1989.
The supplementary data of [Henders et al. 2015] allocated for the 2008-11 period an average of 0.43 Mha/y of observed deforestation to palm oil expansion. This represents 45% of the estimated increase in world planted area of palm oil in that period
.
In a global study for the European Commission, [Cuypers et al. 2013] attributed measured deforestation to different drivers, such as logging, grazing, and various crops, at a national level. Their results imply that 59% of oil-palm expansion was linked to deforestation between 1990 and 2008.
Comparison of regional studies for Indonesia and Malaysia
Estimated percentage of expansion onto forest
|
|
years
|
Malaysia
|
Indonesia
|
ROW
|
% of world palm expansion 2008-15
|
2008-15
|
15%
|
67%
|
17%
|
|
|
Peninsula Malaysia
|
Malaysian Borneo
|
Indonesian Borneo
|
Rest of Indonesia
|
|
% of national expansion 2008-15
|
2008-15
|
19%
|
81%
|
77%
|
23%
|
|
Gaveau et al. 2016
|
2010-15
|
|
75%
|
42%
|
|
|
Abood et al 2015
|
2000-10
|
|
|
>36%
|
|
SARvision 2011
|
2005-10
|
|
52%
|
|
|
|
Carlson et al. 2013
|
2000-10
|
|
|
70%
|
|
|
Gunarso et al. 2013
|
2005-10
|
>6%
|
|
|
|
|
Gunarso et al. 2013
|
2005-10
|
47%
|
37-75%
|
|
Austin et al. 2017
|
2005-15
|
|
>20%
|
|
Vijay et al. 2016
|
2013
|
40%
|
54%
|
13%
|
Vijay et al. 2016
|
2013
|
45%
|
[Abood et al. 2015] found that 1.6 million hectares of deforestation in Indonesia between 2000 and 2010 took place inside concessions granted to industrial palm oil producers. That is 36% of the total expansion of planted palm oil area in that period, according to Indonesian Government figures.
For the same period, [Carlson et al. 2013] estimated a greater % of deforestation: 1.7 Mha of forest loss in palm oil concessions in Indonesian Borneo; about 70% of the harvested area expansion in that region [Malins 2018]. In a later paper, [Carlson et al. 2018] report 1.84 Mha forest loss in palm oil concessions in Indonesian Borneo and 0.55 Mha in Sumatra, for the period 2000-2015.
[SARvision 2011] found that from 2005 to 2010, 865 thousand hectares of forest were cleared inside the boundaries of known palm oil concessions in Sarawak, the Malaysian province in Borneo where most palm oil expansion takes place. This corresponds to about half of the increase in palm oil harvested area in that time
.
[Gaveau et al. 2016] mapped the overlap of deforestation with expansion of industrial (i.e. not smallholder) palm oil plantations in Borneo, over 5-year intervals from 1990 to 2015. They point out that the great majority of palm oil plantations in Borneo were forest in 1973; lower fractions of deforestation come about when one restricts the delay time between clearance and planting of palm oil. Their results show that for industrial palm oil plantations in Indonesian Borneo, ~42% of the expansion from 2010 to 2015 was onto land that was forest only five years earlier; for Malaysian Borneo the figure was ~75%. The assessment applied a more restricted definition of forest than RED2 considering only forest with >90% canopy cover, and excluding secondary forest (i.e. re-grown forest and shrub after historical clearance or fire).
In a later paper, [Gaveau et al. 2018] show for the period 2008-17, that in Indonesian Borneo, 36% of the expansion of industrial plantations (88% of which were palm oil) was onto old-growth forest cleared the same year, whilst in Malaysian Borneo the average was 69%. In Indonesian Borneo, the rate of deforestation by plantations in different years correlated very strongly with the price of crude palm oil in the previous season, whereas in Malaysian Borneo the correlation was weaker, suggesting longer-term centralized planning of deforestation. The results showed that the rate of palm oil expansion has declined since its peak in 2009-12 while the fraction of it that occurs on forest remained stable.
[Gunarso et al 2013] analysed land cover change linked to oil-palm expansion in Indonesia and Malaysia for the Roundtable on Sustainable Palm Oil (RSPO). The most recent changes they report refer to areas of palm oil that were planted between 2005 and 2010. They show the % of this area that was under various land use categories in 2005. Adding the categories that would unequivocally meet the definition of forest in the Directive, a minimum of 37% was obtained for the expansion onto forest for all Indonesia. However, other land use categories reported includes scrubland (which is principally degraded forest, according to the paper), and this would generally also meet the Directive’s definition of forest. This is a large category in Indonesia, as forest near plantations is often degraded by wildfires years before the plantation expands onto that land. Counting these prior land-use types as forest (as they may have been in year 2000), raises the total % deforestation for Indonesia 2005-10 to about 75%, confirming approximately the findings of [Carlson 2013].
