Conference PaperPDF Available

USING MAXENT IN FINDING SUITABLE LOCATIONS FOR ESTABLISHING FALCATA TREE PLANTATIONS IN CARAGA REGION, MINDANAO, PHILIPPINES

Authors:

Abstract and Figures

The Caraga Region in Mindanao, Philippines, is considered a significant contributor in log production, specifically due to Falcata (Paraserianthes falcataria) plantations. Over 80% of the country's Falcata log production came from Caraga Region in 2019. Among the challenges faced by the tree-growers is finding a suitable location for the establishment of new plantations. We used MaxEnt, a machine learning Species Distribution Modeling (SDM) based on Maximum Entropy principles, for this study's Falcata plantation suitability modeling and mapping. This approach used 2,125 Falcata location points distributed in the region, biophysical factors (i.e., Elevation, Slope, Aspect, and the like), and bioclimatic factors (i.e., Annual Mean Temperature, Isothermality, and Annual Precipitation, among others). The model was found to have acceptable model performance based on the average training and test Area Under the Curve (AUC) values of 0.76 and 0.73. A 1 km x 1 km Falcata suitability map was generated using the model. The map shows that 12% of the region has high suitability, while 23% and 30% have moderate and low suitabilities. On the other hand, 35% of the region was not suitable for Falcata plantation establishment.
Content may be subject to copyright.
The 42nd Asian Conference on Remote Sensing (ACRS2021)
22-24th November, 2021 in Can Tho University, Can Tho city, Vietnam
USING MAXENT IN FINDING SUITABLE LOCATIONS FOR ESTABLISHING
FALCATA TREE PLANTATIONS IN CARAGA REGION, MINDANAO, PHILIPPINES
Jojene R. Santillan1,2,3, Arnaldo C. Gagula1,2,3, and Meriam Makinano-Santillan1,2,3
1Caraga Center for Geo-informatics, Caraga State University, Butuan City, Philippines
2Department of Geodetic Engineering, College of Engineering and Geosciences, Caraga State University,
Butuan City, Philippines
3Industrial Tree Plantation Research and Innovation Center, Caraga State University,
Butuan City, Philippines
Email: jrsantillan@carsu.edu.ph, acgagula@carsu.edu.ph, mmsantillan@carsu.edu.ph
KEY WORDS: Falcata, Suitability Mapping, MaxEnt, Maximum Entropy, Caraga Region, Mindanao, Philippines
ABSTRACT: The Caraga Region in Mindanao, Philippines, is considered a significant contributor in log
production, specifically due to Falcata (Paraserianthes falcataria) plantations. Over 80% of the country's Falcata
log production came from Caraga Region in 2019. Among the challenges faced by the tree-growers is finding a
suitable location for the establishment of new plantations. We used MaxEnt, a machine learning Species
Distribution Modeling (SDM) based on Maximum Entropy principles, for this study's Falcata plantation suitability
modeling and mapping. This approach used 2,125 Falcata location points distributed in the region, biophysical
factors (i.e., Elevation, Slope, Aspect, and the like), and bioclimatic factors (i.e., Annual Mean Temperature,
Isothermality, and Annual Precipitation, among others). The model was found to have acceptable model
performance based on the average training and test Area Under the Curve (AUC) values of 0.76 and 0.73. A 1 km x
1 km Falcata suitability map was generated using the model. The map shows that 12% of the region has high
suitability, while 23% and 30% have moderate and low suitabilities. On the other hand, 35% of the region was not
suitable for Falcata plantation establishment.
1. INTRODUCTION
Falcata (Paraserianthes falcataria (L.) Nielsen) (Fig. 1.) is a preferred tree species in industrial tree plantations
because of its fast growth, grows in various soil textures, and is acceptable for wood production (Krisnawati H. et
al., 2011). The 2019 Philippine Forest Statistics of Forest Management Bureau (FMB) of the Department of
Environment and Natural Resources (DENR-FMB, 2020) showed that the Caraga Region contributed 64.12% of the
country's total log production and 87.89% of the Falcata total log production. It is a large tree that grows up to 40
meters in height and 20 to 100 dbh (Alipon, M. et al., 2017). Woods of Falcata is an excellent source to produce
pulpwood and plywood (Krisnawati H. et al., 2011) and the most preferred raw materials for wood (Alipon, M. et
al., 2017).
Figure 1. Falcata Plantation in Caraga Region, Mindanao, Philippines.
An earlier study was conducted to assess the suitability of Falcata plantations in the Caraga Region (Santillan et. al.,
2021). This study utilized MaxEnt, a machine learning Species Distribution Model (SDM), with bioclimatic
datasets (i.e., BIO 2, BIO 4, BIO 6, BIO 7, BIO 14, BIO 15, BIO 17, BIO 18), biophysical datasets (i.e., elevation,
The 42nd Asian Conference on Remote Sensing (ACRS2021)
22-24th November, 2021 in Can Tho University, Can Tho city, Vietnam
aspect, slope, land cover, soil type, and land cover), solar radiation, and wind speed. In selecting the environmental
variables, this study assessed the collinearity of each variable based on the result of Principal Components Analysis
(PCA) of the raster files corresponding to the variables. From this study, 925,700 hectares in the region were found
to be "suitable" for establishing Falcata plantations (Santillan et al., 2021). According to (Feng, X. et al., 2019),
there were two types of collinearity influence in regression type models. First is the effect on model training caused
by the degree of predictor collinearity. The other is the effect on model transfer caused by differences in the
correlation structure of predictor variables between training and testing regions. Therefore, deciding what
collinearity methods are essential.
In this work, we conducted Falcata suitability modeling and mapping using MaxEnt. Specifically, we utilized the
extracted pixel values of bioclimatic, biophysical, solar radiation, and wind speed corresponding to the spatial
locations of Falcata species in the collinearity analysis to find the final set of variables for inclusion in the model.
2. MATERIALS AND METHODS
2.1 Study Area
The Caraga Region (Fig. 2) has long been considered the biggest producer of significant forest products in the
Philippines. It has been dubbed as "the timber corridor of the Philippines." It has a total land area of 1,884,697 ha,
of which 71% are classified as forestlands, and 29% are certified alienable and disposable. The region is
characterized by mountainous areas and flat and rolling lands, with its most productive agricultural area along the
Agusan River Basin. It has a Type II climate, with no pronounced wet and dry season; heavy rains are usually
experienced from November to February (Wikipedia, 2021).
Figure 2. Map of Caraga Region, Mindanao, Philippines. Presented in the map are locations of Falcata Plantations
used in MaxEnt modeling.
2.2 Falcata Location Data
The location data were gathered from the map of confirmed Falcata species derived using Sentinel-2 image using
Maximum Likelihood Classification with greater than 90% accuracy and refined through high resolution google
earth images. Only the Falcata with at least 1 hectare was considered with 1,224 m distance to each other. The
"Random Points Generator" extension in ArcView software was used to select points from Falcata plantations'
centroids randomly. This resulted in 1,065 points which were saved as "training" points, and another 1,060 points
The 42nd Asian Conference on Remote Sensing (ACRS2021)
22-24th November, 2021 in Can Tho University, Can Tho city, Vietnam
were saved as "validation" points (Table 1). A comma space value (CSV) file was generated containing the
Universal Transverse Mercator (UTM 51) WGS 1984 grid coordinates of each point used in MaxEnt.
Table 1. The total number of Falcata location points in the Caraga Region.
Province Name
Training Points
Validation Points
Total Number of Points
Agusan del Norte
208
206
414
Agusan del Sur
557
555
1,112
Surigao del Norte
33
33
66
Surigao del Sur
261
262
523
Dinagat Islands
6
4
10
Total
1,065
1,060
2,125
2.3 Environmental Variables
There are 26 environmental variables identified, including 19 bioclimatic variables, solar radiation, wind speed,
elevation, slope, aspect, soil type, and land cover (Figure 3). The 19 bioclimatic variables (BIO 01-BIO 19),
including the solar radiation and wind speed, were downloaded from
https://www.worldclim.org/data/worldclim21.html. It has a spatial resolution of approximately 1 km and is in
GeoTIFF format, derivatives of the WorldClim version 2.1 climate data for 1970-2000. The biophysical variables
include the Digital Elevation Model that was derived from SAR interferometry with a spatial resolution of 5 meters,
soil data that was obtained from the Department of Agriculture - Bureau of Soils and Water Management (DA-
BSWM), and landcover data from the Philippines' National Mapping and Resource Information Authority
(NAMRIA). The DEM was resampled into 1 km spatial resolution to be consistent with the 19 bioclimatic variables,
solar radiation, and wind speed. The resampled DEM becomes the base layer to generate aspect and slope data
using ArcGIS 10.8 software. The soil and land cover data are also converted to raster data with 1 km spatial
resolution for consistency. The data layers were layer-stacked using ArcGIS 10.8 software to ensure the exact
spatial resolution (1 km) and coverage and raster dimensions. Each layer was then exported to ASCII (*.asc) format
in preparation for MaxEnt modeling.
Figure 3. Example of environmental variables used in MaxEnt Modeling. These 6 variables have the highest percent
contribution on determining suitable sites for Falcata plantations.
The 42nd Asian Conference on Remote Sensing (ACRS2021)
22-24th November, 2021 in Can Tho University, Can Tho city, Vietnam
2.4 Environmental Variable Selection
Collinearity of the environmental variables needs to be quantified so that only variables that are not collinear with
other variables must be considered in the Species Distribution Modeling (SDM). Pixel values were extracted from
the 26 environmental variables using the locations of confirmed Falcata plantations. Afterward, the values
underwent statistical analysis (correlation analysis). The Pearson's correlation coefficients (r) between the variables
were calculated to facilitate collinearity analysis. Variables with a value of r 0.7 are considered correlated, and
only one of them should be selected and included in the MaxEnt model. The variable selection reported on (Garcia,
K. et al., 2013) was used to minimize the finalizing of the environmental variables. This method underwent initial
runs of all the environmental variables. Based on its outputs, only one variable (variable with highest percent
contribution among others) was chosen from the variables with high correlation values in the final MaxEnt Model
(Garcia, K. et al., 2013). As a result, only 14 out of 26 environmental variables were considered (Table 2).
Table 2. List of environmental variables. Highlighted variables are those included in the final MaxEnt model.
BIO1
Annual Mean Temperature
BIO2
Mean Diurnal Range (Mean of monthly (max temp - min temp))
BIO3
Isothermality [(BIO 02/BIO 07) x100)]
BIO4
Temperature Seasonality (standard deviation x 100)
BIO5
Maximum Temperature of Warmest Month
BIO6
Minimum Temperature of Coldest Month
BIO7
Temperature Annual Range (BIO 05 BIO 06)
BIO8
Mean Temperature of Wettest Quarter
BIO9
Mean Temperature of Driest Quarter
BIO10
Mean Temperature of Warmest Quarter
BIO11
Mean Temperature of Coldest Quarter
BIO12
Annual Precipitation
BIO13
Precipitation of Wettest Month
BIO14
Precipitation of Driest Month
BIO15
Precipitation Seasonality (Coefficient of Variable)
BIO16
Precipitation of Wettest Quarter
BIO17
Precipitation of Driest Quarter
BIO18
Precipitation of Warmest Quarter
BIO19
Precipitation of Coldest Quarter
SRAD
Solar radiation
WIND
Wind speed
LCOV
Land-cover type
SOIL
Soil type
ELEV
Elevation
ASPECT
Aspect
SLOPE
Slope
2.5 MaxEnt Modeling
MaxEnt Version 3.4.1, downloaded from https://biodiversityinformatics.amnh.org/open_source/maxent/, was used
in this study. In this study, the model runs in 15 replicates with 5000 iterations to have adequate time for
convergence. Each iteration uses a random partition of the presence data through subsample, in which 75% was
used for modeling, and 25% was used for testing. The convergence threshold used was 0.00001 and 10,000 for
maximum background points. Random seeding, as well as jack-knife test for variable importance, were also enabled.
The model was evaluated using the area Under the Curve (AUC) statistic, calculated from the Receiver Operating
Characteristic (ROC) based on training and testing data. The following differentiation of performance levels was
used: excellent (>0.9), good (0.80.9), accepted (0.70.8), poor (0.60.7), and unsatisfactory (<0.6) (Rojas-Briceño,
The 42nd Asian Conference on Remote Sensing (ACRS2021)
22-24th November, 2021 in Can Tho University, Can Tho city, Vietnam
N. B., et al., 2020). This is the range of the value of the final potential species distribution map, which was
reclassified into four branches of potential habitat: "high" (>0.6), "moderate" (0.40.6), and "low" (0.20.4), and
'not suitable' (<0.2) (Zhang, K. et al., 2019).
3. RESULTS AND DISCUSSIONS
3.1. MaxEnt Model Results
Based on the result, area Under the Curve (AUC) values range from 0.75-0.76 and 0.71-0.74 for training and testing
presence data. The average training and test AUC values are 0.76 and 0.73, respectively, indicating acceptable
MaxEnt model performance.
Figure 4 shows the percent contribution of each environmental variable in the MaxEnt Model. It reveals that BIO
04 (Temperature Isothermality) gains the highest percentage contribution with 32.9%. On the other hand, ASPECT
is the least contributing variable with 0.2%.
Figure 4. Percent Contributions of environmental variables to the Falcata MaxEnt model.
Marginal curves (Figure 5 and Figure 6) were generated to show how each environmental variable affects the
MaxEnt prediction. The curves show how the predicted probability of presence changes as each environmental
variable varies, keeping all other environmental variables at their average sample value. It shows that for BIO
4which has the highest percent contribution, higher logistic output was obtained during the lower temperature. It
means that existing Falcata plantations in the region can be found in areas where temperature seasonality has lower
values.
The 42nd Asian Conference on Remote Sensing (ACRS2021)
22-24th November, 2021 in Can Tho University, Can Tho city, Vietnam
Figure 5. The marginal response curves of BIO 4, land cover, BIO 15, BIO 14, and Slope.
Figure 6. The marginal response curves of soil texture, wind speed, BIO 6, elevation, BIO 17, solar radiation, BIO 7,
BIO 18, and aspect.
3.2 Falcata Suitability Map
Figure 7 shows the Falcata Suitability Map generated using MaxEnt. The result shows 212,985 hectares (12%) of
the Caraga Region was classified as high suitable, 421,747 hectares (23%) were medium suitable, 538,951 hectares
(30%) were low suitable, and 619,380 (35%) were classified as unsuitable. In terms of area, the total suitable areas
in the region are 1,173,683 hectares.
Figure 8 shows the percentage comparison of Caraga Region provinces for Falcata suitability based on its land area
in terms of statistics per province. It shows that Agusan del Sur has more "High Suitable" areas, followed by
Agusan del Norte and Surigao del Sur with 129,864 hectares, 47,839 hectares, and 35,282 hectares, respectively. In
contrary, Agusan del Sur has more “Not Suitable” areas followed by Surigao del Sur, Surigao del Norte, Agusan del
Norte, and Dinagat Islands containing 238,305 hectares, 148,125 hectares, 92,522 hectares, 90,522 hectares, and
49,906 hectares, respectively.
The 42nd Asian Conference on Remote Sensing (ACRS2021)
22-24th November, 2021 in Can Tho University, Can Tho city, Vietnam
Figure 7. The Falcata Suitability Map for Caraga Region, Mindanao, Philippines.
Figure 8. The comparison of Caraga Region provinces in terms of Falcata suitability. Values are in hectares.
Figure 9 shows the comparison of the total suitable areas in each province to the actual total areas of Falcata
The 42nd Asian Conference on Remote Sensing (ACRS2021)
22-24th November, 2021 in Can Tho University, Can Tho city, Vietnam
plantations. It shows that Agusan del Sur has gains the highest suitable areas, followed by Surigao del Sur, Agusan
del Norte, Surigao del Norte, and Dinagat Islands. In addition, Caraga Region provinces have only less than 1% of
the total suitable areas planted with Falcata. This implies that the region still has more areas that are potential in
growing Falcata plantations.
Figure 9. The comparison of the total suitable areas with the total actual area of Falcata plantations in the Caraga
Region. Values are in hectares.
The results differ when applying another collinearity analysis approach in the previously conducted study (Santillan,
J. et al., 2021). Using the extracted values of Falcata points instead of data layers in collinearity analysis, BIO 02
(Mean Diurnal Range) has been removed from the final environmental variables for MaxEnt modeling and the total
suitable areas in the region increased by 247, 983 hectares.
4. CONCLUSION
In this study, we conducted suitability analysis for Falcata in Caraga Region using MaxEnt. The study has
successfully identified areas for Falcata plantations with training and test AUC values of 0.76 and 0.73, respectively.
BIO 4 (Temperature Seasonality) gains the highest percentage contribution with 32.9%, and ASPECT has the
lowest percent contributions among others with 0.2%. The suitability map revealed 1,173,683 hectares of land
where Falcata plantations can be potentially established.
ACKNOWLEDGEMENTS
This work is an output of "Project 1. Development of a Geodatabase of Industrial Tree Plantations (ITP) in Caraga
Region Using Remote Sensing and GIS" under the Niche Centers in the Regions for R&D (NICER) Program:
Industrial Tree Plantations Research and Innovation Center (ITPS) for Upgrading the Wood-Based Industry, funded
and supported by the Philippines Department of Science and Technology (DOST) and the Philippine Council for
Agriculture, Aquatic and Natural Resources Research and Development (PCAARRD).
REFERENCES
Alipon, M., Alcachupas, P., Bondad, Elvina, Cortiguerra, E., 2016. Assessing the Utilization of Falcata [Falcataria
moluccana (Miq.) Barneby & J. W. Grimes] for Lumber Production. Philippine Journal of Science 145 (3): 225-235.
Feng, X., Park, D., Liang, Y., