• Open Access

Extent of industrial plantations on Southeast Asian peatlands in 2010 with analysis of historical expansion and future projections

Authors


Abstract

Tropical peatlands cover over 25 Mha in Southeast Asia and are estimated to contain around 70 Gt of carbon. Peat swamp forest ecosystems are an important part of the region's natural resources supporting unique flora and fauna endemic to Southeast Asia. Over recent years, industrial plantation development on peatland, especially for oil palm cultivation, has created intense debate due to its potentially adverse social and environmental effects. The lack of objective up-to-date information on the extent of industrial plantations has complicated quantification of their regional and global environmental consequences, both in terms of loss of forest and biodiversity as well as increases in carbon emissions. Based on visual interpretation of high-resolution (30 m) satellite images, we find that industrial plantations covered over 3.1 Mha (20%) of the peatlands of Peninsular Malaysia, Sumatra and Borneo in 2010, surpassing the area of Belgium and causing an annual carbon emission from peat decomposition of 230–310 Mt CO2e. The majority (62%) of the plantations were located on the island of Sumatra, and over two-thirds (69%) of all industrial plantations were developed for oil palm cultivation, with the remainder mostly being Acacia plantations for paper pulp production. Historical analysis shows strong acceleration of plantation development in recent years: 70% of all industrial plantations have been established since 2000 and only 4% of the current plantation area existed in 1990. ‘Business-as-usual’ projections of future conversion rates, based on historical rates over the past two decades, indicate that 6–9 Mha of peatland in insular Southeast Asia may be converted to plantations by the year 2020, unless land use planning policies or markets for products change. This would increase the annual carbon emission to somewhere between 380 and 920 Mt CO2e by 2020 depending on water management practices and the extent of plantations.

Introduction

Southeast Asia contains 25 Mha of peatlands, equal to 56% of all tropical peatland areas (Page et al., 2011). These layers of organic soil with an average thickness of 5.5–7.0 m have been formed over thousands of years by accumulation of organic material in anaerobic and often acidic conditions of waterlogged lowlands (Rieley & Page, 2005; Page et al., 2011). Peatlands in this region form a considerable surface carbon deposit containing 11–14% (69 Gt) of global peat carbon (Page et al., 2011), equivalent to nearly eight times the annual amount of carbon released globally into the atmosphere by fossil fuel combustion in 2008 (9 Gt; Boden et al., 2010). Peat swamp forest ecosystems are an important part of the region's natural resources supporting unique flora and fauna endemic to Southeast Asia (Rieley & Page, 2005; Corlett, 2009). In addition, they serve as a refuge for endangered animal species such as orangutan and Sumatran tiger, as well as performing hydrological and other ecosystem functions (Morrogh-Bernard et al., 2003; Giesen, 2004; Rieley & Page, 2005).

Widespread human activities in peat swamp forests invariably set in motion ecosystem degradation. This affects the sensitive peat accumulation process which depends on the delicate balance between hydrology, ecology and landscape morphology (Page et al., 1999). Agricultural activities on peatlands require lowered water table levels that lead to aerobic conditions in the upper peat profile (see, e.g. Hirano et al., 2007; Jauhiainen et al., 2012) resulting in enhanced peat oxidation and increased carbon emissions to the atmosphere (Couwenberg et al., 2010; Hooijer et al., 2010, 2012).

Recent assessments of peatlands in insular Southeast Asia have revealed a dramatic reduction of peat swamp forest cover since 1985 and projections suggest that forest cover on peatlands will be considerably reduced over coming decades (Hooijer et al., 2006; Miettinen & Liew, 2010; Fuller et al., 2011; Miettinen et al., 2011a). Conversion of peat swamps to industrial plantations is often seen as one of the major causes of deforestation (see, e.g. Sarvision, 2011). In particular, oil palm cultivation for biofuel production has caused much controversy (Stone, 2007; Venter et al., 2008; Sheil et al., 2009). However, the quantification of the regional and global environmental consequences of plantation agriculture on Southeast Asian peatlands, both in terms of loss of forest and biodiversity as well as increases in carbon emissions, has been greatly complicated by a lack of objective and up-to-date information on the extent of industrial plantations on peat soil.