For Malaysia, [Gunarso et al 2013] report that from 2006-10, 34% of palm oil expansion was directly onto forest. However, they also reported considerable expansion onto “bare soil” in 2006, and supposed that some of that was bare because it was being converted from forest. From their supplementary information, it can be seen that over a third of the bare soil in 2006 was forest six years earlier, indicating that it is likely to have been areas of forest cleared ready for planting. Including these forest areas would push the fraction of deforestation-linked palm oil expansion up to 47% in Malaysia.
Instead of using satellite images to identify the previous land-cover where Indonesian palm oil plantations expanded, [Austin et al. 2017] referred to land-use maps issued by the Indonesian Ministry of Environment and Forestry. They found that only about 20% of the land used for the expansion of industrial palm oil in 2005-15 had been classified as “forest” on those maps five years before. Their definition of forest specifies >30% canopy cover (instead of >10% in the Directive), and does not include scrub, which would sometimes qualify as forest under the definition of the Directive. A further 40% of the palm oil expansion occurred on land-use categories that included scrub. For these reasons, it is considered that [Austin et al 2017]’s figure of 20% expansion onto forest in 2010-2015 is likely to be an underestimate for the purpose of this report.
JRC estimate of percentage of palm oil expansion onto forest for rest-of-the-world
|
|
year of expansion
|
Latin America
|
Africa
|
rest Asia
|
% of world palm oil expansion 2008-15
|
2008-15
|
9%
|
3%
|
5%
|
Furumo and Aide 2017
|
2001-15
|
20%
|
|
|
Maaijard et al. 2018
|
|
|
6%
|
|
Vijay et al. 2016
|
2013
|
21%
|
6%
|
4%
|
weighted average for ROW
|
2013
|
13%
|
As shown in the table, lower shares of expansion into forest are reported for the rest of the world. Weighting the results for Latin America, Africa and the rest of Asia (excluding Indonesia and Malysia) an average share of expansion of palm oil plantations into forest of 13% was derived.
Overall, taking into account the results from the regional studies on palm oil expansion into high-carbon stock land in Malysia and Indonesia and evidence for such expansion in the rest of the world the world-average share of palm oil expansion onto forest of 45% proposed by [Vijay et al 2016] can be considered a good estimate.
Fraction of oil-palm expansion onto peat
[Abood et al. 2014] found that 21% of known Indonesian palm oil concessions were located over peatlands, and 10% over deep peat (>3 metres), which is supposed to be protected from drainage under a 1990 Indonesian government decree. From 2000-2010, they reported 535 kha of peat-swamp forest were lost on Indonesian palm oil concessions, which is 33% of the palm oil expansion on concessions.
[Miettinen et al. 2012, 2016] analysed high-resolution satellite images to track the spread of mature palm oil plantations onto peatland at times between 1990 and 2015. They used the JRC’s European Digital Archive of Soil Maps to identify peat areas and report that between 2007 and 2015 palm oil plantations expanded 1089 kha onto Indonesian peatland and 436 kha onto Malaysian peatland. Dividing by the increase in mature palm oil area in that time period
, gives a share of 24% palm oil expansion onto peat in Indonesia, and 42% in Malaysia. For the latest period they report, 2010-2015, the corresponding figures are 25% and 36%.
The Malaysian Palm Oil Board published a study of palm oil [Omar et al. 2010], based on GIS identification of palm oil cultivation, and a soil map from the Malaysian Ministry of Agriculture. They report that the percentage of palm cultivation on peat in Malaysia grew from 8.2% in 2003 to 13.3% in 2009, corresponding to 313 and 666 kha respectively. In the same period, their data show the total area of palm oil expanded from 3813 to 5011 kha, so the fraction of that expansion that was on peat was 30%.
[SARvision 2011] found that from 2005 to 2010, 535 thousand hectares of peat-forest were cleared inside the boundaries of known palm oil concessions in Sarawak, the Malaysian province where most palm oil expansion takes place. This corresponds to about 32% of the increase in palm oil harvested area in that time
. This misses peat-forest loss for palm oil outside concession boundaries, and any conversion of peatland that was not forested at the time of conversion.
[Gunarso et al. 2013] report an anomalously low fraction of palm oil expansion on peat in Malaysia (only 6% between 2000 and 2010, according to their supplementary information). This is far below any other estimate, even from the Malaysian sources, so it was discounted
.
For Indonesia, [Gunarso et al. 2013]’s supplementary data show 24% of palm oil expansion between 2005 and 2010 was onto peat-swamp, and this only rises to ~26% if the conversion from peat swamp via “bare soil”are included.
