An analysis of multiple ecosystem services under future oil palm expansion scenarios in Central and West Kalimantan, Indonesia

Based on the current land-use policy, spatial planning and maps of oil palm expansion in Indonesia, this study identified three plausible future scenarios; namely business as usual, conservation and sustainable intensification for future development of oil palm plantations in the study area.This study concludes that ecosystem services are negatively impacted in the business-as-usual scenario due to expansions of oil palm plantations on old-growth forest and regrowth forest. Assuming the lowest intensity of oil palm expansion, the conservation scenario enhances carbon stock and maintains a stable habitat quality relative to the current land use. The sustainable intensification scenario with oil palm expansions only on suitable areas and enhancement of yield generates a positive impact on carbon stock and water yield, whereas habitat quality slightly deteriorates in the study areas. A detailed study at a local level (household or village) that evaluates the economic value of crucial ecosystem services and their trade-offs or synergies is recommended for accurate evaluation to better understand the impacts of oil palm expansion on the local community and the environment.


Executive summary
This study provides an analysis and evaluation of ecosystem services and their trade-offs under the future scenarios of oil palm expansions in two provinces, Central Kalimantan and West Kalimantan, of Indonesia. The future land-use scenarios for oil palm expansions are based on the historical rates, current land-use policies (such as the forest moratorium), spatial planning and the existing suitability maps for oil palm plantations in both provinces. An extensive review of the literature and published documents and reports were used to understand past and future land-use changes, impacts on ecosystem services and socioeconomic conditions. National and provincial land-use policies and factors relevant to oil palm development were explored and analyzed. Further, relevant experts and stakeholders were consulted to consolidate knowledge for future land-use changes and oil palm development scenarios in the study area. The spatial analysis tool in ArcGIS and the Integrated Valuation of Ecosystem Services and Trade-offs Tool (InVEST) were used to analyze historical and future land-use changes, and to evaluate and analyze trade-offs with ecosystem services. This study focused on mapping, quantifying and evaluating the following five most relevant and vital ecosystem services: (i) carbon storage, (ii) habitat quality, (iii) water yield, (iv) palm oil production, and (v) timber production.
Based on the current land-use policy, spatial planning and maps of oil palm expansion in Indonesia, this study identified three plausible future scenarios; namely business as usual, conservation and sustainable intensification for future development of oil palm plantations in the study area. The businessas-usual scenario assumes that oil palm concession license holders will expand their plantations to all concession areas to access the economic opportunities of the high demand for palm oil for domestic as well as international demands. The conservation scenario considers that stringent criteria will ensure environmental integrity and sustainability. The forest moratorium on primary forest and peatland will be supported by strict enforcement and monitoring to avoid the clearing of intact forest and peatland for oil palm and timber plantations. The sustainable intensification scenario considers that future expansions will occur within land suitable for oil palm plantations by applying yield enhancement technology.
Trade-offs and synergies with these ecosystem services were found to depend on the extent and types of land-use transition to oil palm plantations under the selected future land-use scenarios. Except for timber production, there are trade-offs in the four other ecosystem services under the businessas-usual scenario, primarily due to expansions of oil palm plantations on old-growth and regrowth forests. However, the model results showed synergy between increased palm oil yield (due to an extensive expansion of oil palm plantations) and water yield (reducing). This synergy occurred under all future land-use scenarios due to the conversion of agricultural land, scrublands and other land uses into oil palm plantations.
This study concludes that ecosystem services are negatively impacted in the business-as-usual scenario due to expansions of oil palm plantations on old-growth forest and regrowth forest. Assuming the lowest intensity of oil palm expansion, the conservation scenario enhances carbon stock and maintains a stable habitat quality relative to the current land use (2016). The sustainable intensification scenario with oil palm expansions only on suitable areas and enhancement of yield generates a positive impact on carbon stock and water yield, whereas habitat quality slightly deteriorates in the study areas. A detailed study at a local level (household or village) that evaluates the economic value of crucial ecosystem services and their trade-offs or synergies is recommended for accurate evaluation to better understand the impacts of oil palm expansion on the local community and the environment.
Indonesia is the principal producer and exporter of palm oil to global markets (FAOSTAT 2017). Its palm oil production doubled between 2007 and 2017 from 18 million tons (Mt) to 36 Mt and met about 54% of the total global production (IndexMundi 2017). Primary drivers for this increase are considered to be: (i) highest yields (up to 12 tons per hectare per year) among all oilseed crops (Woittiez et al. 2017;Yan 2017); (ii) low production costs (Corley 2009) and high profit margins (Koh and Wilcove 2007); and (iii) growing domestic consumption of vegetable oil (Gaskell 2015) and biofuels (Kharina et al. 2016). The domestic consumption of palm oil is rapidly increasing in Indonesia,reaching 8.6 Mt in 2015 (Pacheco et al. 2017) from 5 Mt in 2010 (Gaskell 2015). If this trend continues, the predicted domestic demand of 10 Mt by 2035 (Gaskell 2015) will be reached much earlier. Further, based on population projections and per capita consumption, the demand for edible oilseed is predicted to double to 240 Mt by 2050, which would require an additional 12 million hectares (Mha) of oil palm plantations (Corley 2009).
The increased production of palm oil has resulted in significant land-use change in Indonesia. In 2015, oil palm plantations accounted for over 11 Mha (Kharina et al. 2016), or 6% of Indonesia's total area. Recent spatial data analysis showed a rapid and extensive expansion of oil palm plantations since the 1990s (Gunarso et al. 2013;Gaveau et al. 2014Gaveau et al. , 2016. Between 1990 and 2015, industrial plantations 1 expanded to about 5.5 Mha in Kalimantan, a greater than ninefold increase in area since 1990 (Gaveau et al. 2016). Between 2005 and 2015, the same study observed an expansion of industrial plantations in Kalimantan by almost three times the average annual expansion (357,000 ha per year) on the previous decade (127,000 ha per year), i.e. between 1995 and 2005.
The unprecedented expansion of oil palm plantation has significantly changed the landscape of Indonesia in the past few decades (Gunarso et al. 2013;Busch et al. 2015;Eldeeb et al. 2015). In addition to oil palm plantation on already cleared land or degraded land, industrial scale plantations were established on old-growth forests in Indonesia. A recent study estimated that out of 14.4 Mha of old-growth forest cleared between 1973 and 2015, oil palm plantations were established on more than half of that area, i.e. 7.8 Mha, and an additional 1.3 Mha area was used for pulpwood plantations (Gaveau et al. 2016). However, the analysis of time series data for landuse change showed that illegal logging and forest fires were the primary drivers of deforestation of old-growth forest before oil palm plantations were established (Gaveau et al. 2016).
