REINWARDTIA Vol. 22. No. 1. pp: 55‒67 DOI: 10.55981/reinwardtia.v22i1.4578 55 MODELLING THE POTENTIAL DISTRIBUTIONS OF SAWO KECIK (MANILKARA KAUKI (L.)) DUBARD USING MAXENT TO SUPPORT CONSERVA- TIONS OF HISTORICAL AND CULTURAL VEGETATIONS IN DAERAH ISTIMEWA YOGYAKARTA PROVINCE Received April 29, 2023; accepted June 6, 2023 ANDRI WIBOWO Department of Biology, Faculty of Mathematics and Natural Sciences, Universitas Indonesia. Kampus UI Gedung E Level 2, Jln. Lingkar Kampus Raya, Pondok Cina, Beji, Depok 16424, Indonesia. Email: awbio2021c@gmail.com. https://orcid.org/0000-0001-7787-5735. ATUS SYAHBUDIN Faculty of Forestry, Universitas Gadjah Mada. Jln. Agro, Bulaksumur, Sleman, 55281, Yogyakarta, Indonesia. https://orcid.org/0000-0002-8614-1925. ADI BASUKRIADI Department of Biology, Faculty of Mathematics and Natural Sciences, Universitas Indonesia. Kampus UI Gedung E Level 2, Jln. Lingkar Kampus Raya, Pondok Cina, Beji, Depok 16424, Indonesia. ERWIN NURDIN Department of Biology, Faculty of Mathematics and Natural Sciences, Universitas Indonesia. Kampus UI Gedung E Level 2, Jln. Lingkar Kampus Raya, Pondok Cina, Beji, Depok 16424, Indonesia. ABSTRACT WIBOWO, A., SYAHBUDIN, A., BASUKRIADI, A. & NURDIN, E. 2023. Modelling the potential distributions of Sawo Kecik (Manilkara kauki (L.)) Dubard using MaxEnt to support conservations of historical and cultural vegeta- tions in Daerah Istimewa Yogyakarta Province. Reinwardtia 22(1): 55‒67. — Sawo kecik or Manilkara kauki (L.) Dubard, of the Sapotaceae family as it is formally known, is one of the species with significant cultural values in Yog- yakarta Province (DIY) culture because it symbolizes social righteousness. In connection with this, Yogyakarta's mu- nicipal and district governments have been encouraged to plant sawo kecik. Despite these efforts, there is still a lack of knowledge regarding the possible range of this species, and this knowledge is essential to promoting the conservation of M. kauki in DIY. With the help of isothermality, precipitation of driest month, precipitation seasonality, precipita- tion of driest quarter, and precipitation of warmest quarter data, this study tries to simulate the probable distributions of M. kauki throughout cities and districts in DIY. The model estimated 1,275 km 2 of DIY areas was suitable for M. kauki that concentrated in the central parts, spanning from the west to the east of DIY. Yogyakarta City followed by Sleman District has the largest areas categorized from high to very high suitable for M. kauki. While, Gunung Kidul followed by Kulonprogo Districts have the largest areas categorized as low suitable. To conclude, M. kauki can adapt areas with moderate precipitation as low as 20 mm during driest month and as low as 100 mm during driest quarter. During warmest quarter, M. kauki requires precipitation with value of 700 mm. The conservation effort and M. kauki planting should then concentrate on Yogyakarta City and Sleman District since such locations are thought to have high appropriateness for the species. Key words: Distr ibutions, M anilk ara k auk i, MaxEnt, Yogyakarta. ABSTRAK WIBOWO, A., SYAHBUDIN, A., BASUKRIADI, A. & NURDIN, E. 2023. Pemodelan potensi distribusi Sawo Kecik (Manilkara kauki (L.)) Dubard dengan MaxEnt untuk konservasi vegetasi bernilai sejarah dan budaya di Provin- si Daerah Istimewa Yogyakarta. Reinwardtia 22(1): 55‒67. — Sawo kecik, atau secara ilmiah dikenal sebagai Manil- kara kauki (L.) Dubard, suku Sapotaceae merupakan salah satu spesies yang memiliki nilai budaya penting dalam kebudayaan Provinsi Daerah Istimewa Yogyakarta (DIY) karena mewakili kebaikan di tengah masyarakat. Terkait hal tersebut, pemerintah provinsi telah mendorong kota dan kabupaten di Yogyakarta untuk menanam sawo kecik. Meski demikian, informasi tentang potensi sebaran jenis ini masih terbatas dan informasi ini sangat penting untuk men- dukung konservasi M. kauki di DIY. Penelitian ini bertujuan untuk memodelkan sebaran potensial M. kauki di kota dan kabupaten di DIY dengan menggunakan metode Maximum Entropy (MaxEnt) sebagai kebaruan dengan menggunakan isotermalitas, presipitasi bulan terkering, presipitasi musiman, presipitasi kuartal terkering, dan presipi- tasi kuartal terhangat variabel. Model memperkirakan wilayah DIY seluas 1.275 km 2 cocok untuk M. kauki yang ter- konsentrasi di bagian tengah, membentang dari barat hingga timur DIY. Kota Yogyakarta diikuti Kabupaten Sleman memiliki wilayah terluas dengan kategori potensi sebaran tinggi hingga sangat tinggi yang cocok untuk M. kauki. Se- dangkan Kabupaten Gunung Kidul yang diikuti Kabupaten Kulonprogo memiliki wilayah terluas dengan kategori po- tensi sebaran rendah. Disimpulkan bahwa M. kauki dapat beradaptasi di daerah dengan curah hujan sedang serendah 20 mm selama bulan terkering dan serendah 100 mm selama kuartal terkering. Pada kuarter terhangat, M. kauki mem- butuhkan presipitasi dengan nilai 700 mm. Kemudian untuk mencapai hasil yang optimal, upaya konservasi dan pena- mailto:afri.irawan31@gmail.com https://dx.doi.org/10.55981/reinwardtia.v22i1.4578 https://orcid.org/0000-0001-7787-5735 https://orcid.org/0000-0002-8614-1925 REINWARDTIA 56 [VOL.22 INTRODUCTION Manilkara kauki (Dubard, 1915) or local name is known as sawo kecik belongs to the Sapotaceae family and is native from the Malay Peninsula to the Pacific. Manilkara kauki was introduced into the West Indies many years ago, they are rarely seen today (Govaerts, 2017). M. kauki grows on beaches in open forest, monsoon forest, vine thickets, and rain forest. This species has at least 11 synonyms. In Indonesia, M. kauki is distributed from Java, Bali, and the Papua Islands. Besides that, this species is also observed on the southern coast of Banyuwangi, the Jakarta coast, the Kari- mun Jawa islands, Bali, Sulawesi, Kagean, Weh Island, and in Nusa Tenggara with elevation limits of 300 m (Alrasyid, 1971). Research by Sudrajat and Megawati (2010) has reported M. kauki presences in Benoa, Mokmer, Kaliurang, Lembar, and Alas Purwo within elevation ranges of 1–332 m. This species is adapted to rainfall rate ranges of 1,500–4,488 mm. Recent research has indicated potential distribu- tion of M. kauki. Regarding this, several methods have been developed to model species distribution at spatial scale. One approach that has been used widely to model the potential spatial distributions of a species is known as Maximum Entropy (MaxEnt) modelling. This model has been used widely to estimate potential distributions of ani- mal (Stephenson et al., 2022), including ticks (Sanchez et al., 2023), vegetation (Dong et al., 2023), and crops. In Indonesia, potential distribu- tion modelling studies using MaxEnt has been developed for X anthomonas campestris (Saputra et al., 2023) and Indigofera tinctoria along Cita- rum Watershed (Usmadi et al., 2021). Besides MaxEnt, there are a growing variety of methods for estimating habitat appropriateness, including geographical based methods (BIOCLIM, DO- MAIN, BIOMAPPER), statistical based methods (generalized additive model/GAM, GLM), ma- chine learning based methods (MaxEnt, Random Forest, Support Vector Machine/SVM), and deep learning based methods (Artificial Neuron Net- work/ANN). Each tool is unique, with its own set of pros and downsides. According to Marcer et al. (2013), among other things, MaxEnt is one of the most often used habitat suitability modeling tools. The need for only species presence data, the abil- ity to run with a little amount of data, the high accuracy of prediction findings, the high repro- ducibility, and the ability to predict the most dis- criminating environmental parameters are all ad- vantages of MaxEnt (Fois et al., 2018). DI Yogyakarta (DIY) is one of the provinces in Indonesia that has promoted the cultivation and planting of M. kauki. According to Governor of DIY Circular No. 194/Kep/2015, entitled Designa- tion of Wanadesa Locations in DIY Year of 2015, it has been ordered to plant M. kauki across 11 vil- lages in Kulonprogo, Bantul, Sleman, and Gunung Kidul Districts. In Bantul District, M. kauki has been assigned as flag species as stated in Mayor of Bantul Circular No. 567/B/Kep/BT/1998. In DIY, M. kauki is among the trees that are selected to be planted inside Kraton (Widayatsari, 2002) because it has important philosophical and cultural values, especially for the Sultanate of Yogyakarta. Manil- kara kauki is representing sarwo bercik value or social righteousness (Wastuty, 2007). Despite this progress, there is still limited information on where the potential distribution of M. kauki is located, mainly in DIY. This information is required in ad- vance to guarantee and sustain M. kauki plantings. Here, this research aims to determine the potential distribution areas for M. kauki in DIY. The novelty of this research is the use of MaxEnt modeling to achieve accuracy in potential distributions. The results of this study will benefit the success of con- servation of historical and cultural vegetation in DIY. MATERIALS AND METHODS The research was conducted on DIY Province. The study methodology followed methods deve- loped by Semu et al. (2021), including species oc- currence, environmental variables, and model eva- luation. The method to estimate and model the po- tential distributions of M. kauki was comprises several steps (Fig. 1). It is started with the species occurrence recording and multicollinearity test to select relevant environmental variables. Study sites This research was conducted in one city (Kodya Yogyakarta) and four districts (Kab. Sleman, Ku- lonprogo, Bantul, Gunung Kidul) of DIY Province (Fig. 2). The geo-coordinates of DIY were 110.0°– 110.9° east longitude (E) and 7.2°–8.3° south lati- tude (S). DIY is bordered by highlands in the north, hilly areas in the east and west, and sea and coastal areas in the south. The southern parts of DIY were karst areas characterized by limited sur- face water and experiencing dryness (Putra & Nur- naman M. kauki sebaiknya difokuskan di Kota Yogyakarta dan Kabupaten Sleman karena daerah tersebut dianggap memiliki tingkat potensi sebaran yang tinggi. Kata kunci: M anilk ara k auk i, MaxEnt, sebar an, Yogyakar ta. WIBOWO et al.: Modelling the potential distribution of Manilkara kauki (L.) Dubard. 2023] 57 jani, 2021). The annual rainfall ranges of DIY are 718.0–2,992.3 mm/year. High rainfalls were ob- served in the northern parts of Sleman District. While low rainfalls were observed in the southern parts of Gunungkidul and Bantul districts, Sleman District has recorded rainfall rates up to 2,992.3 mm/year. While the lowest rainfalls with values of 197.6 mm/year were observed in the southern parts of Gunung Kidul District. According to Hakim & Yuliah (2018), DIY is known for having a high diversity of vegetation. It was recorded that DIY has at least 805 plant spe- cies representing 110 families inhabiting terrestrial ecosystems, 12 families residing in aquatic ecosys- tems, and 4 families under protection. Among those species, at least 33 plant species are catego- rized as very rare, and M. kauki is included in that list. M. kauki is protected under Mayor of Bantul Circular No. 567/B/Kep/BT/1998. Among those species, there are 7 species that have cultural val- ues, including M. kauki, which represents sarwo becik cultural values symbolizing social righteous- ness. Manilkara kauki occurrence surveys and record- ings Explorations or field surveys combined with literature and database reviews were conducted to survey and record the presence of M. kauki in DIY from January to March 2023 covering Yog- yakarta City, Kulonprogo, Sleman, Bantul, and Gunung Kidul Districts. The presence of M. kauki was recorded using direct visual observation or also known as visual encounter survey (VES) and a database (Tabel 1) provided and gathered from literature reviews sourced from journal articles and reports provided by government agencies, including the agency for agriculture and forestry at the Indonesian Ministry of Environment and Forestry. VES was implemented purposefully by visiting parks, household yards, and roadside areas where M. kauki may be planted. The sample size covered all trees planted. The geographical coor- dinates of M. kauki presences in the field were recorded using the Garmin Etrex 30 type Global Positioning System (GPS). The data were convert- Fig. 1. A flowchart of the suitability analysis and geographical distribution modeling. Fig. 2. Location map of the study sites in five districts within DIY Province, Indonesia REINWARDTIA 58 [VOL.22 ed to Microsoft Excel and saved in CSV format for use in MaxEnt habitat suitability modelling. The tree species identification guideline to deter- mine M. kauki was based on identification keys (Backer & van den Brink, 1963; Partomihardjo et al., 2014). Manilkara kauki environmental variables This study included various environmental vari- ables (Table 2) following Dong et al. (2023) and Arshad et al. (2022). For the recent time, bio- climatic variables (Bio 1–Bio 19) from the global climate database WordClim (www.worldclim.org, the new version 2.0) (Hijmans et al., 2005) have been employed extensively in habitat suitability modeling (Khanum et al., 2013) and are widely used in the Asian region (Rana et al., 2017). Those environmental variables were chosen based on selection and utilization of environmen- tal elements having a significant influence in or- der to obtain an accurate and informative habitat suitability model. Jackknife analysis was used to evaluate the contribution of each environmental variable to the resulting model. Some environ- mental variables resulted from Jackknife analysis were not used due to the lack of contribution to the model making (percent contribution = 0). Those environmental variables were variables with a small average contribution (< 6%) or per- mutation importance (< 6%) (Wei et al., 2018). The contribution percentage and permutation are two important factors for understanding and meas- uring the environmental variable’s contribution as well as importance to the MaxEnt model. Multicollinearity test To establish a model that has better perfor- mance with fewer variables and to avoid colline- arity between the variable, a multicollinearity test was performed using Pearson’s correlation tests (Preau et al., 2018) on 19 environmental variables (Bio 1–Bio 19) (Table 2). The variables that have highly cross-correlated variables (r 2 > 0.8) were excluded and variables having r 2 < 0.8 were kept for further analysis for geographical distribution modeling. If multicollinearity occurs, then a varia- ble is strongly correlated with other variables in the model, and its predictive power is unreliable and unstable (As’ary et al., 2023). Based on the multi- collinearity test, the selected environmental varia- bles to be used were Bio 3, 14, 15, 17, and 18. Manilkara kauki suitability analysis This study employed MaxEnt analysis using MaxEnt packages within R platform version 3.6.3 (Mao et al., 2022) to generate predicted suitability maps of M. kauki across DIY. Several R packages required to develop the suitability maps include library ("sp"), library ("dismo") (Khan et al., 2022), library ("maptools"), library ("rgdal") (Bivand et al., 2022), and library ("raster") (Lemenkova et al., 2020). The inputs for MaxEnt included 19 environmental variables (Bio 1–Bio 19).Within the model, the contribution and impact of each environmental variable on the M. kauki habitat suitability model were determined using a Jackknife test (Promnikorn et al., 2019), and the receiving operating curve (AUC) area was used to evaluate the performance model. AUC values range from 0 (least suitability) to 1, with a value less than 0.5 suggesting that the resulting model is no better than random and uninformative data and a value more than 1.0 showing that the resulting model is highly good and informative. The prediction map resulting from MaxEnt mod- els was imported into GIS for presentation and ad- ditional study (Hijmans et al., 2012). According to Wei et al. (2018), habitat suitability levels on the MaxEnt model map can be classified into five suit- ability level included 0: unsuitable, 1: low suitabil- ity, 2: medium suitability, 3: high suitability, 4: very high suitability. City/districts Title Authors Year Bantul Laporan kinerja pengelolaan lingkungan hidup daerah Kabupaten Bantul Dinas Lingkungan Hidup daerah Kabupaten Bantul 2016 Yogyakarta Alternatif pohon buah untuk penghijauan per- mukiman perkotaan berdasarkan pendugaan ting- kat keindahan dan pendapat masyarakat di Kelurahan Rejowinangun, Yogyakarta Annisa, N. S., Irwan, R., S. N. Kurniasih, B. & Ambarwati, E. 2018 Kulonprogo, Gunung Kidul Exploration and characterization of sapodilla (Manilkara zapota (l.) van Royen) in Daerah Istimewa Yogyakarta Rozika, Murti, R. H. & Purwanti, S. 2013 Sleman Kebun Raya Botani di Kabupaten Sleman Jhonson 2019 Table 1. Literature review source materials of M. kauki in Indonesian. WIBOWO et al.: Modelling the potential distribution of Manilkara kauki (L.) Dubard. 2023] 59 Manilkara kauki model evaluation and validation This study's model evaluation follows Reddy et al. (2015) and Song et al. (2023). Area under the curve analysis (AUC) was used to examine the model. The MaxEnt model calculat- ed the percentage contribution of each factor to the species distribution. The percentage contribution represents the value of each factor's contribution to the distributions of the species. The size of the re- ceiver operating characteristic curve (ROC) and the area under the curve (AUC) were used to as- sess model prediction accuracy. The higher the AUC value, the greater the accuracy of the model's prediction outcomes, and the parameters of the MaxEnt model were selected in accordance with Zhao et al. (2018). AUC is an effective and effi- cient independent threshold index with the capacity of assessing the model’s capacity to distinguish the presence and absence. AUC values are categorized in to five different classes based on performance. The performance classes are failing (0.5 to 0.6), bad (0.6 to 0.7), reasonable (0.7 to 0.8), good (0.8 to 0.9) and great (0.9 to 1). Models with values less than 0.5 indicates that the occurrence in the real- life scenario is rare or can be considered as a guesstimate (Shcheglovitova & Anderson, 2013). Jackknife was executed to systematically exclude each variable or evaluate the leading bioclimatic or topographic variables. Jackknife evaluates the leading variables in determining the potential dis- tribution of species. The relationship between the selected environmental factors from 19 environ- mental variables and the potential habitat for the species is determined from the created response curve from the model (Vila et al., 2012). The rela- tive contributions in percentage of the each envi- ronmental variable to the MaxEnt model were cal- culated. RESULTS Manilkara kauki observed morphology M. kauki common morphology was observed based on M. kauki occurrence surveys across DIY. The observed M. kauki has tree height ranges of 10–25 m. Manilkara kauki occurrences Manilkara kauki is mainly present (Fig. 2) between 110.2°–110.5° E and 7.68°–7.84° S in the central parts of DIY, with hilly areas in the east Variables Sources Format Unit Annual mean temperature (Bio 1) www.worldclim.org Image data in Raster °C Mean diurnal range (Bio 2) (mean of monthly (max temp - min temp)) www.worldclim.org Image data in Raster °C Isothermality (Bio 3)* www.worldclim.org Image data in Raster % Temperature seasonality (Bio 4) www.worldclim.org Image data in Raster °C Max temperature of warmest month (Bio 5) www.worldclim.org Image data in Raster °C Min temperature of coldest month (Bio 6) www.worldclim.org Image data in Raster °C Temperature annual range (Bio 7) www.worldclim.org Image data in Raster °C Mean temperature of wettest quarter (Bio 8) www.worldclim.org Image data in Raster °C Mean temperature of driest quarter (Bio 9) www.worldclim.org Image data in Raster °C Mean temperature of warmest quarter (Bio 10) www.worldclim.org Image data in Raster °C Mean temperature of coldest quarter (Bio 11) www.worldclim.org Image data in Raster °C Annual precipitation (Bio 12) www.worldclim.org Image data in Raster mm Precipitation of wettest month (Bio 13) www.worldclim.org Image data in Raster mm Precipitation of driest month (Bio 14) * www.worldclim.org Image data in Raster mm Precipitation seasonality (Bio 15) * www.worldclim.org Image data in Raster dimensionless Precipitation of wettest quarter (Bio 16) www.worldclim.org Image data in Raster mm Precipitation of driest quarter (Bio 17) * www.worldclim.org Image data in Raster mm Precipitation of warmest quarter (Bio 18) * www.worldclim.org Image data in Raster mm Precipitation of coldest quarter (Bio 19) www.worldclim.org Image data in Raster mm Tabel 2. Variables used in this study *: selected variables based on multicollinearity test REINWARDTIA 60 [VOL.22 Fig. 3. Current presence records of Manilkara kauki across one city and four districts within DIY Province. Fig 4. Response curves of suitability predicted values of Manilkara kauki with Bio 3: isothermality, Bio 14: precipitation of driest month, Bio 15: precipitation seasonality, Bio 17: precipitation of driest quarter, and Bio 18: precipitation of warmest quarter. WIBOWO et al.: Modelling the potential distribution of Manilkara kauki (L.) Dubard. 2023] 61 Fig. 5. Distributions of Manilkara kauki current presence records related to environmental variables includ- ing Bio 3: isothermality (%), Bio 14: precipitation of driest month (mm), Bio 15: precipitation seasonality (dimensionless), Bio 17: precipitation of driest quarter (mm), and Bio 18: precipitation of warmest quarter (mm) across one city and four districts within DIY Province, Indonesia. Fig. 6. The Receiver Operating Characteristic (ROC) curve result of the MaxEnt modelling. REINWARDTIA 62 [VOL.22 and west according to the collected occurrence data and survey. In total, there were nineteen oc- currences of M. kauki (Fig. 3), with 84.21% of occurrences dominating the central parts. Only one record near the coastal areas. Manilkara kauki occurs mostly in hilly areas and lowlands with altitude ranges of 33–176 m in DIY. Manilkara kauki was absent in near-coastal areas with an altitude less than 30 m. Manilkara kauki response curves Fig. 4 shows response curves of suitability pre- dicted values of M. kauki potential distributions with Bio 3: isothermality, Bio 14: precipitation of driest month, Bio 15: precipitation seasonality, Bio 17: precipitation of driest quarter, and Bio 18: precipitation of warmest quarter. Among those environmental variables, significant responses were observed for isothermality, precipitation of driest month, and precipitation of driest quarter variables. Manilkara kauki responds immediately toward slight increases of those variables. This condition differs as can be seen for annual precipi- tation of warmest quarter. Manilkara kauki res- ponds gradually toward this variable. Manilkara kauki environmental variables Environmental variable spatial distributions related to the presence of M. kauki across DIY are available in Fig. 5. Those variables include iso- thermality, precipitation of driest month: precipi- tation seasonality, precipitation of driest quarter, and precipitation of warmest quarter. retrieved from the WordClim database. All variables were observed limit the distribution of M. kauki current presence. Coastal areas with low precipitation of warmest quarter < 700 mm has limited M. kauki distributions in here. While high isothermality with values of > 70% and low precipitation sea- sonality with values of < 60 has limited M. kauki distributions in hilly areas. Manilkara kauki can adapt areas with moderate precipitation as low as 20 mm during driest month and as low as 100 mm during driest quarter. During warmest quarter, M. kauki requires precipitation with value of 700 mm. Manilkara kauki was observed clustered near areas with high isothermality with values of > 76% covering north of Kulonprogo and Sleman Districts. MaxEnt model evaluation and validation Model evaluation and validation guarantees the reliability of the model and the reliability is ex- pressed by the area under receiver-operating cha- racteristic (ROC) curve (AUC) obtained by the accuracy test of the ROC curve analysis method. The AUC values were between 0 and 1 and divid- ed into several value classes. When the AUC va- lue was lower than 0.5, the model executed was considered poor than contingency. The MaxEnt model is more precise and descriptive when the AUC test value is closer to 1, indicating better discrimination. In this study, the AUC value showed that the MaxEnt model performed well with AUC value of 0.902 under the model at the current time (Fig. 6).This indicates that MaxEnt model has an accurate prediction on the potential distribution region of M. kauki in DIY. Manilkara kauki MaxEnt model The MaxEnt species occurrence probability output raster for M. kauki was classified, mapped, and evaluated for land area calculation for each city and district considered suitable (Fig. 7). In total, there were estimated 1,275 km 2 of areas in DIY that considered suitable for M. kauki and falls into four suitability levels. The MaxEnt model classified the suitability levels into levels that vary in size among cities and districts. The most suitable habitats for M. kauki in DIY were estimated to be concentrated in the central parts, spanning from the west to the east of DIY Prov- ince. Kulonprogo, Bantul, Sleman districts and Yogyakarta City were observed as regions that have level 3 and level 4 suitabilities or are catego- rized as having high and very high suitabilities (Fig. 8). Level 1 and level 2 suitabilities, catego- rized as low and medium suitabilities, were ob- served in Gunung Kidul, followed by Kulonpro- go, Bantul, and Sleman districts. Each district and city across DIY has a differ- ent composition of potential suitable areas for M. kauki (Fig. 9). Areas of Yogyakarta City were dominated by high suitability, with similar com- positions equaling 50% of total Yogyakarta City Areas. In Sleman, 10% of its areas were consi- dered very suitable for M. kauki, and 20% were considered suitable. Almost all the areas in Gunung Kidul were considered to have low suita- bility for M. kauki, followed by Kulonprogo and Bantul Districts. DISCUSSION This is the first study, especially in Indonesia, to analyze the range expansions of M. kauki based on the MaxEnt species distribution model. To ensure the model's accuracy, the occurrence data and environmental variables have been care- fully selected and validated. Model parameter optimization and evaluation for M. kauki were made using the AUC, and the AUC values indi- cated the high prediction accuracy of the MaxEnt model. Currently, in Indonesia, most studies of M. kauki were mostly focused and limited on tax- onomy and morphological studies (Sofiyanti & Iriani, 2023) and M. kauki for absorbing CO2 (Anggara & Rahmawati, 2021). Here, this study has expanded by estimating potential distributions of M. kauki at province scale. WIBOWO et al.: Modelling the potential distribution of Manilkara kauki (L.) Dubard. 2023] 63 Manilkara kauki distributions on DIY were shown to be limited by coastal areas as can be seen in Gunung Kidul and Kulonprogo areas. This finding is quite contradicted with reports confirm- ing potential distribution of M. kauki that also in- clude coastal areas. The potential absences of M. kauki in coastal areas in Gunung Kidul and Ku- lonprogo areas can be related to the internal and external factor or environmental variables. As an internal variable the presences of M. kauki were determined by seed production during the first year and reproductive event during the second (Cruz-Rodríguez et al., 2009). As an external fac- tor, soil fertility is one of the factors. While soils in Kulonprogo are classified as inceptisol soils. This kind of soil has N, P, and K nutrient deficits (Ciptaningrum, 2022) that limits the suitability of soils for M. kauki. The K deficit will limit the availability of soil pH, K total, K available nutri- ent, and led to K uptake by the plant (Abdillah et al., 2011). While according (Chakradhari et al., 2019), leaves, seed kernel, and seed coat of Manil- kara Genus do have high level of K and this indi- cates the importance of K to determine the suita- ble habitats for M. kauki that in fact K was li- mited in the Kulonprogo areas and explain the low suitability of this area. Besides K deficient, Kulonprogo is a coastal area where most of the soils comprise sandy soils (Sutardi, 2017). According to Hani et al. (2016), the sandy dominated soils in Kulonprogo were having deficient in NPK contents, C organic and also has low pH with ranges of 4.51–5.38. Accord- ing to Falasca et al. (2016), species belong to Sa- potaceae family require neutral soils with pH rang- ing from 6.5 to 7.5. Based on the model, the hill central parts in Bantul, Sleman districts, and Yogyakarta City are considered as the most suitable. The elevation and precipitation of those areas were fulfilling the re- quirement for M. kauki to grow. The growth of M. kauki will reach its optimum results at the eleva- tion ranges from 600 to 2,100 m above sea level (Pohlan et al., 2000). While Manilkara genus be- longs to Sapotaceae can tolerate dry climate, it requires from moderate to high rainfall with ra- nges of 1,000 to 4,000 mm. These suitable rain- falls can be observed in the Bantul, Sleman dis- tricts, and Yogyakarta City. In those areas the M. kauki can adapt areas with moderate precipitation as low as 20 mm during driest month and as low as 100 mm during driest quarter and those areas were considered suitable for M. kauki (Falasca et al., 2016). Another variables that limit the distribution M. kauki is precipitation seasonality and isothermali- ty. Precipitation seasonality (Souza et al., 2016 ) affects the soil moisture and potential distributions of M. kauki (Ayanlade et al., 2021). M. kauki was occurred near areas with high isothermality. Ac- cording to (Huang et al., 2021), wide-range plant groups tended to occur in areas with higher tem- Fig.7. Distribution of potential suitable areas for Manilkara kauki across one city and four districts within DIY Province, Indonesia (Suitability level 0: unsuitable, 1: low suitability, 2: medium suitability, 3: high suitability, 4: very high suitability). REINWARDTIA 64 [VOL.22 perature variability and isothermality. CONCLUSION DIY has considerably suitable areas for M. kauki sizing 1,275 km 2 of DIY areas. The suitable areas were increasing toward central parts of DIY and spanning from the west to the east of DIY. Yogyakarta City followed by Sleman district has the largest areas categorized from high to very high suitable for M. kauki. While, Gunung Kidul followed by Kulonprogo districts have the largest areas categorized as low suitable. To conclude, M. kauki can adapt areas with moderate precipitation as low as 20 mm during driest month and as low as 100 mm during driest quarter. During warmest quarter, M. kauki requires precipitation with value of 700 mm, Then for achieving optimum results, the conservation effort and planting of M. kauki should focus on Yogyakarta city and Sleman dis- Fig.8. Distribution of potential suitable areas in km 2 for Manilkara kauki across one city and four districts within DIY Province, Indonesia related to district areas (Suitability level 1: low suitability, 2: medium suit- ability, 3: high suitability, 4: very high suitability). Fig.9. Percentages of potential suitable areasfor Manilkara kauki across one city and four districts within DIY Province, Indonesia (Suitability level 1: low suitability, 2: medium suitability, 3: high suitability, 4: very high suitability). WIBOWO et al.: Modelling the potential distribution of Manilkara kauki (L.) Dubard. 2023] 65 trict since those areas considered has high suitabi- lity ACKNOWLEDGEMENTS We are deeply indebted to the supports provided by Universitas Indonesia, Universitas Gadjah Ma- da, and The Sultanate of Yogyakarta that has made this research possible. REFERENCES ABDILLAH, A., SYAMSIYAH, J., RIYANTO, D. & MINARDI, S. 2011. The effect of zeolite and K fertilizer on availability and K uptake at sea shore sandy land of Kulon Progo, Yogya- karta. Bonorowo W etlands 1: 1–7. ALRASYID, H. 1971. Keterangan tentang sil- vikultur sawo kecik (Manilkara kauki Durbard, Sapotaceae). Laporan no. 127. Lembaga Peneli- tian Hutan. Bogor. ANGGARA, O. C. & RAHMAWATI, L. A. 2021. The effectiveness of urban forest in absorbing CO2 emission at Rajekwesi type A terminal. Jurnal Ilmu Lingkungan 1: 60–65. ARSHAD, F., WAHEED, M., FATIMA, K., HA- RUN, N., IQBAL, M., FATIMA, K. & UM- BREEN, S. 2022. Predicting the suitable cur- rent and future potential distribution of the na- tive endangered tree Tecomella undulata (Sm.) Seem. in Pakistan. Sustainability 14: 7215. DOI: 10.3390/SU1412721. AS’ARY, M., SETIAWAN, Y. & RINALDI, D. 2023. Analysis of changes in habitat suitability of the Javan Leopard, 2000–2020. Diversity 15: 529. DOI: 10.3390/d15040529. AYANLADE, A., JEJE, O., NWAEZEIGWE, J., ORIMOOGUNJE, O. & OLOKEOGUN, O. 2021. Rainfall seasonality effects on vegetation greenness in different ecological zones. Envi- ronmental Challenges 4. DOI: 10.1016/j.envc. 2021.100144 BACKER, C. A. &VAN DEN BRINK, R. C. B. 1963. Flora of Java (Spermatophytes Only). Volume I. N. V. P. Noordhoff-Groningen, Netherlands. BIVAND, R. 2022. R packages for analyzing spa- tial data: a comparative case study with areal data. Geographical Analysis 54. DOI: 10.1111/ GEAN.12319. CHAKRADHARI, S., DEB, M. K., PATEL, K. S., MARTÍN-GIL, J., TOWETT, E. K. & MAR- TÍN-RAMOS, P. Evaluation of the phytochemi- cal and mineral characteristics of some selected sapotaceae plants. European Journal of Medici- nal Plant 27(4): 1–16. CIPTANINGRUM, E. 2022. Evaluasi Kesuburan Tanah dengan Omission Trial N, P, K yang Di- tambahkan Biochar pada Jagung Semi di In- ceptisol, Kalibawang, Kulon Progo. Ilmu Tanah. Universitas Gadjah Mada. [Undergra- duate thesis]. CRUZ-RODRÍGUEZ, J. A., LÓPEZ-MATA, L. & VALVERDE, T. 2009. A. comparison of traditional elasticity and variance-standardized perturbation analyses: a case study with the tropical tree species Manilkara zapota (Sapotaceae). Journal of Tropical Ecology 25 (2). DONG, H., ZHANG, N., SHEN, S., ZHU, S., FAN, S. & LU, Y. 2023. Effects of climate change on the spatial distribution of the threat- ened species Rhododendron purdomii in Qin- ling-Daba mountains of Central China: impli- cations for conservation. Sustainability 15(4): 3181. DOI: 10.3390/SU15043181 DUBARD, M. M. M. 1915. Annales Du Musee Coloniale De Marseille Ser. 3 (3): 9. FALASCA, S. L., PITTA-ALVAREZ, S. & UL- BERICH, A. 2018. The potential growing are- as for A rgania spinosa (L.) Skeels (Sapotaceae) in Argentinean drylands. International Journal of Agronomy 2018: 9262659. DOI: 10.11 55/2018/9262659. FOIS, M., CUENA-LOMBRAÑA, A., FENU, G. & BACCHETTA, G. 2018. Using species distribution models at a local scale to guide poorly known species, review: method- ological issues and future directions. Ecologi- cal Modelling 385: 124–132. DOI: 10.1016/ J.ECOLMODEL.2018.07.018. GOVAERTS, R. 2017. World Checklist of Sapo- taceae. Richmond, UK: Royal Botanic Gar- dens. HAKIM, L. & YULIAH. 2018. Peran B2P2BPTH Yogyakarta dalam pelestarian jenis-jenis khas Daerah Istimewa Yogyakarta. Seminar Nasio- nal Pendidikan Biologidan Saintek III (2018): 329–337. HANI, A., INDRAJAYA, Y., SURYANTO, P. & BUDIADI. 2016. Dryland agroforestry practi- ces in Menoreh Hills, Kulon Progo. A grivita 38(2). HIJMANS, R. J., CAMERON, S. E., PARRA, J. L., JONES, P. G. & JARVIS, A. 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965–1978. DOI: 10.1002/ JOC.1276. HIJMANS, R. J., GUARINO, L. &MATHUR, P. 2012. Diva-GIS Version 7.5 Manual. HUANG, E., CHEN, Y., FANG, M., ZHENG, Y. & YU, S. 2021. Environmental drivers of plant distributions at global and regional scales. Global Ecology and Biogeography 2021: 1–13. DOI: 10.1111/geb.13251Cita KHAN, A. M., LI, Q., SAQIB, Z., KHAN, N., HABIB, T., KHALID, N., MAJEED, M. & TARIQ, A. 2022. Maxent modelling and im- pact of climate change on habitat suitability REINWARDTIA 66 [VOL.22 variations of economically important chilgoza pine (Pinus gerardiana Wall.) in South Asia. Forests 13: 715. DOI: 10.3390/F13050715. KHANUM, R., MUMTAZ, A. &KUMAR, S. 2013. Predicting impacts of climate change on medicinal asclepiads of Pakistan using MaxEnt modeling. Acta Oecologica 49: 23–31. DOI: 10.1016/J.ACTAO.2013.02.007. LEMENKOVA, P. 2020. Using R Packages 'tmap', 'raster' and 'ggmap' for cartographic vis- ualization: an example of dem-based terrain modelling of Italy, Apennine Peninsula. ZbornikRadova - Geografski Fakultet Univer- ziteta U Beogradu. 68: 99–116. DOI: 10.5937/ ZRGFUB2068099L. MAO, M., CHEN, S., K. E/ Z, QIAN, Z. & XU, Y. 2022. Using Maxent to predict the potential distribution of the little fire ant (W asmannia auropunctata) in China. Insects 13: 1008. DOI: 10.3390/INSECTS13111008. MARCER, A., SÁE, L., MOLOWNY-HORAS, R., PONS, X. & PINO, J. 2013. Using species distribution modelling to disentangle real- ised versus potential distributions for rare species conservation. Biological Conservation 166: 221–230. DOI: 10.1016/ J.BIOCON.2013.07.001. PARTOMIHARDJO, T., ARIFIANI, D., PRATAMA, B. A. & MAHYUNI, R. 2014. Jenis–jenis Pohon Penting di Hutan Nusa- kambangan. LIPI Press, Jakarta. POHLAN, J., BORGMAN, J. & GEHRKE- VELEZ, M. R. 2000. Perspectives of Sapota- ceae species for fruit production in the Soco- nusco Region, Chiapas, Mexico. A cta Horticul- turae. DOI: 10.17660/ActaHortic.2000.531.5 PRÉAU, C., TROCHET, A., BERTRAND, R. & ISSELIN-NONDEDEU, F. 2018. Modeling potential distributions of three European am- phibian species comparing Enfa and MaxEnt. Herpetological Conservation and Biology 13 (1): 91–104. PROMNIKORN, K., JUTAMANEE, K. & KRAICHAK, E. 2019. Maxent model for predicting potential distribution of V itex glabrata R.br. in Thailand. Agriculture and Natural Resources 53: 44–48. PUTRA, Q. I. & NURJANI, E. 2021. Study of meteorological drought and its impact on rain- fed paddy productivity in Gunung Kidul Re- gency. E3S Web of Conferences. 325. 01017. DOI: 10.1051/E3SCONF/202132501017. RANA, S. K., RANA, H. K., GHIMIRE, S. K., SHRESTHA, K. K. & RANJITKAR, S. 2017. Predicting the impact of climate change on the distribution of two threatened Himalayan medicinal plants of Liliaceae in Nepal. Journal of Mountain Science 14(3): 558–570. DOI: 10.1007/S11629-015-3822-1. REDDY, M. T., BEGUM, H., SUNIL, N., PAN- DRAVADA, S. R. & SIVARAJ, N. 2015. As- sessing climate suitability for sustainable vege- table roselle (Hibiscus sabdariffa var. sabdariffa L.) cultivation in India using Maxent. A gricul- tural and Biological Sciences Journal 1(2): 62 –70. SAPUTRA, M., SUTOMO, HUMAIDA, N. & HADIYAN, Y. 2023. Smart farming: modeling distribution of X anthomonas campestris pv. oryzae as a leaf blight-causing bacteria in rice plants. IOP Conference Series: Earth and Envi- ronmental Science: 012026. SÁNCHEZ-PÉREZ, M., FERIA-ARROYO, T. P., VENEGAS-BARRERA, C. S., SOSA-GUTI- ÉRREZ, C., TORRES, J., BROWN, K. A. & GORDILLO-PÉREZ, G. 2023. Predicting the impact of climate change on the distribution of Rhipicephalus sanguineus in the Americas. Sustainability 15: 4557. DOI: 10.3390/SU15 054557. SEMU, A. A., BEKELE, T., LULEKAL, E., CARIÑANOS, P. & NEMOMISSA, S. 2021. Projected impact of climate change on habitat suitability of a vulnerable endemic V achel- lianegrii (Pic.Serm.) Kyal.&Boatwr (Fabaceae) in Ethiopia. Sustainability 13: 1127510. DOI: 3390/SU132011275. SHCHEGLOVITOVA, M. & ANDERSON, R. P. 2013. Estimating optimal complexity for eco- logical niche models: a Jackknife approach for species with small sample sizes. Ecological Modelling 269: 9–17. SOFIYANTI, E. & IRIANI, D. 2023. Anatomical, histochemical, and epidermal characteristic of Sawo Kecik (Manilkara kauki (L.) Dubard - Sapotaceae) leaves. Jurnal Biologi Udayana 27 (1): 14–25. SONG, D., LI, Z., WANG, T., QI, Y., HAN, H. & CHEN, Z. 2023. Prediction of changes to the suitable distribution area of Fritillaria przewal- skii Maxim. in the Qinghai-Tibet plateau under Shared Socioeconomic Pathways (ssps). Sustainability 15: 2833. DOI: 10.3390/su 15032833. SOUZA, R., FENG, X., ANTONINO, A., MON- TENEGRO, S., SOUZA, E. & PORPORATO, A. 2016. Vegetation response to rainfall season- ality and interannual variability in tropical dry forests. Hydrological Processes 30(20): 3583– 3595. DOI: 10.1002/hyp.10953. STEPHENSON, K., WILSON, B., TAYLOR, M., MCLAREN, K., VEEN, R., KUNNA, J. & CAMPBELL, J. 2022. Modelling climate change impacts on tropical dry forest fauna. Sustainability 14(8): 4760. DOI: 10.3390/ SU14084 760. SUDRAJAT, D. & MEGAWATI. 2010. Keraga- man morfologi dan respon perlakuan praperke- cambahan benih dari lima populasi sawo kecik (Manilkara kauki (L.) Dubard. Jurnal Penelitian Hutan Tanaman 7: 67–76. DOI: 10.20886/JPHT.2010.7.2.67-76. WIBOWO et al.: Modelling the potential distribution of Manilkara kauki (L.) Dubard. 2023] 67 SUTARDI. 2017. Kajian minus one test dan kesuburan lahan pasir untuk budidaya tanaman bawang merah. Balai Besar Pengkajian dan Pengembangan Teknologi Pertanian. USMADI, D., SUTOMO, IRYADI, R., HANUM, S. F, DARMA, I. & WIBAWA, I. 2021. Pre- dicting species distribution for true indigo (Indigofera tinctoria L.) in Citarum watershed, West Java, Indonesia. Journal of Tropical Bio- diversity and Biotechnology 6: 1–12. VILÀ, M., GASSÓ, N., THUILLER, W. & PINO, J. 2012. Potential distribution range of invasive plant species in Spain. Neobiota 12: 25–40. WASTUTY, P. W. 2007. Eko arsitektur pada ka- wasan Siliran Lor tinjauan pada vegetasi. Info Teknik 8(1): 64–71. WEI, B., WANG, R., HOU, K., WANG, X..&WU, W. 2018. Predicting the current and future cul- tivation regions of Carthamus tinctorius using Maxent model under climate change in China. Global Ecology and Conservation 16: E00477. DOI: 10.1016/J.GECCO.2018.E00477. WIDAYATSARI, S. 2002. Tata ruang rumah bangsawanYogyakarta. Dimensi Teknik A rsi- tektur 30(2): 122 – 132. ZHAO, W. L., CHEN, H. G., LIN, L., CUI, Z. J. & JIN, L. 2018. Distribution of habitat suitability for different sources of Fritillariaecirrhosae bulbus. Chinese Journal of Ecology 37: 1037– 1042. ZHU, G. P., GARIEPY, T. D., HAYE, T. & BU, W. J. 2017. Patterns of niche filling and expan- sion across the invaded ranges of Halyomor- phahalys in North America and Europe. Jour- nal of Pest Science 90: 1–13. DOI: 10.1007/ s10340-016-0786-z. REINWARDTIA 68 [VOL.22