[Austin et al. 2017] report that the fraction of Indonesian palm oil expansion onto peat remained at ~20% for all the time-periods they studied (1995-2015), without any correction for “bare soil”. The reason why Austin’s results are lower than others is the use of the “BBSDLP
” peat map from the Indonesian Ministry of Agriculture (H. Valin, private communication, 5 December 2018). The BBSDLP map does not include areas with less than 0.5 m depth of peat
, and this is partly why it shows 13.5% less peat area than maps from Wetlands International, which themselves probably underestimate peat area by about 10-13%, according to ground surveys. [Hooijer and Vernimmen 2013].
Quantitative data for the fraction of palm expansion onto peatland in the rest of the world is not available. From 2008-15, 9% of palm oil expansion was in Latin America, 5% in the rest of Asia and 3% in Africa. There are considerable areas of tropical peat in South America, especially in Peru, Bolivia, Venezuela and along the Amazon, but these are not significant production areas of palm oil. However, the world’s largest tropical peat swamp is in the Congo basin. There, already at least one huge palm oil concession, of 470 kha (e.g. 10% of the entire area of palm oil in Malaysia), has been granted, and it lies 89% on peat [Dargie et al. 2018]. The fear is that as production growth in SE Asian countries slows, more investment will flow into developing palm oil on peatlands in Africa and Latin America.
Placing the most weight on the results of [Miettinen et al. 2012, 2016], which can be considered as the most advanced pieces of scientific literature, and assuming zero peatland drainage for palm in the rest of the world, gives an interpolated weighted average estimate of 23% expansion of palm oil onto peat for the whole world between 2008 and 2011.
Sugar cane
More than 80% of global sugar cane expansion took place in Brazil from 2008 to 2015.
[Cuypers et al. 2013] estimated that 36% of world sugar cane expansion between 1990 and 2008 was onto land that was previously forest. However, that is likely an over-estimate for the purposes of the analysis: deforestation was allocated between forestry, expansion of pasture, and expansion of different crops, at national scale. Little deforestation was attributed to pasture land, because it hardly showed any net expansion; by contrast, sugar cane expanded greatly and therefore received a high allocation of the national deforestation. However, the regions of Brazil where sugar cane expanded mostly do not overlap with areas of high deforestation, and this was not considered in the analysis of [Cuypers et al. 2013].
[Adami et al. 2012] reported that only 0.6% of sugar cane expansion in the Centre-South of Brazil went onto forest between 2000 and 2009. Although the region accounted for about 90% of world sugar cane expansion in that time period, there was some expansion in other regions of Brazil not covered by this study.
[Sparovek et al. 2008] agreed that in 1996-2006 sugar cane expansion in the Brazilian Centre-South was almost entirely onto pasture or other crops (as there is very little forest left in that region); however, another 27% of expansion occurred in “peripheral” areas around and inside the Amazon biome, in the Northeast and in the Atlantic Forest biome. In these peripheral regions, there was a correlation between forest loss per municipality and sugar cane expansion. However, no figures on the share of expansion onto forest are given in the paper.
As a result no adequate quantification of deforestation by sugar cane could be derived from the literature.
Maize
Cereals are not usually thought of as causing deforestation, because most production is in temperate zones where deforestation is generally modest. However, maize is also a tropical crop, often grown by smallholders, and also often rotated with soybeans on large farms. And a disproportionate part of maize expansion happens in tropical regions where deforestation is more common and carbon-intensive.
The expansion in China was concentrated onto marginal land in the North-East of the country [Hansen 2017], which one supposes to be mostly steppe grasslands rather than forest. The expansion in Brazil and Argentina could be assigned the same % deforestation as soy in Brazil.[Lark et al. 2015] found that, of US maize expansion between 2008 and 2012, 3% was at the expense of forest, 8% shrubland and 2% wetlands. Nevertherless, it is difficult to make a global estimate without looking into detail at what is happening in each country.
References
[Abood et al. 2015] Abood, S. A., Lee, J. S. H., Burivalova, Z., Garcia-Ulloa, J., & Koh, L. P. (2015). Relative Contributions of the Logging, Fiber, Palm oil, and Mining Industries to Forest Loss in Indonesia. Conservation Letters, 8(1), 58–67. http://doi.org/10.1111/conl.12103
[Adami et al. 2012] Adami, M., Rudorff, B. F. T., Freitas, R. M., Aguiar, D. A., Sugawara, L. M., & Mello, M. P. (2012). Remote Sensing Time Series to Evaluate Direct Land Use Change of Recent Expanded Sugarcane Crop in Brazil. Sustainability, 4, 574–585.
http://doi.org/10.3390/su4040574
[Agroicone 2018] Moriera, A, Arantes,S., and Romeiro, M. (2018). RED II information paper: assessment of iLUC risk for sugarcane and soybean biofuels feedstock. Agroicone, Sao Paulo 2018.