The majority of all industrial plantations on peatlands are large-scale oil palm and pulp wood (Acacia) plantations. These types of plantations can be delineated using visual interpretation of high spatial resolution (<30 m) satellite images (Miettinen & Liew, 2010; Wahid et al., 2010). In addition to large-scale industrial plantations, smaller areas of oil palm, coconut, pineapple, sago palm, rubber and other plantations exist, often managed by small-holder farmers. However, the heterogeneous structure and small size of the plantations make it practically impossible to assess the extent of various plantation crops cultivated by small-holder farmers at the regional level using currently available remote sensing datasets.

In this study, we investigated the extent of oil palm, pulp and other types of large-scale industrial plantations on the peatlands of Peninsular Malaysia, Sumatra and Borneo (Fig. 1). We used visual interpretation of high-resolution (10–30 m) satellite data to map industrial plantations in the region with an unprecedented level of detail and accuracy. The objectives of this paper are (1) to report the extent and spatial distribution of industrial plantations in 2010, (2) to analyse the historical development trends of industrial plantation agriculture since 1990 and (3) to derive projections for near-future plantation expansion on the peatlands of insular Southeast Asia based on ‘business-as-usual’ scenarios.

Figure 1.

Plantation distribution on the peatlands of the study area in 2010. Locations of insets presented in Fig. 2 are marked as black boxes. For colour legend please refer to Fig. 2.

Materials and methods

Peatland maps

Peatland areas in Sumatra and Kalimantan (Indonesian part of Borneo Island) were outlined using the Wetlands International 1 : 700 000 peatland atlases (Wahyunto et al., 2003, 2004). For Malaysia, information on the extent of peatlands was derived from the European Digital Archive of Soil Maps (Selvaradjou et al., 2005) as described in Miettinen & Liew (2010). The small country of Brunei, located in the northern coast of Borneo Island, was excluded from the analysis due to the unavailability of peatland maps.

Mapping of industrial plantations

For the 2010 mapping of industrial plantations, 100% of peatland areas were covered with 74 Landsat 7 ETM+ images acquired between 1 January 2010 and 11 March 2011. The images had 28.5 m spatial resolution. RGB band combination of 742 was used for on-screen viewing (band 7: 2.08–2.35 μm, band 4: 0.76–0.9 μm and band 2: 0.52–0.6 μm). Industrial plantations were manually delineated on screen based on visual interpretation of the images at a scale of 1 : 50 000.

For the historical analysis of the expansion of industrial plantations, the extent of plantation areas in 1990 and 2007 was derived from the land cover maps created earlier by Miettinen & Liew (2010) with Landsat and Satellite Pour l'Observation de la Terre (SPOT) data using the same visual interpretation-based mapping methodology described above. And finally, the 2000 plantation mapping was performed in the present study on the 2000 GeoCover product which is a 14.25 m resolution pansharpened mosaic of Landsat 7 ETM+ images acquired 1997–2003. The 2000 mapping was undertaken using the same methodology which had been used for all the above-mentioned mappings and the same RGB:742 used for the Landsat data in the 2010 mapping.

For consistency in the historical analysis, only areas covered by valid data (i.e. free of cloud cover, thick haze or other artefacts) in all four observations (1990–2000–2007–2010) were used. The overlapping valid data coverage over all observed time slices was 81% of peatlands in the entire study area. Lack of data coverage affected especially Peninsular Malaysia and East Kalimantan Province where only 52% and 36% of peatland areas were covered with valid data, respectively. This needs to be noted when assessing the results of the historical expansion of plantations in these provinces.

Plantation species identification

The plantation species identification was mainly performed on the set of SPOT images described by Miettinen & Liew (2010). Altogether 121 high spatial resolution (10–20 m) SPOT scenes were available. Due to the persistent cloudy weather conditions in this region, the data acquisition ranged over 3 years (2006–2008) with four additional images captured in 2005. Twenty seven of the images were captured in 2006, 58 in 2007 and 32 in 2008.