The impacts of land-use change have been used to understand the gain or loss of ecosystem services in Indonesia. Many researchers have assessed and reviewed the impacts of oil palm plantations on the availability of multiple ecosystem services (e.g. Sumarga and Hein 2014;Dislich et al. 2016;Petrenko et al. 2016;Sumarga and Hein 2016;Vijay et al. 2016). They reported a significant loss of ecosystem services in landscapes modified by oil palm plantations compared with previous land uses, such as primary forests and peatlands. Sumarga and Hein (2014) observed the loss of timber production, destruction of orangutan habitat and increased carbon emissions resulting from the conversion of peatlands to oil palm plantations in Central Kalimantan, Indonesia. A review of studies confirms that the biodiversity loss is more evident in oil palm plantations than in forest and other tree covers (Fitzherbert et al. 2008). A global study highlighted that the potential expansion of oil palm into vulnerable forests posed the threat of extinction of mammal and bird species (Vijay et al. 2016). Further, Dislich et al. (2016) found that conversion of forest into oil palm plantation results in a reduction of 11 out of 14 ecosystem functions (or services) available from the forest, with the exception of food and raw materials.
However, the change in ecosystem services depends on the previous land use prior to establishing an oil palm plantation. Dislich et al. (2016) also reported that oil palm plantations on already cleared or deforested land can enhance the land's capacity by producing a high-value commodity (i.e. palm oil), as well as offering some positive impacts on multiple ecosystem services such as climate regulation, water regulation and erosion control relative to the deforested land. It is evident that oil palm plantation offers an additional sequestration of carbon on above-and belowground biomass. Oil palm plantation enhances carbon stocks by an average of 50.5 tC per hectare (Sumarga and Hein 2014) relative to the deforested land.
Despite the reported negative impacts on the delivery of ecosystem services, oil palm plantations have become an important contributor to Indonesia's national economy and support the livelihoods of rural people. Indonesia's export revenue from palm oil increased by over fivefold from US$3.4 billion in 2004 to US$17.5 billion in 2014 (Pacheco et al. 2017) and will continue to rise with the growth in palm oil production. A study in Riau Province found that industrial oil palm plantations enhanced local economic growth with positive impacts on livelihoods through increasing access to basic needs (e.g. school, health) and other development opportunities (e.g. roads, electricity, banks) (Budidarsono et al. 2013). Conversion of about 93,000 ha of peatlands to smallholder oil palm plantation in Siak generated over 37,000 jobs and increased the average household income by US$4556 per year (Agustira et al. 2016). The net economic benefit was estimated to be US$1036 million per year to the community after accounting for the environmental cost of US$1116 million per year due to conversion of peatlands to oil palm plantation.
However, the expansion of oil palm plantation depends on permits issued by government authorities and is guided by existing land-use policies and plans. The forest moratorium initially adopted in May 2010 applies to all new oil palm concessions and restricts further clearing of primary forests and peatlands for oil palm development (Murdiyarso et al. 2011;Austin et al. 2014). In 2011, the indicative moratorium map delineated a total of 22.5 Mha, which includes 7.2 Mha of primary forests, 11.2 Mha of peatlands and an additional 4.1 Mha of other land across Indonesia (Murdiyarso et al. 2011). Although the moratorium is criticized for its narrow scope in applying only to primary forests and peatlands (Busch et al. 2015), this policy can be effectively used to restrict oil palm expansion to outside these areas and also to support Indonesia's voluntary commitment to abolish deforestation from palm oil production and supply chains by 2020 (UN Climate Summit 2014).
The recent history and driving forces for oil palm expansion imply that future expansion of plantations is inevitable in Indonesia for the following reasons: (i) ever-growing demand for palm oil in both domestic and global markets (e.g. Abdullah 2011; Kharina et al. 2016); (ii) contribution to national and local economic growth (e.g. Agustira et al. 2016;Pacheco et al. 2017); and (iii) expansion of oil palm in smallholder farms, which accounted for about 40% of the total oil palm plantation area in 2013 (Daemeter Consulting 2015). Understandably, many factors, including regulatory, social, economic and environmental, interact to determine the actual expansion of oil palm plantation on the ground. However, there have been attempts to delineate the areas for future expansion of oil palm plantation based on land use and characteristics. The available maps showing potential areas for future oil palm expansion in Indonesia include: (i) the provincial spatial land-use planning maps developed by Central Kalimantan and West Kalimantan provincial government, which identify lands based on their functions and delineate the land available for other uses including agricultural land and oil palm expansion (CIFOR 2014); (ii) the World Resource Institute's (WRI) oil palm concession maps, which show all oil palm plantation areas where the palm industries have already secured rights to develop oil palm as of 2010 -these companies have permits | 3 from relevant government authorities to clear forest, plant and manage oil palm plantations; and (iii) the suitability maps that identify potential areas for oil palm expansion using various suitability criteria for sustainable palm oil production by avoiding clearing of forests, peatland and threats to loss of biodiversity (e.g. Mantel et al. 2007;Gingold et al. 2012).
This study is designed to examine the change in ecosystem services and their trade-offs for the period between 2017 and 2035. It focuses on oil palm expansion in two provinces, Central Kalimantan and West Kalimantan, of Indonesia. The future land-use scenarios for oil palm expansion are based on the historical rates, current land-use policies, such as the forest moratorium, spatial planning and the existing suitability maps for oil palm plantations in both provinces. The specific objectives of this study are to: • identify plausible future land-use scenarios associated with oil palm expansion in Central Kalimantan and West Kalimantan • quantify, evaluate and map key ecosystem services using ecosystem services mapping and modeling tools such as InVEST • analyze trade-offs or synergies among multiple ecosystem services under different future landuse scenarios • estimate and compare the total economic valuation of the ecosystem services under contrasted future oil palm expansion scenarios.
The economic valuation of ecosystem services enables a comparison of the gain or loss of the associated values of the ecosystem services due to land-use change under the future land-use scenarios. The research outcomes will inform management decisions at local, regional and national levels and assist stakeholders, such as government and nongovernment officials, the private sector and communities, to visualize the future expansion of oil palm in the study areas and to understand their impacts on the key ecosystem services. The economic valuation of the ecosystem services illustrates their value in monetary terms and provides evidence and justification for future sustainable land-use planning. However, this study is a desktop analysis only and is restricted to examining the impacts of future oil palm expansion on ecosystem services in the study areas examined.