[Austin et al. 2017] Austin, K. G., Mosnier, A., Pirker, J., McCallum, I., Fritz, S., & Kasibhatla, P. S. (2017). Shifting patterns of palm oil driven deforestation in Indonesia and implications for zero-deforestation commitments. Land Use Policy, 69(August), 41–48. http://doi.org/10.1016/j.landusepol.2017.08.036
[Carlson et al. 2013] Carlson, K. M., Curran, L. M., Asner, G. P., Pittman, A. M., Trigg, S. N., & Marion Adeney, J. (2013). Carbon emissions from forest conversion by Kalimantan palm oil plantations. Nature Clim. Change, Retrieved from https://www.nature.com/nclimate/journal/v3/n3/pdf/nclimate1702.pdf
[Curtis et al. 2018] Curtis, P. G., Slay, C. M., Harris, N. L., Tyukavina, A., & Hansen, M. C. (2018). Classifying drivers of global forest loss. Science, 361(6407), 1108–1111.
http://doi.org/10.1126/science.aau3445
[Cuypers et al. 2013] Cuypers, D., Geerken, T., Gorissen, L., Peters, G., Karstensen, J., Prieler, S., van Velthuizen, H. (2013). The impact of EU consumption on deforestation : Comprehensive analysis of the impact of EU consumption on deforestation. European Commission.
http://doi.org/10.2779/822269
[Dargie et al. 2018] Dargie, G.C., Lawson, I.T., Rayden, T.J. et al. Mitig Adapt Strateg Glob Change (2018).
https://doi.org/10.1007/s11027-017-9774-8
[FAOstat 2008], Food and Agriculture Organization of the United Nations, Searchable database of crop production statistics,
http://www.fao.org/faostat/en/#data/QC
[Fehlenberg et al. 2017] Fehlenberg, V., Baumann, M., Gasparri, N. I., Piquer-Rodriguez, M., Gavier-Pizarro, G., & Kuemmerle, T. (2017). The role of soybean production as an underlying driver of deforestation in the South American Chaco. Global Environmental Change, 45(April), 24–34. http://doi.org/10.1016/j.gloenvcha.2017.05.001
[Furumo & Aide 2017] Furumo, P. R., & Aide, T. M. (2017). Characterizing commercial palm oil expansion in Latin America: land use change and trade. Environmental Research Letters, 12(2), 024008.
http://doi.org/10.1088/1748-9326/aa5892
[Gaveau 2016] Gaveau, D.L.A., Sheil, D., Husnayaen, Salim, M.A., Arjasakusuma, S., Ancrenaz, M., Pacheco, P., Meijaard, E., 2016. Rapid conversions and avoided deforestation: examining four decades of industrial plantation expansion in Borneo. Nature - Scientific Reports 6, 32017.
[Gaveau 2018] Gaveau, D.L.A., Locatelli, B., Salim, M.A., Yaen, H., Pacheco, P. and Sheil, D. Rise and fall of forest loss and industrial plantations in Borneo (2000–2017). Conservation Letters. 2018;e12622. https://doi.org/10.1111/conl.12622
[Gibbs et al. 2015] Gibbs, H. K., Rausch, L., Munger, J., Schelly, I., Morton, D. C., Noojipady, P., Walker, N. F. (2015). Brazil’s Soy Moratorium: Supply-chain governance is needed to avoid deforestation. Science, 347(6220), 377–378. http://doi.org/10.1126/science.aaa0181.
[Graesser et al. 2015] Graesser, J., Aide, T. M., Grau, H. R., & Ramankutty, N. (2015). Cropland/pastureland dynamics and the slowdown of deforestation in Latin America. Environmental Research Letters, 10(3), 034017. http://doi.org/10.1088/1748-9326/10/3/034017
[Gunarso et al. 2013] Gunarso, P., Hartoyo, M. E., Agus, F., & Killeen, T. J. (2013). Palm oil and Land Use Change in Indonesia, Malaysia and Papua New Guinea. RSPO.
http://doi.org/papers2://publication/uuid/76FA59A7-334A-499C-B12D-3E24B6929AAE
Supplementary materials:
https://rspo.org/key-documents/supplementary-materials
[Hansen et al. 2017] Hansen, J., M.A. Marchant, F. Tuan, and A. Somwaru. 2017. "U.S. Agricultural Exports to China Increased Rapidly Making China the Number One Market." Choices. Q2.
http://www.choicesmagazine.org/choices-magazine/theme-articles/us-commodity-markets-respond-to-changes-in-chinas-ag-policies/us-agricultural-exports-to-china-increased-rapidly-making-china-the-number-one-market
[Henders et al 2015] Henders, S., Persson, U. M., & Kastner, T. Trading forests: Land-use change and carbon emissions embodied in production and exports of forest-risk commodities. Environmental Research Letters, 10(12), 125012. http://doi.org/10.1088/1748-9326/10/12/125012
http://doi.org/10.1088/1748-9326/10/12/125012
[Hooijer and Vernimmen 2013] Hooijer, A. and Vernimmen, R. 2013 “Peatland maps: accuracy assessment and recommendations” Report by Deltares & Euroconsult Mott MacDonald for Implementation of Agentschap NL 6201068 QANS Lowland Development edepot.wur.nl/251354
[Jusys 2017] Jusys, T. (2017) A confirmation of the indirect impact of sugarcane on deforestation in the Amazon, Journal of Land Use Science, 12:2-3, 125-137, DOI: 10.1080/1747423X.2017.1291766
[Lark et al. 2015] Lark, T.J, Salmon, M.J, & Gibbs, H. (2015). Cropland expansion outpaces agricultural and biofuel policies in the United States. Environmental Research Letters. 10. 10.1088/1748-9326/10/4/044003.