The plantation species identification was based on visual appearance (i.e. tone and texture) of the plantation areas, spatial arrangement of the plantation canals and roads, location, context, field knowledge of the interpreters and available land use allocation information. Plantations which could not be allocated to either oil palm or pulp due to lack of evidence or which were known to be of other plantation species were assigned to the ‘other/unknown’ class.

It was assumed that the plantation species did not change over the 20 year study period. This was considered to be a reasonable assumption since species suitable for industrial plantations on peatlands are very limited in number and infrastructure for palm oil and pulp processing remains in place for decades. We acknowledge that some sporadic exceptions to this rule may exist especially in those parts of the study area that have a long history of plantation development (e.g. Peninsular Malaysia and Sabah), but we believe that their effect on the results is marginal. Thus, the 1990 and 2000 plantation areas were assigned the same species as detected in 2007 in the specific locations. Species for new plantation areas established between 2007 and 2010 were determined by the location, appearance and spatial arrangement of the plantation area in the Landsat 7 ETM+ images together with personal knowledge and available land use allocation information.

Accuracy assessment

Very high-resolution satellite imagery acquired between 2004 and 2010 available in Google Earth was utilized to evaluate the accuracy of the methodology used in this study by comparing the very high-resolution samples to the 2007 map. The very high-resolution data available in Google Earth included images from the IKONOS and GeoEye-1 satellites operated by GeoEye as well as the Quickbird, WorldView-1 and WorldView-2 satellites operated by Digital Globe. These satellites have 0.5–1.5 m spatial resolution and enable accurate detection of industrial plantation areas including species identification.

The data used for the accuracy assessment covered 4% of the peatlands spread around 30 different sites. Within these sampling sites, 600 sample plots were selected using stratified random sampling. Half of the plots were selected from outside industrial plantations to estimate the level of omission errors of the mapping. Half of the sample plots were selected within areas classified as industrial plantations to evaluate the accuracy of the plantation species identification and the level of commission errors in the mapping. The 300 sample plots selected within plantation areas were distributed between different plantation species based on their proportions on the 2007 map.

Projections of future expansion

The projections of the potential expansion of industrial plantations by 2020 were based on recent expansion trends. The simplest (steady increase) projection assumed that plantation extent in coming years would expand at the same rate as it did in the most recent period of analysis, 2007–2010. This is a conservative approach, as the expansion rate has accelerated continuously over the last 20 years. A second model was derived by using the degree to which the increase from 2000 to 2010 had been greater than that over the 1990s. This takes into account the longer-term acceleration although still not the sharp acceleration since 2007.

It is important to understand that both of the future projection methods used in this study are purely based on historical trends of plantation expansion over the past 20 years and do not take into account any economical, political or other factors that may affect future plantation expansion rates. Therefore, the projections approximate the rate of future plantation development in ‘business-as-usual’ scenarios assuming that no significant changes in the circumstances affecting plantation expansion rates and distribution in the region would take place. The projections for 2020 plantation extent were, however, tested against the total area of available peatland that was not yet converted to plantation by 2010. For Indonesia, the projection was also tested against land allocation maps for 2011. For Malaysia, such maps were not available to us.

Calculation of carbon emission from industrial plantations on peatland

Recent extensive field analyses on heterotrophic carbon emission from plantations on tropical peatland (Jauhiainen et al., 2012) and of carbon loss as measured through subsidence (Hooijer et al., 2012) conclude that unit emissions at five or more years after drainage can be considered to be 75 t ha−1 yr−1 CO2e at water table depths around 0.75 m that are representative for actual conditions in most relatively well-managed plantations. Although the majority of the measurements in the above-mentioned papers were conducted in Acacia plantations, the similarity of the environmental factors regulating peat decomposition and a smaller amount of measurements conducted in oil palm plantations indicate that the emission estimate is generally applicable for all large-scale industrial plantations in Southeast Asia (Hooijer et al., 2012). This estimate agrees well with earlier findings on carbon emissions from peat decomposition in the region (Couwenberg et al., 2010; Hooijer et al., 2010). If ‘best practice’ water management is introduced and water levels could be kept at 0.6 m, these emissions could potentially be lowered to around 63 t ha−1 yr−1 CO2e following the relations reported by Hooijer et al. (2012) and Jauhiainen et al. (2012). If, however, the initial high carbon emission peak occurring within the first 4 years after the establishment of the plantation highlighted by Hooijer et al. (2012) is included, and water levels remain at 0.75 m depth, plantations can be expected to emit 100 t ha−1 yr−1 CO2e, calculated as an average over a 25 year period after drainage.