Study area
The study areas comprised two provinces in Kalimantan, the Indonesian part of Borneo, i.e. Central Kalimantan and West Kalimantan ( Figure  1). This region has been identified as a part of one of the 25 biodiversity hotspots for conservation priorities in the world because of its high number of endemic species and the threat of habitat loss (Myers et al. 2000). Both provinces encompass some parts of the 'Heart of Borneo', the central, mountainous region of Borneo with intact natural rainforests and biodiversity, which supports the livelihoods of millions of people, including forestdwelling indigenous Dayaks (WWF 2013).
Central Kalimantan (Kalimantan Tengah) is the second largest province in Indonesia, with a total territorial area of 153,564 km 2 . Geographically, it is located at latitude 0°45'N-3°30'S and longitude 110°45'-115°50'E. This province comprises 13 regencies with a total population of over 2.2 million in the last census 2010 (BPS 2017). More than the half of the province has been assessed as high conservation value (HCV) areas. However, about two-thirds of the HCV areas are considered at high risk due to planned development activities (Ibie et al. 2016 Figure 2 shows the framework outlining the research methods and tools used in this study to accomplish the objectives. An extensive review of the literature and published documents/ reports helped explain past and future landuse change, impacts on ecosystem services and socioeconomic conditions. National and provincial land-use policies and factors relevant to oil palm development were explored and analyzed. Further, relevant experts and stakeholders were consulted to consolidate knowledge and understanding for future land-use change and oil palm development scenarios in the study area. A total of 35 stakeholders, representing government, nongovernment, industry and smallholder farmers, participated in a field-level consultation workshop held in Palankaraya, Central Kalimantan on 21 November 2016. Three potential future oil palm expansion scenarios were identified based on the existing land-use policies, suitability maps for oil palm in the study areas and the stakeholders' perspectives on the future of oil palm. The spatial analysis tools in ArcGIS version 10.2 were used to derive future land-use maps under the selected oil palm expansion scenarios.

Research framework
The Integrated Valuation of Ecosystem Services and Trade-offs Tool (InVEST) was chosen to evaluate ecosystem services and analyze tradeoffs. The key reasons for the selection of this particular tool were: (i) freely available computer software; (ii) widely used in terrestrial ecosystems and tested globally; (iii) capacity for generating spatially explicit maps; (iv) enables analysis of trade-offs between different land-use scenarios; and (v) provides specific models for various ecosystem services. The InVEST modules for the selected key ecosystem services were run for each scenario and the outputs were analyzed for spatial and temporal distribution of the ecosystem services and their trade-offs resulting from land-use change under these scenarios.

Analysis of oil palm expansion and mapping of future land use under oil palm expansion scenarios
To analyze the expansion of oil palm plantations between 2000 and 2016, this study used a recent spatial dataset from Borneo developed by Gaveau et al. (2016). The dataset represents a comprehensive methodology involving digitization of LANDSAT time series imagery followed by map refinement through ground verification, expert review and crosschecking with other datasets.   palm plantations, oil palm on smallholder farms has significantly increased in Indonesia (Jelsma et al. 2009;Sheil et al. 2009 Table 2 defines the land-use classes in the 2016 land-use map for the study areas. Section 3 provides further detail on these land-use classes and also presents maps of the study areas.

Identifying and mapping of future oil palm expansion scenarios
Based on the current land-use policy, spatial planning and existing maps of oil palm expansion in Indonesia, three future plausible scenarios, namely, business as usual, conservation and sustainable intensification were identified in the study areas. Table 3 summarizes the key features of these scenarios.
These future land-use scenarios and mapping procedures are provided below.

Business-as-usual scenario
The business-as-usual scenario assumes that oil palm concession license holders will expand their plantations to all concession areas to access economic opportunities created by the high demand for palm oil for domestic as well as international use. In addition, the demand for wood is also growing for biofuel energy (UN Energy 2007) and timber supply (Hetsch 2009;Penna 2010 (2015) Oil palm plantation IOPPs as mapped by Gaveau et al. (2016); does not show smallholder oil palm plantation areas Timber plantation Pulpwood plantations as mapped by Gaveau et al. (2016) Water bodies Rivers, lakes or any forms of water in the study area concession map depicting the potential areas of timber plantation expansion in both provinces allowed timber plantation expansion to be considered in this scenario. Hence, favorable market conditions (i.e. increasing demand) and a profitable investment return drive a rapid expansion of both oil palm and timber plantations to all concession areas not planted before 2016 by 2035 in this scenario.
The concession maps for oil palm and timber were extracted for the study areas from the concession maps of Kalimantan compiled by the WRI based on data from the Ministry of Forestry prior to 2010 (Gingold et al. 2012). Overlaying the concession maps of oil palm and timber onto the study area land-use maps in 2016 shows the areas for future expansions of oil palm and timber plantations under this scenario.

Conservation scenario
The conservation scenario considers stringent criteria to ensure environmental integrity and sustainability. Strict reinforcement and monitoring to avoid the clearing of intact forest and peatland for oil palm and timber plantations ensures conformation to the forest moratorium on primary forest and peatland. Further, logged forest and regrowth forest are also excluded from clearing, abolishing the deforestation of logged forest, which conforms to the 'zero deforestation' criterion set out to ensure sustainable palm oil production. In addition, this scenario also adheres to provincial spatial planning by restricting future expansion of oil palm and timber in the non-forest estate, i.e. the Areal Penggunaan Lain (APL) land class. The APL land class delineates the non-forest land and is legally available for other uses including agricultural land and settlement (Rosenbarger et al. 2013). The exclusion of all land classes except APL guarantees that future expansions of oil palm and timber only occur in the non-forest estate. This scenario will protect HCV and high carbon stock (HCS) areas due to the protection of the intact forest, logged forest, regrowth forest, peatland and restricts future expansion only on the APL land class.
The exclusion areas (intact forest, logged forest, regrowth forest, peatland and land classes other than APL) on the spatial planning map were combined to delineate the areas of exclusion for oil palm and timber expansions under this scenario. The exclusion or constraint map was generated by integrating areas for: (i) intact forest, logged forest and regrowth forest on the 2016 land-use map; (ii) peatland; and (iii) the non-APL areas on the spatial planning map. The 2014 peatland map was developed and validated by Wahyunto et al. (2014). The peatland map classifies the peatland depth into four classes: 50-100 cm, 101-200 cm, 201-400 cm and above 400 cm. Before integrating the peatland map into the constraint map, all peat Table 3. Key features of the future oil palm expansion scenarios.

Scenarios Key features
Business as usual • This scenario assumes that oil palm and timber plantation expansions occur on the respective concession areas as depicted on the concession area maps for 2035. • The high demand for palm oil for domestic use as well as from international markets is a main driver for oil palm expansion. • The growing demand for wood for biofuel energy and timber supply leads to expansion of timber plantations on concession areas.

Conservation
• No oil palm expansion occurs on forest cover (intact forest, logged forest and regrowth forest) and peatlands to conform to the forest moratorium and zero deforestation commitment. • Oil palm expansions are confined to the non-forest estate (APL: Areal Penggunaan Lain). • Oil palm expansions mainly occur on agricultural areas, regrowth forest, scrubland and other land. • High conservation value (HCV) and high carbon stock (HCS) areas are protected.