[LCAworks 2018] Strapasson,A., Falcao, J., Rossberg, T., Buss, G., and Woods, J. Land use Change and the European Biofuels Policy: the expansion of oilseed feedstocks on lands with high carbon stocks. Technical report prepared by LCAworks Ltd., in collaboration with Sofiproteol, France.
[Machedo et al. 2012] Macedo, M. N., DeFries, R. S., Morton, D. C., Stickler, C. M., Galford, G. L., & Shimabukuro, Y. E. (2012). Decoupling of deforestation and soy production in the southern Amazon during the late 2000s. Proceedings of the National Academy of Sciences of the United States of America, 109(4), 1341–6. http://doi.org/10.1073/pnas.1111374109
[Malins. 2017] Malins, C. (2017). For peat’s sake - Understanding the climate implications of palm oil biodiesel. Cerulogy and Rainforest Foundation Norway, London 2017. Retrieved from
http://www.cerulogy.com/uncategorized/for-peats-sake/
[Malins 2018] Malins, C. (2018). Driving deforestation: the impact of expanding palm oil demand through biofuel policy, London 2018. Retrieved from http://www.cerulogy.com/palm oil/driving-deforestation/
[Meijaard et al. 2018] Meijaard, E., Garcia-Ulloa, J., Sheil, D., Wich, S.A., Carlson, K.M., Juffe-Bignoli, D., and Brooks, T. . (2018). Palm oil and biodiversity.
http://doi.org/https://doi.org/10.2305/IUCN.CH.2018.11.en
[Miettinen et al. 2012] Miettinen, J., Hooijer, A., Tollenaar, D., Page, S. E., & Malins, C. (2012). Historical Analysis and Projection of Palm oil Plantation Expansion on Peatland in Southeast Asia. Washington, D.C.: International Council on Clean Transportation.
[Miettinen et al. 2016] Miettinen, J., Shi, C., & Liew, S. C. (2016). Land cover distribution in the peatlands of Peninsular Malaysia, Sumatra and Borneo in 2015 with changes since 1990. Global Ecology and Conservation, 6, 67–78.
http://doi.org/10.1016/j.gecco.2016.02.004
[Morton et al. 2006] Morton, D. C., DeFries, R. S., Shimabukuro, Y. E., Anderson, L. O., Arai, E., del Bon Espirito-Santo, F., … Morisette, J. (2006). Cropland expansion changes deforestation dynamics in the southern Brazilian Amazon. Proceedings of the National Academy of Sciences of the United States of America, 103(39), 14637–14641.
http://doi.org/10.1073/pnas.0606377103
[Omar et al. 2010] Omar, W., Aziz,N.A.,Mohammed A.T., Harun, M.H. and Din, A.K.; “Mapping of oil palm cultivation on peatland in Malaysia, Malaysian Palm Oil Board Information series 529, MPOB TT No. 473, June 2010. ISSN 1511-7871.
[Page et al. 2011] Page, S.E., Morrison, R., Malins, C., Hooijer, A., Rieley, J.O. Jaujiainen, J. (2011). Review of Peat Surface Greenhouse Gas Emissions from Palm oil Plantations in Southeast Asia. Indirect Effects of Biofuel Production, (15), 1–77.
[Richards et al. 2017] Richards, P. D., Arima, E., VanWey, L., Cohn, A., & Bhattarai, N. (2017). Are Brazil’s Deforesters Avoiding Detection? Conservation Letters, 10(4), 469–475.
http://doi.org/10.1111/conl.12310
[SARVision 2011] SARVision. (2011). Impact of palm oil plantations on peatland conversion in Sarawak 2005-2010, (January 2011), 1–14.
http://archive.wetlands.org/Portals/0/publications/Report/Sarvision%20Sarawak%20Report%20Final%20for%20Web.pdf
[Searle & Giuntoli 2018] Searle, A. S., and Giuntoli, J. (2018). Analysis of high and low indirect land-use change definitions in European Union renewable fuel policy.
[Sparovek et al. 2008] Sparovek, G.; A. Barretto; G. Berndes; S. Martins; and Maule, R. (2008). “Environmental, land-use and economic implications of Brazilian sugarcane expansion 1996–2006.” Mitigation and Adaption Strategies for Global Change,14(3), p. 285.
[USDA 2008] United States Department of Agriculture Foreign Agricultural Service. Searchable database of Production, Supply and Distribution data of crops.
https://apps.fas.usda.gov/psdonline/app/index.html#/app/advQuery
[Vijay et al. 2016] Vijay, V., Pimm, S. L., Jenkins, C. N., Smith, S. J., Walker, W., Soto, C., … Rodrigues, H. (2016). The Impacts of Palm oil on Recent Deforestation and Biodiversity Loss. PLOS ONE, 11(7), e0159668.
http://doi.org/10.1371/journal.pone.0159668
[Waroux et al. 2016] Waroux, Y., Garrett, R. D., Heilmayr, R., & Lambin, E. F. (2016). Land-use policies and corporate investments in agriculture in the Gran Chaco and Chiquitano. Proceedings of the National Academy of Sciences, 113(15), 4021–4026.
http://doi.org/10.1073/pnas.1602646113
[Yousefi et al. 2018].Yousefi, A., Bellantonoio, M, and Hurowitz,G., The avoidable Crisis, Mighty Earth, Regnskogfondet and FERN, March 2018,
http://www.mightyearth.org/avoidablecrisis/
ANNEX 2
GIS analysis
1.
Method
In order to estimate deforestation and related emissions associated with the expansion of biofuel crops since 2008 into areas with a tree canopy cover density greater than 10 %, a geospatial modelling approach was used to combine a deforestation map from Global Forest Watch (GFW) with crop type maps from MapSPAM and EarthStat. Further details of the approach are summarized below, and data sources used in the analysis are listed in the Table below. The analysis was undertaken using a pixel size of approximately 100 hectares at the equator.
Data Sources
Crop Data
At present, globally consistent maps showing the expansion of all individual biofuel crops through time are not available, although research is ongoing to achieve this for palm oil and soybean through the interpretation of satellite imagery. For this analysis, we relied on two sources for single-year, single-crop maps: MapSPAM (IFPRI and IIASA 2016), which captures the global distribution of 42 crops in the year 2005
, and EarthStat (Ramankutty et al. 2008), which maps crop and pasture areas in the year 2000. Both sources of crop data result from approaches that combine a variety of spatially-explicit input data to make plausible estimates of global crop distribution. Data inputs include production statistics at the scale of administrative (subnational) units, various land cover maps produced from satellite imagery, and crop suitability maps created based on local landscape, climate and soil conditions.
Given the lack of up-to-date global maps for individual crops as well as the lack of consistent information about their expansion through time, a major assumption used in our analysis is that total deforestation and associated GHG emissions occurring within an area since 2008 can be allocated to a specific crop based on each crop’s proportional area relative to the total agricultural land area, including pasture, present in the same pixel of the crop map.
Deforestation Data
Published maps of global annual tree cover loss derived from Landsat satellite observations, available on Global Forest Watch for years 2001 through 2017, formed the basis of our deforestation analysis. The tree cover loss data are available at a 30-meter resolution, or a pixel size of 0.09 hectares. The original tree cover loss data of Hansen et al. (2013) do not distinguish permanent conversion (i.e., deforestation) from temporary loss of tree cover due to forestry or wildfire. Therefore, for this analysis we included only the subset of tree cover loss pixels that fell within areas dominated by commodity-driven deforestation, as mapped at a 10-kilometer resolution by Curtis et al. (2018)
. Thus areas where other drivers, such as forestry or shifting agriculture, are dominant were excluded from analysis. Within the commodity-driven deforestation class, only pixels with a percent tree cover above 10 percent were considered for analysis, with “percent tree cover” defined as the density of tree canopy coverage of the land surface in the year 2000. Given the specific criteria included in RED2 (see “b” and “c” in Background above), analysis results were disaggregated into deforestation for the years 2008 through 2015 for areas with greater than 30 percent tree cover and areas with 10-30 percent tree cover.
Curtis et al. (2018) point out that multiple forest loss drivers may be present within a landscape at any given time, and the dominant driver may vary in different years during the 15-year study period; their model assigned only one dominant driver that contributed to the majority of tree cover loss within that landscape during the study period. One assumption used in this analysis was that all tree cover loss within areas dominated by commodity-driven deforestation was for the expansion of new agricultural areas. This assumption would tend to over-estimate the effect of commodity-crops in those pixels. On the other hand, agriculture may also be expanding in areas dominated by shifting agriculture or forestry; other classes from the Curtis et al. (2018) map that were excluded from our analysis. This implies that the method could under-estimate the deforestation due to crops. However, the footprint areas of the nine crops included in this analysis fell primarily into the commodity-driven deforestation class, and therefore crop areas outside this class were assumed to have small area ratios (see Crop Allocation Model section below) and therefore the contribution of these areas to the final totals should be small.