In this study, these emission estimate numbers were multiplied with the number of hectares of plantation to quantify total carbon emissions from peat decomposition in industrial plantations in the region. The resulting total emission numbers are considered low estimates since the loss of preplantation above ground vegetation biomass and fire induced emissions during plantation development were excluded. It is also important to understand that the emission estimates presented in this study are based on the actual plantation area coverage alone and do not take into account impacts beyond the plantation limits, which can be substantial (see, e.g. Hooijer et al., 2012).

Preplantation vegetation cover

Taking advantage of two recently published land cover maps (Miettinen & Liew, 2010; Miettinen et al., 2012a), we were able to analyse the preplantation vegetation cover of new plantation areas. In this analysis, the plantation maps were overlaid on maps describing the land cover distribution in a given area a decade before a plantation was detected. Those areas that were first detected as plantations in 2000 were compared with 1990 land cover. Similarly, the plantation areas that had been established between 2000 and 2010 were compared with 2000 land cover. It has to be acknowledged that the plantations have most likely not directly replaced the land cover type that was detected in the given location a decade earlier. Plantation establishment often is a gradual process which typically goes through several phases. Nevertheless, we believe that the land cover on a given area, a decade before a new plantation is detected, describes adequately the type of vegetation that was present in the future plantation areas when the plantation development process was set in motion.

Results

Accuracy assessment results

Overall accuracy of the industrial plantation mapping including species identification reached 94% with a kappa value of 0.91. Both oil palm and pulp wood plantations had user's and producer's accuracies of 90% or better (Table 1). The accuracy assessment also revealed that around half of the plantations classified as ‘other/unknown’ were in fact shown to be oil palm plantations. In theory, this could lead to an underestimation of the extent of oil palm plantations. Owing to the very small proportion of ‘other/unknown’ plantations (~3%), however, any potential underestimation due to this cause can be expected to be in the order of only 1–2%. On the whole, these statistics indicate that the maps used in this study enable reliable analysis on the current extent and historical expansion of industrial plantations in the study area.

Table 1. Error matrix for mapping of industrial plantations
 ReferenceTotalUser's accuracy
Oil palmPulpOther/unknownOther mapped
LC map
 Oil palm18504619594.9
 Pulp489069989.9
 Other/unknown4011616.7
 Other mapped60429030096.7
Total199899303600 
Producer's accuracy93.0100.011.195.7  

Extent of industrial plantations in 2010

Our results show that over 3.1 Mha of peatland had been converted to industrial plantations by 2010 (Table 2), equal to 20% of peatlands in the study area. Industrial plantations were, however, not evenly distributed in the region. Over 60% were found in Sumatra Island which contains less than half (47%) of peatlands in the study area (Fig. 1). Overall, industrial plantations covered 26% of the peatland areas in Sumatra and 13% of peatlands in Borneo. Furthermore, there was a clear difference between the Malaysian part of Borneo (i.e. Sarawak and Sabah states) and the Indonesian part of the island (i.e. Kalimantan). In Sarawak and Sabah, 36% and 27% of peatlands had been converted to plantations, respectively, whereas none of the provinces in Kalimantan had more than 10% plantation cover in 2010. Overall, Riau and South Sumatra provinces in Sumatra and Sarawak state in Borneo contained 62% of all industrial plantations in the region (Fig. 2).

Figure 2.

Extent of plantations on the peatlands of Riau, South Sumatra and Sarawak in 2010. Locations of the insets within the region are outlined in Fig. 1.

