Maxent model application for conflict mitigation of Bornean Orangutan (Pongo pygmaeus wurmbii) in oil palm plantation

The Bornean orangutan (Pongo pygmaeus wurmbii) is a flagship species and is protected by the Government of Indonesia and the global community. Orangutan habitat is in a tropical rainforest ecosystem. Orangutan exists in conservation area and development area (outside conservation). Development activities, such as oil palm plantations, cause fragmentation of orangutan habitat and triggers conflicts with humans. This study aims to map the suitability class of orangutan habitat and areas with potential conflicts around oil palm plantations and determine mitigation measures. The study site is located in Ketapang, West Kalimantan, Indonesia, consisting of forested areas and areas for development. This study uses the Maximum Entrophy (MaxEnt) model, which can simulate the probability of orangutans’ presence based on influencing environmental variables. Primary data is from 2019 – 2020 survey, and secondary data is collected from various sources. The study area is divided into three classes of orangutan habitat suitability: high, medium, and low. The dominant environmental variables in determining suitability classes are land use and land cover. The high and medium habitat suitability classes are mostly located in forested areas. The suitable habitat for orangutans in the study area is fragmented by community cultivation areas and oil palm plantations. Oil palm plantations are the variable with the highest contribution to the mapping of potential conflict vulnerabilities. The mitigation zone in the study area is divided into four zones. Zones 1, 2, and 3 can be linked by developing an orangutan corridor to connect with the nearby orangutan habitat in Gunung Palung National Park. Zone-1 is a Gunung Tarak Protected Forest, a buffer forest area for the Gunung Palung National Park. Zone-1 is the core orangutan conservation zone in the study area. This study suggests that oil palm companies in zones 1, 2, and 3 prioritize mitigation measures with passive preventive actions, while those in zone 4 can take active preventive mitigation actions.


Introduction
One of the protected flagship species in Indonesia is the Bornean Orangutan (Pongo pygmaeus wurmbii Linn).This large primate is classified as a rare animal and is only found in Southeast Asia.Its conservation status according to the IUCN Red List (2021) is Critically Endangered, has been registered in CITES APPENDIX 1 (2020), and is protected by the Government of Indonesia (PP No. 7 of 1990).Orangutan habitat is not limited to conservation areas, as set by the Government of Indonesia, but also in areas reserved for agriculture [1].The tropical rain forest that is the habitat of orangutans, continues to shrink, degrade and become fragmented.This situation is the main trigger of human-orangutan conflict [2,3].The increasing human needs for limited land and natural resources have created conflict and competition between humans and orangutans [4,5].This conflict creates a feeling of insecurity for both of them [6,7].This impacted the increasing pressure and threats to the existence and survival of orangutans.There is also concern about damage to their crops from the human side [8][9][10].Some of the common causes of conflict in Indonesia include (i) conversion of forests to agricultural areas, mining, and settlements, (ii) forest and land fires, (iii) logging, both legal and illegal, (iv) labour, and (v) infrastructure development.
Orangutans in Kalimantan are often reported to be in agricultural areas adjacent to conservation areas and still have some forested areas [11,12].Orangutans can be trapped in the agricultural areas, especially plantations, and lose access to their natural habitat [13,14].These areas were previously the natural habitats of orangutans that are disturbed due to agricultural activities, shrinking in size, degraded, fragmented, and can lead to extinction [15,16].Agricultural activities that get the most spotlight for degrading orangutan habitats are oil palm plantations.The need for extensive land causes orangutans to share space.This situation creates conflicts that cause pressure and threats to the sustainability of the orangutan species and habitat.From the plantation side, the presence of orangutans is also considered anuisance, pest, and causing legal, and credibility problems.
Orangutan conflict mitigation by oil palm plantations should ideally carried out from the planning to plantation operations.Planning should consider spatial allocation for conservation and the presence of orangutans.Some stakeholders have been implementing mitigation measures called Integrated Conservation Landuse Planning (ICLUP), before opening the plantation [17].Companies can use ICLUP to minimize conflicts with orangutans.However, some plantations only know the existence of orangutans once they are operational, and conflicts cannot be avoided.Therefore, plantations must adapt to the presence of orangutans in the plantation area and its surroundings [18,19].
Conflict mitigation and adaptation of plantations to the presence of orangutans can be carried out with a comprehensive understanding of their behavior towards habitat, food and water resources, roaming patterns and accessibility, and potential threats and disturbances.These are referred as the orangutans habitat suitability level.Knowledge of surrounding areas suitable for orangutan habitats and the potential threats and disturbances can be used by plantations and other stakeholders to plan mitigation and adaptation to conserve orangutans and their habitats.One of the methodologies for mapping orangutan habitats and threats using spatial model approach is the Maximum Entrophy (MaxEnt) model [20,21].
Orangutan conservation around oil palm plantations can provide benefits for plantations.On the ecological aspect, the presence of orangutans contributes to ecosystem balancing and the energy chain in the plantation and its surroundings.On the economic part, orangutan conservation can increase the value of plantation companies through sustainable product certification such as ISPO (Indonesia Sustainable Palm Oil), RSPO (Roundtable Sustainable Palm Oil), ISCC (International Sustainability and Carbon Certification), etc.All sustainability certification schemes require the conservation management of wildlife habitats, including orangutans.This certification scheme also guarantees that oil palm plantations make a real contribution to orangutan conservation.This study aims to (i) identify and map the orangutan habitat suitability level around existing oil palm plantations, to determine the potential existence and presence of orangutans, (ii) identify and map areas based on the vulnerability level that can trigger orangutan conflicts with humans, (iii) identify options for mitigating and adapting oil palm plantations to minimize conflicts with orangutans.

Study location
The study location is in district Matan Hilir Utara Nanga Tayap, Ketapang Regency, West Kalimantan Province, Indonesia.Based on the forest area map, the designated areas are agricultural area (APL), protected forest (HL), production forest (HP), and convertible production forest (HPK).In the APL area, 1243 (2023) 012022 IOP Publishing doi:10.1088/1755-1315/1243/1/0120223 there are oil palm plantations run by companies and communities.The oil palm plantation companies i.e.KAL, GMS, LSM, DAS, and LS.The HL area is Gunung Tarak protected forest, a buffer forest for Gunung Palung National Park to the north of the study area (Figure 1).

Materials and tools
The data used in this study are primary and secondary.The primary data is from the geographic distribution of the Bornean orangutan, from orangutan monitoring data conducted by the BGA group in 2019-2020.Data is collected using the field observation method and camera traps placement.Field observations were carried out using the direct encounter and indirect encounter methods, resulting in direct encounter data, nests, feces, food residues, and other traces that can indicate the presence of orangutans [22][23][24].Other data are the location of settlements around the study location, road network, rivers, and land conditions in and around the research location [15].For the installation of camera traps, purposive sampling was placed based on the field observations on tracks, nests, feces, and feeds [25,26,27].The total number of orangutan encounter points is 384 points.In addition to the encounters point, monitoring data also identified the types of plants that became a source of food.
The secondary data used consisted of Sentinel-2 satellite imagery acquired in September 2020, a 1:25000 scale map of the Indonesian landscape, a 1990 land system map of Indonesia, and a 1:50000 scale semi-detailed soil map (Table 1).Using ArcGIS software, these primary and secondary data were processed and analyzed as input variables from the MaxEnt model.

MaxEnt model
Mapping orangutan habitat and threats can be conducted using spatial models such as the Maximum Entrophy (MaxEnt) model.This model is a set of machine learning modeling algorithms in software that can be used to determine the suitability level of a species habitat [28].This model can also be usedto predict and determine the geographic distribution of a species [29,30].The environmental variables in this model are prepared with the ArcGis application so that all have georeferenced values.Suitable habitat is determined by the relationship between the presence of species as the dependent variable and environmental characteristics as the independent variable [31,32].
The Maxent model can be run with two types of data: training sample data and environmental variable maps.The training sample data is numerical data in *.csv format, consisting of information on the presence of orangutans and their location (latitude and longitude).The model used 50% of encounter data for developing the algorithm and the rest for validation.The environmental variable map is a thematic map sheet in ASCII format, the independent variable for MaxEnt model input.The pixel size used in this study is 100x100 meters.The environmental variables used are representations of habitat biophysical conditions and threat indicators (Table 1).Environmental variables numbers 1-6 are indications of the carrying capacity of the habitat, while numbers 7-10 indicate the presence of threats.Multicollinearity testing was carried out before environmental variables were used as input for MaxEnt modeling.This test aims to determine the correlation between the environmental variables used.Multicollinearity can produce overconfident and/or underestimate models caused by the influence of two or more mutually related variables, both positively or negatively [33].The threshold value that indicates a set of variables has multicollinearity if the correlation value is |0.75|, therefore one of the environmental variables must be removed.To determine the multicollinearity between variables used multivariate statistical analysis with Spatial Analyst Tools in ArcGIS.
The output of the MaxEnt model is the probability of the presence of orangutans in each pixel or a specific location.This probability value will determine suitable areas for orangutan habitats and sites that have risk of threats and cause conflict for orangutans.Determination of suitable habitats is also supported by primary data from surveys of orangutan food sources.This model has a limited number of training samples or encounter points that are not evenly distributed throughout all locations.Therefore, weighted extrapolation will be used in determining habitat suitability using the Percent contribution and Permutation importance values for each environmental variable resulting from the MaxEnt model.All analyses of the MaxEnt model used ArcGIS software.

Model performance
The MaxEnt model can predict the probability of an orangutan encounter or presence using ten environmental variables.The output of the MaxEnt model consists of a spatial prediction of the location of the presence of orangutans, AUC (Area Under Curve), the response curve of each variable, the percentage contribution of each environmental variable to the probability of presence, and the Jackknife test.A good model can be determined by a curve that maximizes sensitivity to positive fractional values.This can be measured by calculating the Under Curve Area (AUC) (Fourcade et al., 2014).The performance of the MaxEnt model in the study has an AUC value of 0.942.(Figure 2).This means that the prediction model for the presence of orangutans has explained most of the encounter locations.

Habitat suitability class
The biophysical aspects that determine the level of suitability of orangutan habitat and the proportion of its contribution in this model are (i) land use (42.7%), topography (28.6%), water sources (14.9%),ecosystem type (9.4 %) and land slope (4.3%).This level of conformity is divided into three classes, namely high, medium and low (Figure 3).The area with high suitability class is 27,083 ha (20.5%), medium is 29,118 ha (22.1%) and low is 75,633 ha (57.4%).For each variable, the suitability class is determined by the probability of the presence of orangutans at that location.The level of habitat suitability shows suitable locations as orangutan habitat only based on biophysical aspects, without considering the threat factors originating from human activities.Threat factors will be used to determine the probability of presence in a certain area that is prone to conflict.Some high and medium habitat class areas are located near and even within oil palm plantations and in bush and field areas.This habitat fragmentation can trigger conflict.However, information on the existence of these habitats is also important for mitigating conflicts, for example, by developing corridors connecting suitable habitats.The most suitable land use and cover as orangutan habitat is forest.However, not all forests in this area are in the high suitability class because other environmental variables influence them.Compared to other types of land use, the probability of forest being the habitat of orangutans is highest, and after that is shrubs and scrub (Figure 4).In these three types of land cover, there is natural vegetation in whichthere are types of orangutan food sources [34,35].The potential for orangutan food trees is also concentrated in forest-covered areas.Of the 198 plant species identified, there are 83 species of plants with the potential as food trees, 70 species of which are found in forest land cover classes, 45 species are found in thickets and 7 species in shrubs.Among the forage plants of the forest people were Artocarpus anisophyllus, Ficus annulata, Artocarpus elasticus, Ficus obscura, Piper aduncum, Mallotus paniculatus, Litsea sp., Durio sp., Eugenia sp., Knema sp.etc.The availability of water for orangutans is represented by the distance to water sources, rivers, and swamps/open water bodies.The probability of the presence of orangutans to the distance of water sources is below 0.5 (Figure 5).This indicates that orangutans depend not only on the rivers and swamps identified in this study but also on other sources.This can be demonstrated by the presence of orangutans in wetland ecosystem types such as peat swamp forest.Orangutans have the ability to adapt and choose water sources for their needs [36].

Figure 5. Probability of orangutan presence according to Euclidean distance from water source
In general, the ecosystem in the study area is tropical rainforest which is the natural habitat of orangutans [37].Included in tropical rain forests are lowland forests, swamps, mountains, heath, and peat [35,[38][39][40].All types of ecosystems resulted in the probability of the presence of orangutans above 0.62.The forest cover factor in each ecosystem is the dominant factor determining the probability of orangutans' presence.Of the ten ecosystems in the study area, shallow peat swamp forest (MDW) and deep peat (GBT) ecosystems have a higher probability of suitability than other ecosystems.MDW habitat suitability is higher than GBT.This is due to the higher diversity of forage plants in this ecosystem, as it is a transitional ecosystem to mineral soils (Figure 6).The mixed hill dipterocarp forest ecosystem on metamorphic rocks (HJA and TWI) on Mount Tarak also has a high probability of presence.
The altitude in the study area is suitable as orangutans habitat with a probability value of more than 0.6.Orangutans live in the lowlands and the highest density is found at an altitude of about 200-400 m above sea level [24,41].The slope factor shows variations in the probability level of suitable habitat for orangutans.However, the model results show the highest probability of orangutans presence is on a land slope of less than 15% (Figure 7 dan Figure 8).This information on topography and land slope also show the level of accessibility of orangutans [42,43].

Potential areas of conflict
Potential conflict areas are determined based on environmental variables such as public roads, plantation roads, company and community plantation roads, and settlements [44,45,46].The variable is a representation of the place where humans carry out activities.In addition to these four variables, the potential for conflict is also influenced by topography and slope, limiting humans' access and utilize land [47,48,49].The measure used to determine potential areas of conflict between orangutans and humans is the probability of the presence of orangutans at locations of human activity using the Euclidean distance measure.The probability of orangutan presence is influenced by habitat suitability, about 51.5%, and the variables that trigger conflict are around 48.5%.Habitat locations with high suitability classes around oil palm plantations cause oil palm plantation variables to make a significant contribution to the probability of the presence of orangutans.The contribution of each conflict trigger variable to the probability of the presence of orangutans starting from the highest is oil palm plantations 39.8%, topography 23.0%, company and community plantation roads 15.4%, settlements 12.7%, public roads 7.6 %, and a slope of 1.5%.The probability of the presence of orangutans in a location becomes the basis for determining potential conflict-prone areas (Figure 9).Although the presence of these orangutans in areas with high habitat suitability, getting closer to areas of human activity also increases the potential for conflict.

Figure 9. Areas with high levels of potential vulnerability to conflict between orangutans and humans
The response curve for the distance of oil palm plantations to the probability of the presence of oil palm shows that at a distance of 0-1800 meters shows a high probability, above 0.5.The further away from oil palm plantations, the probability of the presence of orangutans that can cause conflict is smaller.The safe distance of oil palm plantations from the presence of orangutans is 5600 meters from the orangutan habitat (Figure 10).This situation is also reinforced by the response curve of the company's and the community's plantation road distances which show that at a Euclidean distance of 2700 -15000 m, the probability of the presence of orangutans is above 0.7.Meanwhile, at a distance of less than 2700 m, the average probability of their presence is ±0.5 (Figure 11).This shows that the probability of orangutan presence near the company and community plantation blocks is high.Orangutans are able to roam as far as 1-2 km depending on the availability of food sources and for males, it is also determined by the distribution of adult females [50,51].Plantation blocks become important in mitigation planning.
Potential conflict areas, according to altitude, show that at an altitude of less than 150 m above sea level, the probability of orangutan presence is up to 0.6.At this altitude, human access to land is easier than at the height above it.Field observations show that in areas that have high and medium habitat suitability at these altitudes, humans have accessed a lot, for example, in land clearing activities for cultivation and oil palm plantations, logging, as well as opportunistic activities.The probability of the presence of orangutans based on the ease of access to the location by humans is also shown by the 1243 (2023) 012022 IOP Publishing doi:10.1088/1755-1315/1243/1/01202210 probability of land clearing, where the steeper it is, the probability of the presence of orangutans also decreases (Figure 12 and Figure 13).
The potential for conflict around settlement areas is low.The probability of the presence oforangutans above 0.5 can occur at a distance of 7000 meters from the settlement (Figure 14).Settlementsin this area are generally located on public roads and banks of the Pawan river.The habitat suitability map also shows that the Pawan riverbank area has a low probability of orangutan presence.Likewise, the response to the probability of residential land use and distance from public roads (Figure 15).

Discussion
Information on the probability of presence according to environmental variables that trigger conflict is important in mitigation planning, bearing in mind that each orangutan individu has a characteristic pattern of specific distribution and roaming patterns.Orangutans are a semi solitary species.Distribution and roaming patterns are determined by the availability of feed resources, social status, and for adult males, it is also determined by the distribution of adult females [51].According to this roaming pattern, the type of orangutan is divided into three types, namely settlers, travelling, and nomads [52].Individuals with high social status, such as adult females (and their offspring), and adult males are settlerindividuals who spend most of the year in certain habitat areas with good quality, with a home range ofabout 2-10 km [37,53].Adult and young male individuals who regularly stay in a specific habitat area for weeks or months and then move to other habitats and then move back to their original habitat are the traveling types.This travelling type cruise 1-2 km/day will have a long range.The last type is the nomad type.Individuals with this type are generally young males, who never return or very rarely return to their original habitat for at least three years [54] This study has produced a map of habitat suitability and potential conflicts between orangutans and humans.The results of this study can be used as a basis for planning mitigation and adaptation measures for oil palm plantations for the presence of orangutans.Before determining the stages and techniques of mitigation, this study proposes the division of orangutan habitat zoning to facilitate mitigation planning.Zoning is carried out by considering connectivity between regions [18].
In the northern part of the study area, there is a forest habitat center connected to the study area through the protected forest area of Mount Tarak.Zoning is conducted by choosing a habitat area of more than 10 ha.According to [50], the home range of forest people in two different locations is 16.9 ha and 11.7 ha.Meanwhile, according to [51], who researched Mount Palung, the home range of female orangutans was 36 ha, and adult males were 43 ha.This study uses the minimum habitat area value as a precautionary principle and the lack of orangutan habitat in the study area.
There are areas with high and medium habitat suitability classes in the study area, which are fragmented into 68 polygons with a size ranging from 10.1 ha -43,239.8ha.The polygons are divided into four zones (Figure 16 and Table 2).This zoning division is based on the aggregation of orangutan habitat polygons.The following is the distribution of orangutan habitat zoning in the study area: x Zone-1 is the Gunung Tarak Protected Forest zone, a buffer area for Gunung Palung National Park.Zone 1 has one complete polygon covering an area of 6922.7 ha.Around 14 polygons are fragmented, with a size ranging from 10-500 ha.Zone-1 can serve as a corridor connecting Zone-2and Zone-3 to the center of orangutan habitat in Gunung Palung National Park.x Zone 2 is orangutan habitat fragmented by oil palm plantations and community farming.Zone-2 has one intact polygon with an area of 1568.6 ha and 17 polygons with less than 500 ha.The orangutan habitat in Zone-2 which is included in the PT KAL concession, has been designated as a conservation area or HCV [51].x Zone-3 is a habitat located in the production forest area between the Tulak and Pawan rivers.
Zone-3 is the largest orangutan habitat in the study area, with one intact polygon covering an area of 43,871.0ha.In the Northeastern part of zone-3, an HCV conservation area belongs to PT DAS (BGA Group).Zone-3 is the core zone of orangutan habitat in the study area.x Zone-4 is a habitat located in the South-Southeast of the Pawan River.In zone-4, the orangutan habitat has been fragmented into small polygons with an area of less than 110 as many as 12 polygons.It only has one intact polygon with an area of 702.7 ha.Zone-4 is difficult to connect with other zones, because the Pawan River bounds it.This river is a means of transportation with settlements on the border.
Oil palm companies can mitigate orangutan conflicts by taking preventive and curative measures.Preventive action is to prevent conflicts or disturbances caused by orangutans.Compared to Curative measures, these Preventive measures will be more sustainable and last longer.Based on the interaction with the managed object, Preventive actions can be categorized into passive and active actions.Passive action is related to managing land use, obstacles and avoiding direct confrontation with orangutans.Active Preventive action is prevention before conflict occurs by interacting directly with orangutans, such as relocation/translocation.This Active Preventive action is supplemented by Curative actions carried out by the competent authorities and experts in their fields, for example, BKSDA (wildlife authority).
The results of this study suggest more Passive Preventive Mitigation.Oil palm plantations have limited human resources, who are experts in handling wildlife and have authority.Preventive Curative measures are only carried out in Zone-4.This matter is also an adaptation of oil palm plantations, where managing orangutan conflicts is integral to sustainable plantation management.The followings are the stages and Passive and Active Preventive Mitigation actions that can be carried out by oil palm plantations, both companies and communities in each zone.The presence of OU and the management of conflict with OU could add to the ecological value of the plantation, meaning that oil palm can provide some living for the OU.If an oil palm plantation could prove to the world that it can also protect OU,

Conclusion
The research area is divided into three classes of orangutan habitat suitability: high, medium, and low.The dominant environmental variables in determining suitability classes are land use and cover.The high and medium habitat suitability classes are primarily located in forested areas, inside or outside forest administrative areas.The suitable habitat for orangutans in the study area is fragmented by community farming areas and oil palm plantations.Human activities in oil palm plantations contributed the most in determining the potential conflict vulnerability map.The presence of orangutan near oil palm plantations has a high probability.The map of the habitat suitability class and the potential conflict vulnerability level from the study can determine conflict mitigation measures.The mitigation zone in the study area is divided into four zones.Zones 1, 2, and 3 can be linked by developing an orangutan corridor to connect with the orangutan habitat center in Gunung Palung National Park.Zone-1 is a Gunung Tarak Protected Forest, a buffer for the national park.Zone-1 is the core orangutan conservation zone in the study area.This study suggests that the oil palm companies in zones 1, 2, and 3 prioritize

Figure 1 .
Figure 1.Study Area (Legal boundaries of forest area, HP: Production forest; HPK: production forest can be converted; HL: protected forest; APL: other landuse areas or cultivated land)

Figure 2 .
Figure 2. Area under curve of model perform

Figure 3 .
Figure 3. Class of orangutan habitat suitability in the study area

Table 2 .
Number of polygons (n) by category of polygon area and habitat zoning