Peatland Data
Peatland extent was defined using the same maps as Miettinen et al. 2016, who mapped changes in land cover from 1990 to 2015 in the peatlands of Peninsular Malaysia, Sumatra and Borneo. For Sumatra and Kalimantan, Miettinen et al. (2016) included peat from the Wetlands International 1:700,000 peatland atlases (Wahyunto et al. 2003, Wahyunto et al. 2004), where peat was defined as follows: “soil formed from the accumulation over a long period of time of organic matter such as the remains of plants”. Peat soil is generally waterlogged or flooded all year long unless drained.” As outlined in Wahyunto and Suryadiputra (2008), the peatland atlases in turn compiled data from a variety of sources which primarily used imagery (satellite, radar, and aerial photography data), as well as survey and soil mapping, to map peat distribution. For Malaysia, peat from the European Digital Archive of Soil Maps was used (Selvaradjou et al. 2005).
An analysis specific to deforestation from palm oil expansion in peat soils was undertaken due to the importance of peat in this biofuel crop’s overall land use and GHG footprint. Using industrial palm oil expansion data from Miettinen et al. 2016, the area of tree cover loss that occurred before the year of known palm oil expansion from 2008 through 2015 was estimated.
GHG Emissions Data
Emissions from deforestation since the year 2008 were estimated as the loss of carbon from the aboveground biomass pool. Emissions are expressed in units of megatons of carbon dioxide (Mt CO2).
Emissions from aboveground biomass loss were calculated by overlaying the map of tree cover loss (from 2008 through 2015) with a map of aboveground live woody biomass in the year 2000. The biomass map, produced by Woods Hole Research Center and derived from satellite and ground observations, is available on Global Forest Watch. All biomass loss was assumed to be “committed” emissions to the atmosphere upon clearing, although there are lag times associated with some causes of tree loss. Emissions are “gross” estimates rather than “net” estimates, meaning that the land use after clearing, and its associated carbon value, was not considered. The carbon fraction of aboveground biomass was assumed to be 0.5 (IPCC 2003) and carbon was converted to carbon dioxide using a conversion factor of 44/12, or 3.67. One advantage of using a pixel-based forest biomass map with continuous values, rather than assigning categorical carbon stock values to different land cover types (e.g., forest, shrubland, IPCC Tier 1 values, etc.) is that the data used for estimating biomass loss is completely independent of the choice of land cover map used to estimate land cover change.
Emissions associated with other carbon pools, such as belowground biomass (roots), dead wood, litter and soil carbon, including peat decomposition or fires, were excluded from the analysis.
Analysis Extent
The extent of the global analysis was defined by overlaying the commodity-driven deforestation map (Curtis et al. 2018) with the biofuel-relevant crops of interest (palm oil, coconut, wheat, rapeseed, maize, soybean, sugar beet, sunflower and sugar cane). Only pixels that were included in one of the nine crops of interest and that touched the commodity-driven deforestation class were considered in the analysis.
Crop Allocation Model
Total deforestation and emissions within a given 1-kilometer pixel were allocated to different biofuel crops of interest based on the proportion of each crop present in the pixel (“Crop X”, e.g. soy) relative to the total area of agricultural land in the pixel, defined here as the sum of cropland and pasture land. In this way, each biofuel crop’s relative contribution to the pixel’s total agricultural footprint served as the basis for allocating its associated deforestation and GHG emission footprint.
Because a single, globally consistent and up-to-date map of agricultural land disaggregated by crop type was not readily available, we applied a two-step process to approximate each biofuel crop of interest’s relative role in deforestation and emissions in a given location (Eq. 1). In the first step, we used crop data for the most recent year available (MapSPAM, Year 2005) to calculate the ratio of Crop X to total cropland within a pixel. In the second step, we used EarthStat data (Year 2000) to calculate the ratio of total cropland to total pasture+cropland within a pixel. (EarthStat data were used because MapSPAM does not include maps of pasture land, and the expansion of pasture land also plays a role in deforestation dynamics.) Combining these two steps made it possible to approximate the relative contribution of Crop X to the total agricultural footprint within a given pixel, albeit using different data sources from different time periods.
Equation 1:
Final Calculations
Once the crop allocation maps were created for each biofuel crop of interest, we multiplied the total deforestation and GHG emissions by the proportion of Crop X in each 1- kilometer pixel, and calculated global summary statistics disaggregated by deforestation and emissions occurring on land with greater than 30 percent tree canopy density and on land with 10-30 percent tree canopy density.
The GIS results show the deforestation observed during the 8 calendar years 2008 to 2015, which associated with different crops. To see what % of the crop expansion is associated with deforestation, the total area of deforestation during these years was divided by the corresponding increase in crop area. To take into account that a crop can still cause deforestation even when the overall global crop area declines but expands in some countries, the shares were calculated based on the gross increase in global crop area, which is the sum of the increases in crop area in countries where it did not shrink.
Further, data on harvested areas was adjusted to obtain information on planted areas: for annual crops, the increase in crop area was assumed to be the same as the increase in harvested area. For (semi-)permanent crops, the fraction of the crop area, which is not harvested because the plants have not yet reached maturity, was taken into account. Sugar cane needs to be replanted about every five years, but there are only four harvests, as it is still immature after the first year. Oil-palm is replanted about every 25 years and bears fruit in the last 22 years.
For most crops, the database [FAOstat 2008] was used, which shows the area harvested by calendar year. Only for palm oil, data from [USDA 2008] was chosen, because it reports data on all mature palm oil areas, including in years where harvesting was impeded by flooding. The database also includes more countries for this crop.
Table: Summary of Data Sources in WRI GIS-analysis.
Dataset
|
Source
|
Forest and Peat Extent
|
Tree Cover 2000
|
Hansen et al. 2013
|
Peatlands
|
Miettinen et al. 2016
|
Deforestation
|
Tree Cover Loss
|
Hansen et al. 2013 (+ annual updates on GFW)
|
Commodity-driven deforestation
|
Curtis et al. 2018
|
Palm oil Expansion, 2000-2015 (for estimation of deforestation on peat)
|
Indonesia, Malaysia
|
Miettinen et al. 2016
|
GHG Emissions
|
Aboveground Biomass
|
Zarin et al. 2016
|
Crop and Pasture Extent Data
|
MapSPAM (physical area)
|
IFPRI and IIASA 2016
|
EarthStat
|
Ramankutty et al. 2008
|
References
Curtis, C., C. Slay, N. Harris, A. Tyukavina, M. Hansen. 2018. “Classifying Drivers of Global Forest Loss.” Science 361: 1108-1111. doi: 10.1126/science.aau3445.
Graesser, J., Aide, T. M., Grau, H. R., & Ramankutty, N. (2015). Cropland/pastureland dynamics and the slowdown of deforestation in Latin America. Environmental Research Letters, 10(3), 034017.
http://doi.org/10.1088/1748-9326/10/3/034017
Hansen, M. P. Potapov, R. Moore, M. Hancher, S. Turubanova, A. Tyukavina, D. Thau, S. Stehman, S. Goetz, T. Loveland et al. 2013. “High-Resolution Global Maps of 21st-Century Forest Cover Change.” Science 341: 850-853. doi: 10.1126/science.1244693.
International Food Policy Research Institute (IFPRI) and International Institute for Applied Systems Analysis (IIASA). 2016. “Global Spatially-Disaggregated Crop Production Statistics Data for 2005 Version 3.2”, Harvard Dataverse 9. doi: 10.7910/DVN/DHXBJX.
IPCC 2003: Penman J., M. Gytandky, T. Hiraishi, T. Krug, D. Kruger, R. Pipatti, L. Buendia, K. Miwa, T. Ngara, Ngara, K. Tanabe et al. 2003. “Good Practice Guidance for Land Use, Land-Use Change and Forestry.” Institute for Global Environmental Strategies for the IPCC. Japan.
Miettinen, J., C. Shi, and S. C. Liew. 2016. “Land Cover Distribution in the Peatlands of Peninsular Malaysia, Sumatra, and Borneo in 2015 with Changes since 1990.” Global Ecology and Conservation 6: 67−78. doi:
10.1016/j.gecco.2016.02.004
Ramankutty, N., A. Evan, C. Monfreda, and J. Foley. 2008. “Farming the planet: 1. Geographic distribution of global agricultural lands in the year 2000.” Global Biogeochemical Cycles 22. doi:10.1029/2007GB002952.
Selvaradjou S., L. Montanarella, O. Spaargaren, D. Dent, N. Filippi, S. Dominik. 2005. “European Digital Archive of Soil Maps (EuDASM) – Metadata on the Soil Maps of Asia.” Office of the Official Publications of the European Communities. Luxembourg.
Wahyunto, S. Ritung, H. Subagjo. 2003. “Maps of Area of Peatland Distribution and Carbon Content in Sumatra, 1990-2002.” Wetlands International – Indonesia Programme & Wildlife Habitat. Canada.
Wahyunto, S. Ritung, H. Subagjo. 2004. “Maps of Area of Peatland Distribution and Carbon Content in Kalimantan, 1990-2002.” Wetlands International – Indonesia Programme & Wildlife Habitat. Canada.
Zarin, D., N. Harris, A. Baccini, D. Aksenov, M. Hansen, C. Azevedo-Ramos, T. Azevedo, B. Margono, A. Alencar, C. Gabris et al. 2016. “Can Carbon Emissions from Tropical Deforestation Drop by 50% in 5 Years?” Global Change Biology 22: 1336-1347. doi:
10.1111/gcb.13153