Sustainable intensification
• This scenario considers oil palm expansion on limited areas of the potential suitable land as mapped by WRI and SEKALA Project (Gingold et al. 2012). • Oil palm plantations are established by using improved varieties of cultivars and intensively managed to enhance palm oil productivity to 5.1 tons crude palm oil per hectare per year. • The expansion of timber plantations is not taken into account in this scenario because a suitability map for timber plantations was not available.
| 9 depth classes were merged to exclude plantation on all peatland with a depth of more than 50 cm. The non-APL areas were extracted from the spatial planning maps for Central Kalimantan and West Kalimantan obtained from CIFOR. To incorporate a conservation perspective into this scenario, the plantation intensity is assumed to be one-quarter of the plantation rate between 2000 and 2016 in the respective province.

Sustainable intensification scenario
The sustainable intensification scenario considers that future expansions will occur within the land suitable for oil palm plantations and applies yield enhancement tools and technology. It uses the existing suitability map to allocate the future expansion of oil palm in the study areas. The expansion of oil palm plantations is only considered on non-forest lands including scrubland, agricultural land and other land, which are characterized by low carbon stock, i.e. with less than 35 tons of carbon per hectare and low biodiversity value (Gingold et al. 2012). However, the expansion of timber plantations is not taken into account in this scenario because a suitability map for timber plantations was not available.
The suitability maps for oil palm are based on various criteria for IOPP for Indonesian Kalimantan. Mantel et al. (2007) used relevant biophysical criteria, constraints and management needs to determine suitability ratings for oil palm plantation in Kalimantan. The suitability map developed by WRI and SEKALA uses 21 indicators of sustainable palm oil production set out by the Roundtable on Sustainable Palm (RSPO) standard (Gingold et al. 2012). Since the latter oil palm suitability map is more comprehensive and complies with the criteria for sustainable oil palm production, it was used to predict future oil palm expansion. All natural forests, including primary forests, secondary forests and peatland, are protected in the study area.
The suitability map classifies approximately 3.2 Mha or about 21% of the total area in Central Kalimantan and 5.2 Mha or about 35.3% of total land in West Kalimantan as potentially suitable for sustainable oil palm plantations. However, some suitable areas are located on the forest cover and peatland areas. This intersection of suitable land with forest and peatland can be attributed to the difference in forest cover and peatland datasets used in the suitability mapping. To avoid losing these areas, a constraint map encompassing forest cover and peatland was combined with the land-use map to generate a scenario map that avoided clearing forest cover and peatland.
This scenario conservatively assumes that future oil palm plantations (2017-2035) increase at similar historical expansion rates between 2000 and 2016. Hence, an average annual expansion of 76,000 ha and 71,000 ha was applied to Central and West Kalimantan, respectively. Since oil palm plantations are established on suitable lands by using improved varieties of cultivars and are intensively managed to significantly enhance the oil palm productivity, this scenario assumes an average yield of 5.1 tCPO ha −1 yr −1 on par with the average yield under the RSPO-certified plantation (WWF 2012).
The suitability map for oil palm represents the potential areas for oil palm expansion in future. The proximity to existing road networks and mills may have a role in determining the locations of future expansion of oil palm plantations. With a massive road network and high density of mills across the study areas, all suitable areas have an equal chance of being used for oil palm plantation. Hence, the subset feature in Geostatistical Analyst of ArcGIS 10.2 was used to randomly identify an equivalent area for expansion between 2017 and 2035 based on a similar annual rate of oil palm expansion between 2000 and 2016.

Mapping, quantifying and valuating of ecosystem services under future land-use and landcover scenarios 2.5.1 Key ecosystem services
This study focused on the five most relevant and important ecosystem services from the oil palm landscape based on stakeholder consultation, literature reviews (Bhagabati et al. 2014;Sumarga and Hein 2014;Vijay et al. 2016) and expert knowledge. These ecosystem services include: i) Climate regulation service: Carbon storage; ii) Habitat quality; iii) Water yield; iv) Provisioning Service: Palm oil production; and v) Provisioning Service: Timber production.

Climate regulation service: Carbon storage
The land-use and land-cover changes (LULC) can represent a significant source of greenhouse gas (GHG) emissions in Indonesia. In 2005, about 63% of Indonesia's emissions were attributed to deforestation, particularly through peatland and forest fires (Government of Indonesia 2015). The expansion of oil palm plantations is considered a major driver of forest conversion and the resulting significant emissions. Oil palm development is predicted to contribute over a quarter (i.e. 26%) of net carbon emissions in West Kalimantan under the business-as-usual scenario by 2010 (Kimberly et al. 2012). Hence, carbon storage was selected as a key ecosystem service for this study. The change in carbon stock was analyzed to assess the impacts on climate regulation and understand trade-offs or synergies with other key ecosystem services under a different trajectory of future land-use scenarios. The InVEST Carbon Storage and Sequestration model was used for mapping, quantifying and evaluating carbon stored and sequestered under a LULC scenario. It applied carbon stock in the above-and belowground biomass, soil and dead organic matter to the current and future LULC maps (Equation 1). The model estimates the economic value of this ecosystem service to society by applying the carbon price and the discount rate, i.e. annual rate of devaluation of money. The model outputs include biophysical model outputs, comprising the carbon stock maps for the current and future LULC scenarios, and the valuation model output, comprising an economic valuation map and a sequestration map for future LULC (Sharp et al. 2016).
Carbon stock data were mainly sourced from Sumarga and Hein (2014), who compiled the carbon stock data from various sources and applied it to Central Kalimantan (Table 4). The carbon stock in peatland was obtained from Suwarna et al. (2012). They estimated carbon stock in the soil and vegetation of tropical peat forest in Indonesia. A global average carbon price of US$18.7 per tC was applied an average price of US$5.1 per tCO2-eq for the carbon from forestry and land-use sectors in the voluntary markets in 2017 (Hamrick and Gallant 2017). A default value of 7% was used to estimate net present value (NPV).

Habitat quality
The direct measurement of biodiversity (in terms of genes, species or ecosystems and their abundance and frequency) is beyond the scope of this study because measurements are complex (e.g. Baral et al. 2014), and require significant resources and time to collect. Instead, habitat quality is used as a proxy measure for biodiversity and assessed in a landscape by employing threats of different land-use classes and their relative vulnerability, accessibility and proximity to the identified threats (Arunyawat and Shreatha 2016; Sharp et al. 2016). This habitatbased approach is relatively simple and requires minimal data inputs for mapping biodiversity conservation status in a landscape. Hence, the InVEST Habitat Quality Model was used by applying threats from agricultural development and road networks. The data inputs require current and future LULC maps, sources of threat, accessibility to threat, sensitivity of LULC classes to each threat and a half-saturation value. The half-saturation constant represents a landscape degradation value. This parameter was set to the default value of 0.5 (Sharp et al. 2016). Equation 2 calculates the habitat quality of a grid cell x in land-use class j.
Several studies have acknowledged the significance of water-related ecosystem services and investigated the impacts on water yield and water quality due to land-use change (e.g. Schmalz et al. 2015;Ardaban et al. 2016Ardaban et al. , 2017Arunyawat and Shrestha 2016). During our field trip, many local people highlighted the change in water yield following land-use change and expressed their interest in better understanding the impact of oil palm plantations on water yield. Water yield is considered to be equal to net water run-off in given environmental and climatic conditions. Given the simplicity of the model and its low data requirement, InVEST Water Yield Model was chosen to assess the impact on water yield under future land-use scenarios. This model applies the Budyko curve and annual average precipitation to estimate water yield for a defined LULC class as given by Equation 3 (Sharp et al. 2016).
is the annual water yield for the grid cell x in a landscape.
represents the annual actual evapotranspiration for the grid cell x and is the annual precipitation on the grid cell x.
The following input datasets were obtained from various sources: • Root restricting layer depth: A GIS raster dataset for root restricting layer depth was derived from WRI's soil depth map for Kalimantan. • Precipitation map: This map was extracted from WRI's average annual precipitation data set for Kalimantan. • Plant available water content: A GIS raster dataset for this input was developed by using the soil profile in FAO's soil map (Fischer et al. 2008) and the corresponding value for the available water fraction was obtained from Soil-Plant-Atmosphere-Water (SPAW) at https:// hrsl.ba.ars.usda.gov/SPAW/Index.htm. • Average annual reference evapotranspiration: This was obtained by multiplying Pan Evapotranspiration data by 0.7 based on Allen et al. (1998). The Pan Evapotranspiration data set was sourced from http://www.cgiar-csi.org/ data/global-aridity-and-pet-database. • Evapotranspiration coefficient (KC): This is a factor for potential evapotranspiration for each plant type and was derived based on the sample evapotranspiration coefficient for different vegetation types in Sharp et al. (2016). • Z parameter: The model default value of 0.5 was applied.

Provisioning Service: Palm oil production
An assessment of palm oil yield under future land-use scenarios, in conjunction with impacts on other ecosystem services, provides a holistic understanding of trade-offs or synergies between the ecosystem services. Despite the focus of this study being on a spatially explicit assessment of ecosystem services, the lack of discrete spatial data on oil palm plantations (for example, plantation location, age, typologies (industrial or smallholder), management regime and productivity) prevented any spatial analysis of palm oil production. Instead, a simple quantitative approach was applied using average palm oil productivity in oil palm plantations and plantation area to estimate total palm oil production under these scenarios as described by Equation 4.
is total value of palm oil production in year t (t = 1,2,3…T) in US$.
represents the palm oil yield in tCPO ha −1 yr −1 .
represents the area of existing oil palm plantation in hectares and is the new oil palm plantation at 5 years and over s (s = 5,6,8 ……18 years). The plantations reach 18 years in 2035.
is the price of palm oil per tCPO in US$.
Many studies reported great variability in palm oil productivity across regions and typologies (industrial or smallholder) (Shiel et al. 2009;Sumarga and Hein 2014;Fargeon 2015;Woittiez et al. 2017). With uncertainty and variation in the yield data, this study applied an average yield of 3.8 tCPO ha −1 yr −1 (Woittiez et al. 2017). However, the sustainable intensification scenario uses an average yield of 5 tCPO ha −1 yr −1 due to the productivity enhancement.
A recent price of US$665 per tCPO from the Malaysian Palm Oil Council (MPOC) 4 was used to evaluate the oil palm asset price. New 4 http://www.mpoc.org.my/Daily_Palm_Oil_Prices.aspx plantations are assumed to expand at the annual rate determined for the respective future landuse scenario and are considered to produce palm oil after 5 years as suggested by Sumarga and Hein (2014).

Provisioning Service: Timber production
Timber plantation is also a driver for land-use change in the study areas and is likely to grow in the concession areas under the business-asusual and conservation scenarios. Hence, timber production is a provisioning ecosystem service from the timber plantation landscape. The total value of timber production is estimated by multiplying the area of timber plantation, the allowable cut per hectare and the timber price as given by Equation 5.

Total economic valuation of ecosystem services
The economic valuation of ecosystem services has been popularized to appreciate the significance of the natural ecosystem by expressing the values in monetary terms or a dollar value (Costanza et al. 1997;de Groot et al. 2012;Kubiszewski et al. 2013;Anderson et al. 2017 and used for economic valuation of ecosystem services (Kubiszewski et al. 2013;Costanza et al. 2014). The benefit transfer method is the most widely used in the emerging field of ecosystem services valuation and is recommended for applying the economic values measured in one location to a similar location (Costanza et al. 1997) 2012) and Costanza et al. (2014), many studies have used biome-level values to demonstrate the significance of natural ecosystems to human well-being and to inform decision making on natural resource management and investment (e.g. Kubiszewski et al. 2013;Kundu et al. 2016;Anderson et al. 2017;Tolessa et al. 2017). These studies justified the application of biome-level data through the lack of country-level, regional or local data and did not account for geographical variability and uncertainty in valuation. Although these values are not the market value or a price for ecosystem services, the economic values help all stakeholders to understand and appreciate the relative values of the ecosystem services. They also enable comparison of the change in these values over time and space in conjunction with the drivers of change.
There is little data available for the economic valuation of ecosystem services for Indonesia. Despite some data on economic valuation of ecosystem services from protected areas ( Based on the current vegetation structure and disturbances, the forest covers were categorized into intact forest, logged forest and regrowth forest. Applying the mean values for ecosystem services of the tropical forest biome to these forest covers ignores the difference in their capacity to provide ecosystem services. This valuation leads to overestimation or underestimation of ecosystem services depending upon the proportion of these forest covers. In highly disturbed areas, the mean value approach overestimates the total value, whereas in less disturbed areas, the total value is underestimated. To differentiate the landscape's capacity to deliver ecosystem services, unique landscape indexes are increasingly used to value ecosystem services (Burkhard et al. 2009Sohel et al. 2015). This approach could be used to identify and assign relative indexes for different conditions of a landuse class. By considering the valuation of ecosystem services in the original database, the minimum and maximum values encompass the differences in the biome conditions and their capacities in terms of delivery of ecosystem services.  Jackson et al. 2005) and reducing pressure on natural forests (e.g. Warman 2014). To conform to these findings and to account for the ecosystem services from plantations, the minimum values for ecosystem services for the tropical forest biome were applied to oil palm and timber plantations. In addition, the market values of palm oil and timber yields from these plantations were added to the provisioning services for TEV.
Scrublands were previously forested areas (oldgrowth and selectively logged forests) and became degraded due to deforestation, drought and recurrent fires. To value ecosystem services from scrublands, these areas were considered equivalent to a woodland biome in a poor state and were assigned the minimum value for this biome. However, applying the minimum economic value of US$1274 for the woodland-to-scrubland area remarkably overestimates the value. Scrublands are in a degraded state and experience high disturbance. In the absence of an economic valuation of habitat services from scrubland in the study area or the region, this study applied a minimum value of habitat services from tropical forest. Since agricultural land was not included in the de Groot et al. (2012) biome list, this study used the economic value of ecosystem services from the 'crop' category compiled by Costanza et al. is the TEV for scenario, s in year, t (t = 1, 2, 3, 4…T). r is the discount rate.

Current land use in the study areas
The land-use map (2016) represents the nine landuse classes in both provinces (Figures 3 and 4). Table 6 presents the spatial extent of each land-use class in both provinces. Oil palm plantation covers about one-tenth of the area of both provinces. Oil palm mainly dominates in the southwest region in regencies Kotawaringin Barat, Seruyan and Kotawaringin Timur in Central Kalimantan. In West Kalimantan, plantation mainly occurs in the central part in regencies Landak, Sanggau, Sekadau, Sintang and in the south in Kab Ketapang regency. ITP was mapped onto approximately 112,000 ha in Central Kalimantan and 67,000 ha in West Kalimantan.

Oil palm expansion between 2000 and 2016
Central Kalimantan and West Kalimantan share over 60% of the total IOPP (5.06 Mha) of Indonesian Kalimantan. One-third of the IOPP has been developed on the forest cover, shrub and nonforest in Kalimantan. Figure 5       scenarios by 2035. The major features of these scenarios are presented in Table 7 for Central and West Kalimantan.

Conservation scenario (CON)
Applying the assumptions of the conservation scenario, oil palm plantations slightly increase by 216,000 ha, whereas timber plantation areas expand by about 17,000 ha by 2035 in Central Kalimantan. Altogether over 13.7 Mha are excluded from the future expansions of oil palm and timber plantations under this scenario, including forest cover (0.98 Mha), peatland (2.66 Mha) and areas other than APL land in the spatial planning map (12.8 Mha). However, the existing oil palm plantations (1.6 Mha) and timber plantations (109,000 ha) in the constrained area before 2017 are continually used for palm oil and timber productions under this scenario (Figure 8b).
In West Kalimantan, excluding the forest cover and peatland, and restricting plantations to the designated land for agriculture and other uses (APL) on the spatial planning map, oil palm will expand by an additional 0.54 to 2.1 Mha. Timber plantations will also increase by about 74,000 ha to 141,022 ha (Figure 9b).

Sustainable intensification scenario (SUS-INT)
In

Carbon storage and sequestration
By applying the secondary source of carbon stock data for land-use classes with and without peatland, the model estimates that 6516 million tons of carbon (million tC) are stored in the current land use in Central Kalimantan. Carbon storage is reduced by about 9% (591 million tC) under the business-as-usual scenario, whereas conservation and sustainable intensification scenarios result in an increase of the carbon stock by 10 million tC and 39 million tC, respectively (Table 8).

Transitions to Oil Palm Plantations in Thousand Hectares
Business as usual Conservation Sustainable Intensi cation Obviously, the reduction of carbon stock is expected in business as usual due to the conversion of HCS areas, i.e. old-growth and regrowth forests. However, the carbon stock increases in conservation and sustainable intensification due to avoiding clearance of the HCS areas, including forests and peatland, and most of the expansion occurring on agricultural land and scrubland.
In West Kalimantan, the current land use stores about 3925 million tC, which is approximately 60% of the carbon stock in Central Kalimantan. Aggressive expansion of oil palm and timber plantations into HCS areas results in about 20% loss of carbon stock (779 million tC) in the business-as-usual scenario. Carbon stock increases by 26 million tC in the conservation scenario to 3951 million tC, i.e. 1.6-times more than in Central Kalimantan due to the relatively high intensity of oil palm expansion on agricultural land and scrubland, i.e. low carbon stock areas in West Kalimantan. However, the carbon stock increases by the same amount, i.e. 39 million tC, as in Central Kalimantan under the sustainable intensification scenario to 3964 million tC in West Kalimantan (Table 9).
Regarding the NPV, the loss of carbon stock in the business-as-usual scenario results in a negative NPV of (−)US$6.4 billion in Central Kalimantan and (−)US$8.5 billion in West Kalimantan at a 7% discount rate without an increase in the carbon price of US$18.72 per tC. The NPV is proportional to the amount of carbon stock loss or gain in the future scenarios. Due to the increase in carbon stock in conservation and sustainable intensification scenarios, the NPVs are positive with the highest return of US$429 million in sustainable intensification in West Kalimantan (Tables 8 and 9).
The spatial distribution of carbon stock under the current land use and future scenarios is visualized in Figure 12a-d for Central Kalimantan and Figure 13a-d for West Kalimantan. Since the change in carbon stock is not prominent on the provincial maps, a small area is highlighted to focus on carbon stock change as a result of plantation expansions in these scenarios.

Habitat quality
The model estimated relative habitat quality values between 0 and 1 across the province under the future land-use scenarios. Habitat quality is expected to decline in future land-use scenarios due to expansions of oil palm plantations at the expense of forest cover. However, the conservation scenario presents an exception with almost no change in habitat quality in both provinces. This is attributed to the protection of old-growth and regrowth forests under this scenario. Under the businessas-usual scenario, the habitat quality declines significantly with an almost 12% reduction in the average habitat quality value from 0.572 in current land use to 0.503, in Central Kalimantan. Similarly, the habitat quality decreases by over 14%      from 0.418 to 0.358 in West Kalimantan. The old-growth and regrowth forests in the concession areas are cleared for expansion of oil palm and timber plantations, which entails a remarkable loss of habitat quality in both provinces. Besides, these expansions also open up the habitat frontiers in the adjacent areas of old-growth and regrowth forest by increasing the level of threat due to proximity to these areas causing a decline in habitat quality. The model predicts a very nominal loss of habitat quality, i.e. less than 1% in both provinces from the oil palm expansions under the sustainable intensification scenario. The change in habitat quality is less distinct on the provincial maps.
Hence, a small area is highlighted to demonstrate the impact on habitat quality as a result of oil palm expansion under three scenarios (Figures 14  and 15).
Under the conservation and sustainable intensification scenarios, the old-growth and regrowth forests are protected ensuring no loss of habitat quality. Hence, high habitat quality areas are located in the northeast regions of both provinces, which correspond to the areas of forest cover. Habitat quality tends to decline from a conversion of scrublands to plantation, while it slightly increases due to conversion of agricultural land to plantation. Habitat quality is considered to deteriorate on plantations (oil palm and timber) relative to scrubland. The conversion of scrubland to oil palm plantation evidently causes loss of habitats of native flora and fauna and may also negatively impact aquatic ecosystems due to water contamination. However, by employing some conservation strategies, the habitat quality might improve in plantations (oil palm). A small increase in the number of forest-dwelling species was reported in oil palm plantations after implementation of conservation efforts in these plantations (Koh 2008).

Water yield
The model predicts a decline in water yield under all future land-use scenarios relative to the current land use. In Central Kalimantan, the water yield will decrease by 0.62%, 0.14% and 0.13% under sustainable intensification, business-as-usual and conservation scenarios, respectively. In contrast, the business-as-usual scenario will experience the highest loss of water yield (1.7%), followed by the sustainable intensification scenario (0.54%) in West Kalimantan. The lowest decline in water yield is predicted for the conservation scenario in both provinces. Although, these reductions in water yields do not show any significant loss at the provincial level, the changes in water yield are evident at the local level, as illustrated in Figures 16 and 17. With the same bioclimatic and biophysical variables, the reduction in water yield is solely attributed to a change in current land use to plantation (oil palm or timber) in the future scenarios at the provincial level. The expansion of oil palm and timber plantations on agricultural land, scrublands and other lands increases the transpiration from these areas. Thus, the water yield is negatively impacted due to a decrease in water available as run-off. In both provinces, the highest percentage loss of water yield is supported by increased conversion of these land uses to plantations. For instance, in the sustainable intensification scenario in Central Kalimantan, above 90% or over 1.0 Mha of oil palm expansions occur on agricultural land (57%), scrublands (33%) and other land (4%), which corresponds to the highest percentage loss in water yield. Similarly, about 2.7 Mha of these lands are converted to oil palm and timber plantations resulting in the highest loss in water yield in the business-as-usual scenario in West Kalimantan.

Palm oil production
In Central Kalimantan, the business-as-usual scenario produces 8.45 million tons of crude palm oil per year (Mt CPO yr −1 ) from an average area of 2.22 Mha based on the average yield of 3.8 tCPO per ha per year between 2017 and 2035. Assuming an enhancement of productivity to 5.1 tCPO per ha per year equivalent to the average yield under the Roundtable for Sustainable Palm Oil (RSPO), the sustainable intensification scenario supplies almost the same amount of palm oil as the business-as-usual scenario, i.e. 8.37 Mt CPO yr −1 from 2.05 Mha. In the conservation scenario, the existing oil palm plantation dominates the palm oil production with an annual yield of 6.49 Mt CPO from 1.71 Mha (Figure 18). Based on the current price of US$665 per t CPO from the MPOC, 6 the gross revenue is estimated to US$5.61 billion for business-as-usual, US$5.56 billion for sustainable intensification and US$4.31 billion for the conservation scenarios.   In contrast to Central Kalimantan, the businessas-usual scenario produces 16% more palm oil, i.e. 9.85 Mt CPO yr −1 , in West Kalimantan due to an average oil palm area of 2.6 Mha between 2017 and 2035. This yield is about 20% more than in the sustainable intensification scenario (8.08 Mt CPO yr −1 ). While the existing oil palm plantation accounts for about 6.0 Mt CPO yr −1 , the conservation scenario yields a slightly higher annual yield of about 6.63 Mt CPO yr −1 (Figure 19). Based on the current price of palm oil US$665 per t CPO, the gross revenue per year is estimated to about US$6.55 billion for business-as-usual, US$5.37 billion for sustainable | 33 intensification and US$4.41 billion for the conservation scenarios.

Timber production
Based on the allowable cut of 0.86 m 3 per ha per year (Suwarno et al. 2016), the total timber production increases to about 0.42 million m 3 per year (m 3 yr −1 ) in the business-as-usual scenario from about 0.10 million m 3 yr −1 in the current land use in Central Kalimantan. Due to a restriction on expansion of timber plantation on forest cover, timber production decreases to about 0.11 million m 3 yr −1 and 0.03 million m 3 yr −1 in conservation and sustainable intensification scenarios, respectively. In West Kalimantan, timber production is projected to increase by many fold to 1.62 million m 3 yr −1 in business as usual from 0.06 million m 3 yr −1 in the current land use. In contrast to Central Kalimantan, timber plantation increases by 74,000 ha in the conservation scenario resulting in a rise in timber production to 0.12 million m 3 yr −1 in West Kalimantan. However, sustainable intensification produces the same quantity of timber, i.e. 0.06 million m 3 yr −1 as in the current land use because the timber plantation area does not change in this scenario relative to current land use ( Figure 20).    Figure 21).

Total economic valuation of ecosystem services
The NPVs of the TEV of the ecosystem services were estimated for the current land use and the three future land-use scenarios at the discount rates of 3.5%, 5%, 10% and 15% (Table 11). The negative sign (−) indicates a total loss of ecosystem services under the future land-use scenarios relative to current land use. The NPVs of the ecosystem services decrease with an increase in the discount rate from 3.5% to 15%. At the lowest discount rate of 3.5%, business as usual results in an equivalent to ( The reductions in these TEV are directly attributed to the loss of old-growth and regrowth forests in the business-as-usual scenario. The conservation scenario offers the highest TEV among the three scenarios because of the protection of all forest covers from the expansion of oil palm and timber plantations.

Trade-offs or synergy between ecosystem services under future landuse scenarios
The model results in Section 3 show a synergy between increased palm oil yield (due to an extensive expansion of oil palm plantations) and water yield (reducing). The reduction in water yield is considered a positive ecosystem service for reducing or avoiding the negative consequences of increased water flows, such as flooding, landslides and loss of property and lives (Arunyawat and Shrestha 2016). This synergy occurs under all future land-use scenarios due to the conversion of agricultural land, scrublands and other land uses into oil palm plantations. Water yield is significantly reduced through the increased rate of transpiration compared with land uses before conversion. The expansions of oil palm on oldgrowth or regrowth forests under the business-asusual scenario result in trade-offs between palm oil yield and carbon stock as well as habitat quality. In the business-as-usual scenario, the increased palm oil yield results in a significant loss of the carbon stock and habitat quality due to the conversion of HCS and habitat quality forest covers to oil palm plantation. Table 12 summarizes the relative availability of four ecosystem services and the TEV of the land-use classes relative to intact forest. Tradeoffs or synergies are evident due to the change in the ecosystem services attributed to the landuse change under a future scenario. Under the business-as-usual scenario, the expansion of oil palm plantations across intact forest results in trade-offs of all four ecosystem services. The TEV of ecosystem services show a reduction in value of 80% due to the conversion of intact forest to oil palm plantation. The development of oil palm plantation on logged forest or regrowth forest also results in trade-offs as the TEV of ecosystem services reduce by over 50%. Oil palm expansions onto agricultural land increases the value of provisioning ecosystem services by about 75% due to the revenue from palm oil production, whereas regulating and cultural services are significantly reduced. After clearing intact forest, timber plantations retain about 9% of the ecosystem services from the previous land use (Table 12). Regarding water yield, several hydrological studies suggest that forest vegetation with high evapotranspiration rates reduces water yield, while agricultural crops or urbanization increases surface run-off or overland flow (Hamilton 2008;Suryatmojo et al. 2013). The InVEST Water Yield Model demonstrated a correlation between water yield (reducing) and expansion of oil palm plantation on agricultural land or scrublands in both study areas. Water yield increases with loss of vegetation cover, whereas it decreases with an increased vegetation cover. This result is consistent with water yield predictions for the agriculturedominated sub-watershed in Northern Thailand using the InVEST Water Yield Model (Arunyawat and Shreatha 2016). Since this study did not validate the model outputs, the results are an indicative representation of water yield in the study area due to expansion of oil palm plantation. Further, the model might not have fully captured the impact on water yield due to the hydrology of peatland in the study area, which needs to be taken into account for an accurate understanding of the relationship between land-use change and water yield. To improve confidence and the reliability of outputs for its application in decision making (e.g. Guswa et al. 2014), water yield prediction needs to be performed at a district or sub-watershed level using accurate data and subsequent model calibration and sensitivity analysis (e.g. Sanchez-Canales et al. 2012;Hamel and Guswa 2016).

Trade-offs or synergies and economic valuation of ecosystem services
Several studies have explored the trade-offs or synergies between multiple ecosystem services under specific land-use and management regimes or land-use change scenarios (e.g. Power 2010; Klasen et al. 2016;Sumarga and Hein 2016). The management regime or land-use change can cause trade-offs or synergies due to either an enhancement or a reduction of multiple ecosystem services. Such knowledge is needed to understand the realities of whether the specific ecosystem service of interest is enhanced or lost from the land-use change or management intervention  (Dislich et al. 2016). This study highlighted that some of the trade-offs are irreversible with much broader impacts (e.g. climate regulation, habitat loss) and showed that the intensity of trade-offs increases if the oil palm plantation replaces forest cover or peatlands.
The trade-offs for oil palm plantations are evident between the provisioning service from palm oil production and losses of regulating, cultural and habitat functions. Trade-offs or synergies between ecosystem services are evaluated against the previous land-use or the land management regime or between the current and future land use. Oil palm can offer climate regulation services where the previous land use has a lower capacity for carbon sequestration and storage. For example, Germer and Sauerborn (2008) observed a synergy for climate regulation from the conversion of tropical grassland to oil palm plantation with net carbon sequestration in biomass and soil. This study confirmed the trade-offs or synergies observed in the above studies by demonstrating (i) trade-offs in four ecosystem services due to oil palm expansion on old-growth forest and regrowth forest under the business-as-usual scenario and (ii) synergies for climate regulation from oil palm expansion on degraded or agricultural land by enhancing carbon sequestration and storage in green biomass and soil in all future land-use scenarios.
This study focused on the entire province level and generated spatial maps for five key ecosystem services. These spatial maps showed the change in ecosystem services and their trade-offs or synergies under three future land-use scenarios. Another study (Klasen et al. 2016) demonstrated that trade-offs tend to vary at different spatial scales, and that a study focusing on local (i.e. households or village) or regional (i.e. district) scales can be used to undertake a detailed analysis of trade-offs or synergies between ecosystem services due to oil palm expansion. Such a study can assist land-use planners or decision makers in making betterinformed land-use decisions.
The global value of ecosystem services at the biome level might not provide accurate values for ecosystem services in the study area. However, global values have been used to understand the economic values of ecosystem services at the global level (e.g. Costanza et al. 1997Costanza et al. , 2014de Groot et al. 2012) and at landscape (Kundu et al. 2016;Tolessa et al. 2017) and country (Kubiszewski et al. 2013;Anderson et al. 2017) levels. By applying global ecosystem services values to the representative biome, a preliminary value of ecosystem services can be generated in any geographic location and thus help to develop an understanding of the contribution of these ecosystem services to humans. This approach enabled an assessment and evaluation of ecosystem services TEV in the study area under three future land-use scenarios and demonstrated a significant loss of ecosystem services under the business-asusual scenario. This supports the perspective that intensive expansion of oil palm, as in the businessas-usual scenario, could seriously harm the province's capacity to supply ecosystem services. The high associated cost signifies the gravity of the loss of these services. However, valuation of these ecosystem services based on local-or provinciallevel data would increase the accuracy and confidence in the valuation of ecosystem services in the study area.

Conclusion
The analyses of future land-use scenarios have shown significant impacts on the supply of ecosystem services based on the intensity of oil palm expansions under these scenarios. The business-as-usual scenario results show detrimental impacts on ecosystem services due to an intensive expansion of oil palm plantation including across areas of old-growth forest and regrowth forest. Assuming the lowest intensity of oil palm expansion, the conservation scenario enhances carbon stock and maintains a stable habitat quality relative to the current land use (2016). The sustainable intensification scenario with oil palm expansions only on suitable areas and enhancement of yield generates a positive impact on carbon stock and water yield, whereas habitat quality deteriorates slightly in the study areas.
Based on the TEV of ecosystem services, the conservation scenario generates the highest values of ecosystem services under the three future land-use scenarios in both study areas. However, the sustainable intensification scenario offers a compromise solution for future expansion of oil palm by ensuring a supply of ecosystem services comparable to the conservation scenario, and without significantly affecting palm oil yield. This scenario would help meet the future demand for palm oil through area expansion of oil palm and yield enhancement. However, a potential issue of food security might emerge because of the extensive conversion of agricultural land to oil palm plantations under the sustainable intensification scenario. Therefore, future oil palm expansion should be approached cautiously to achieve sustainable land use and balance human and environmental needs.
The TEV of ecosystem services must not be taken as a market value or for payment of ecosystem services. These values should be understood in relative terms between these scenarios and used to enhance awareness regarding the impacts of future expansion of oil palm plantations on key ecosystem services. A detailed study at a local level (household or village) that evaluates the economic values of key ecosystem services and their trade-offs or synergies is recommended for an accurate valuation of ecosystem services to better understand the impacts of oil palm expansion on the local community and the environment.