Table 2. Distribution of industrial plantations (IP) in Peninsular Malaysia, Sumatra and Borneo in 2010a (in 1000 ha)
 Oil palm IPPulp IPOther/unknown IPTotal IPIP% of total peatland
Area% Within island% Within study areaArea% Within island% Within study areaArea% Within island% Within study areaArea% Within island% Within study area
  1. a

    Note that the 2010 mapping covers 100% of peatland areas.

Peninsular   Malaysia238NA110NA023NA20262NA829
 Aceh4642000000462117
 North Sumatra19819900015120010657
 Riau48746234645554185515968513124
 West Sumatra8984000000895342
 Jambi848463870001468520
 Bengkulu200000414470013
 South Sumatra13112630937369277449231431
 Lampung1010000000101011
Total Sumatra10471004983610098321002819151006226
 Sarawak4945823000315026525561736
 Sabah5062000242526227
 West   Kalimantan133166106311322111571759
 Central   Kalimantan1141350004631181344
 South   Kalimantan2831000463313110
 East   Kalimantan39526371813753628
Total Borneo8581004016100262100539361003013
Total study   area2143 100852 100118 1003113 10020

Over two-thirds (69%) of all industrial plantations in 2010 were used for oil palm cultivation, with the remainder being mainly pulp wood (27%) and a small extent of other (4%) plantations (Table 2). While oil palm plantations occurred in all parts of the study area, pulp wood plantations were very unevenly distributed, with 98% being found in three Sumatran provinces: Riau, Jambi and South Sumatra. The distribution of pulp wood plantations is dependent on the location of pulp mills, which require far larger investments than the relatively small-scale infrastructure necessary for palm oil processing. They are, therefore, limited in number (there are two pulp mills in Riau) and slow to be developed in new areas.

Historical expansion of industrial plantations with future projections

Analysis of the overlapping areas of the plantation maps of 1990, 2000, 2007 and 2010 (81% of total study area) allowed us to investigate the expansion of industrial plantation areas in the region since 1990. Apart from Peninsular Malaysia, where plantation agriculture on peatlands was already well established by 1990, nearly all peatland plantations have been created over the past 20 years (Fig. 3). Furthermore, plantation development has accelerated fast with 70% of plantations established over the decade 2000–2010 and 27% within the 3 years of 2007–2010 alone. The acceleration of plantation development since 2000 has been particularly fast in South Sumatra, Riau and Sarawak which together accounted for 75% of all new plantations established since 2000 (Table 3). The recent increase in plantation development in West Kalimantan must also be highlighted, with 48% of the plantations existing in 2010 being established since 2007. Nevertheless, throughout the study period, Riau province experienced the main concentration of new plantation development (Table 3). With this widespread and rapid rate of plantation development and large extent of peatland (26% of the peatland in the study area), Riau Province dominated all Provincial (Indonesian) and State (Malaysian) level peatland plantation statistics in the region.

Figure 3.

Expansion of industrial plantations on peatland in Sumatra, Borneo and Peninsular Malaysia since 1990 and projection of future expansion until 2020.

Table 3. Distribution of new plantation establishment in the 1990s and in the 2000s (in %)
 1990–20002000–2010Proportion of peatland mapped
Within islandTotalWithin islandTotal
Peninsular MalaysiaNA4NA152
Aceh332178
North Sumatra21165383
Riau5946493397
West Sumatra753284
Jambi337581
Bengkulu110093
South Sumatra33342397
Lampung210097
Total Sumatra100771006893
Sarawak7213621978
Sabah1832180
West Kalimantan7122779
Central Kalimantan204175
South Kalimantan005280
East Kalimantan005136
Total Borneo100181003173
Total study area 100 10081

Based on the historical expansion of plantation areas presented above, using two different projection models (steady increase and decadal acceleration), we estimated that in 2020 industrial plantations would cover between 6.0 and 9.2 Mha of peatland in the study area (Fig. 3). Further analysis on potential limitations for future expansion (see more details in Miettinen et al., 2012b) revealed that in no Province in Indonesia, nor in the Malaysian States in Borneo, would the total peatland area limit further plantation expansion by 2020. For Indonesia, the projection was also tested against land allocation maps for 2011, and it was found that the area allocated for plantation expansion would easily accommodate the projections. This may not, however, be the case in some parts of Malaysia where available land resources are scarcer. Limited availability of land for agricultural development on mineral soils, e.g. in Sarawak, may increase the pressure on conversion of peatlands leading to shortage of area available for further plantation development. This shortage of peatland in Malaysia may become effective even sooner if land allocation laws begin to limit availability of the remaining peatlands; however, the status of such laws currently remains unknown to us.

Preplantation vegetation cover

The analysis of the preplantation vegetation cover revealed that in the 1990s plantation development took place almost entirely in forested areas. Eighty-five percent of all new plantations detected in 2000 were covered by peat swamp forest in 1990 (Table 4). Between 2000 and 2010, the proportion of forest conversion to plantation dropped to 56%. However, this was mainly due to the seemingly low peat swamp forest conversion rates for new plantations in South Sumatra and Central, South and East Kalimantan provinces where vast areas of forest had already burnt in the late 1990s (Miettinen et al., 2011a). The three high intensity plantation development areas of Riau, Sarawak and West Kalimantan, on the other hand, continued to show around 80% conversion rates from peat swamp forest to industrial plantations since 2000 (Table 4).

Table 4. Preplantation vegetation (in %)
 1990 vegetation of plantations detected in 20002000 vegetation of plantations detected in 2010
PSFRegrowthMosaicOpenPSFRegrowthMosaicOpen
  1. Class descriptions: PSF stands for peat swamp forest. The regrowth-class includes all natural and planted secondary vegetation excluding industrial plantations. The mosaic-class includes areas with a mixture of woody vegetation and open land. Typically, this land cover class occurs in either sparsely vegetated newly cleared areas or small-holder farming areas. The open-class includes agricultural fields and areas that are either bare or covered by ferns typically less than 2 m in height.

Peninsular Malaysia56892733391216
 Aceh97012464077
 North Sumatra7641195524911
 Riau93106831035
 West Sumatra88201054221113
 Jambi8314034136194
 Bengkulu4728322072262
 South Sumatra751825117298
 Lampung13150720000
Total Sumatra863110523467
 Sarawak92431741745
 Sabah78621433242617
 West Kalimantan9221586941
 Central Kalimantan23227319303615
 South Kalimantan00000313732
 East Kalimantan10000015233032
Total Borneo88426671698
Total854110562977

Carbon emissions from peat decomposition in industrial plantation area

Assuming average water table depths of 0.75 m, the industrial plantation area of 2010 is estimated to produce a carbon emission from peat decomposition of at least 233 Mt yr−1 CO2 equivalents, up from 79 Mt yr−1 CO2e in 2000 and 20 Mt yr−1 CO2e in 1990. Of the total emission in 2010, 161 Mt yr−1 CO2e (69%) is caused by oil palm plantations. If the initial peak in emissions within the first 5 years after drainage is included, and the total emissions are averaged over a 25 year period, the yearly emissions from the industrial plantation area in 2010 would reach up to 311 Mt yr−1 CO2e. Depending on the plantation expansion projection method, the 2020 emission from industrial plantations on peatlands in this part of Southeast Asia would be somewhere between 447 and 688 Mt yr−1 CO2e (including emissions of 308–465 Mt yr−1 CO2e from oil palm plantations alone) based on the assumption of 0.75 m water table depth or up to 596–917 Mt yr−1 CO2e (of which 410–620 Mt yr−1 CO2e is from oil palm) if the initial peak in emissions is included. If, however, we assumed that ‘best practice’ water management could be achieved in all plantations in future, and if we exclude the initial emission peak, total emissions from industrial plantations in 2020 would be between 376 and 578 Mt yr−1 CO2e (including emissions of 258–391 Mt yr−1 CO2e from oil palm plantations).

Discussion

In this paper, we have presented the extent and spatial distribution of industrial plantations in the peatlands of Peninsular Malaysia, Sumatra and Borneo in 2010, with analysis of expansion since 1990 and projections for future development. In general, our results agree well with previous studies. Hooijer et al. (2006) estimated that in 2010 around 3 Mha of Southeast Asian peatlands would be covered by industrial plantations. In this study, we revealed that 3.1 Mha, or 20% of peatlands in the study area, had been converted to industrial plantations by 2010. Based on a newly published 250 m resolution regional land cover map, large-scale oil palm plantations covered 0.88 Mha in the peatlands of the study area around the year 2000 (Miettinen et al., 2012a), compared with our high-resolution (10–30 m) assessment resulting in 0.93 Mha.

In a recent study performed with high-resolution satellite data, Wahid et al. (2010) estimated that less than 0.7 Mha of peatland was covered by oil palm plantations in Malaysia in 2009. This is significantly less than the over 0.8 Mha estimated in this study for 2010. It must be remembered, however, that oil palm expansion is currently very fast in some parts of Malaysia as documented by both Sarvision (2011) and our study. In their analysis, Wahid et al. (2010) used satellite data acquired in 2008–2009, as opposed to some images in our study which were acquired as late as March 2011. This indicates that in some areas, the difference in acquisition dates of satellite data may have been nearly 3 years.

Overall, it is important to remember that this study concentrated on the western part of insular Southeast Asia. Plantations in the Indonesian provinces of Papua and West Papua, as well as the countries of Papua New Guinea, Brunei and Thailand, were excluded from the analysis. The total extent of industrial plantations on peatland in the entire Southeast Asian region will be higher than presented in this paper, although none of the areas missed in this study are known to have had extensive plantation agriculture established on peatlands by 2010.

The projection methods used in this study simply assume continuation of recent plantation development trends in ‘business-as-usual’ scenarios, assuming that various aspects regulating the rate and distribution of the expansion of industrial plantations in this region would not change significantly. We acknowledge that this approach has several limitations. First, it does not take into account the demand drivers, under different scenarios for economic development, and potential biofuel market regulation. Second, as the longer term risks and costs of draining peatlands, namely land subsidence and reduced drainability in time, become clearer to companies, they may potentially choose to focus on plantation development on mineral dryland soils, where tens of millions of hectares of deforested land (particularly in Indonesia) are currently unproductive.

Moreover, it is not known what the response of local Governments will be to the international and national pressures to reduce greenhouse gas emissions from peatlands and to conserve remaining wetland forests. Enforcement of existing and potential future rules banning drainage and deforestation of peatlands is still possible. One example of a recent political development is the 2 year moratorium on allocation of new concession areas on forested peatlands as part of a bilateral cooperation between Indonesia and Norway (Murdiyarso et al., 2011). However, the final forms of implementation of the moratorium at a practical level are still taking shape and the developments beyond the 2 year period are uncertain. Therefore, it is unclear how and to what extent the moratorium and future political decisions will affect the expansion of industrial plantations on the peatlands of Southeast Asia during this decade. In the absence of information on the abovementioned issues, we believe that our projection estimates represent a valid indication of the future direction of plantation expansion in this region under ‘business-as-usual’ scenarios that continue the practices that have prevailed over the past 20 years.

We estimate that in 2010 industrial plantations produced carbon emission to the atmosphere of at least 233 Mt yr−1 CO2 equivalents, or up to 311 Mt yr−1 CO2e if the initial peak in carbon emissions after plantation development (Hooijer et al., 2012) is included. These figures exceed the total annual carbon emission from fossil fuel burning of a country like Malaysia (208 Mt yr−1 CO2e in 2008; Boden et al., 2010). Note also that the values presented in this paper, based on updated knowledge on carbon emissions from peat decomposition and on extensive new high-resolution plantation mapping, are substantially higher than some earlier estimates; e.g. a recent paper (Koh et al., 2011) suggested 17 Mt yr−1 CO2e emissions for 0.88 Mha of oil palm plantations around the year 2000 as opposed to our estimate of 70 Mt yr−1 CO2e for 0.93 Mha in 2000. Furthermore, it is important to understand that the 233 Mt yr−1 CO2e emission estimate presented in this study excludes emissions from drained peatlands under other land covers in the region, including adjoining areas that are affected by plantation drainage, as well as emissions from peatlands outside the study area (e.g. Papua). In comparison, biomass burning carbon emissions from the entire equatorial Southeast Asia, largely attributed to peatland fires in this region, averaged around 700 Mt yr−1 CO2e between 1997 and 2009 (van der Werf et al., 2010). However, this estimate was heavily affected by an extreme fire episode in 1997–1998. Since 1999, fire induced carbon emissions have ranged between 80 and 1300 Mt yr−1 CO2e, averaging at 400 Mt yr−1 CO2e.

Amid the intense debate related to industrial plantation development on Southeast Asian peatlands, it is important to remember that plantation development is not the only threat that tropical peatlands are currently facing. A combination of the industrial plantation mapping performed for this study and a recently published regional land cover map (Miettinen et al., 2012a) allowed us to analyse the proportion of industrial plantations on all deforested peatlands (Table 5). In addition to an overall 20% coverage of industrial plantations, secondary regrowth and small-holder plantations were estimated to cover 27% of peatlands in the study area. Furthermore, a mosaic of vegetated and nonvegetated areas (including both natural and managed land use types) covered 9% and open clearances or extremely degraded areas 10%. Only 34% of the peatlands in the study area remained forested in 2010 (Miettinen et al., 2012a), down from around 75% in 1990 (Miettinen & Liew, 2010). This deforestation rate of nearly 4% yr−1 substantially exceeds earlier analyses of historical and projected forest losses (e.g. Hooijer et al., 2006) and is indeed dramatically higher than deforestation levels generally around the world (Hansen et al., 2010) and in other parts of this region (Miettinen et al., 2011b), especially when it is considered that much of the remaining peat swamp forest is also degraded to some extent (Miettinen & Liew, 2010).

Table 5. Land cover distribution on the peatlands of Peninsular Malaysia, Sumatra and Borneo in 2010
 Peat swamp forestRegrowthMosaicOpenIndustrial plantationOther nonforestTotal
1000 ha%1000 ha%1000 ha%1000 ha%1000 ha%1000 ha%1000 ha
  1. Please refer to Table 4 for land cover class descriptions.

Peninsular Malaysia23026182201281476926229131890
 Aceh108399434186104461700277
 North Sumatra25769203392162005700348
 Riau13823410512632682637968242414014
 West Sumatra20962293416178783710211
 Jambi181252553675106081462010717
 Bengkulu00244614267147131152
 South Sumatra10775373713492201544931201450
 Lampung33242667454910113392
 Other provinces24323345461115001174
Total Sumatra185026215030643965591904263207234
 Sarawak380264032847382652536601443
 Sabah392049261583317522732191
 West Kalimantan10426031818945101615793121743
 Central Kalimantan1454487802629110357121184803009
 South Kalimantan419629461414945311031329
 East Kalimantan25036207306291101653871688
Total Borneo316943185325555783211936135717403
Total52493441862713269156310310220102115528

The results reported in this paper highlight the accelerating nature of industrial plantation expansion on peatland since 1990. Projections indicate that half of the peatland area on the islands of Sumatra and Borneo and in Peninsular Malaysia may be covered by plantations by 2020 if current expansion trends persist. Furthermore, under current policies, there is no indication that expansion will stop in 2020. This expansion is also likely to result in degradation well beyond the plantation limits as peatland drainage and access roads have an impact over at least 2 km (Hooijer et al., 2012). Bearing in mind that other causes of peatland deforestation, including small-holder agriculture and logging, are also still expanding, it would appear that very little peatland forest is likely to remain in Southeast Asia by the end of the current decade unless land use planning policies are changed or markets for palm oil and pulp products from these areas are reduced.

Acknowledgements

This research was mainly funded by the International Council on Clean Transportation (ICCT). J.M., C.S. and S.C.L. also acknowledge financial support from the Agency for Science, Technology and Research (A*STAR) of Singapore to the Centre for Remote Imaging, Sensing and Processing (CRISP).

Ancillary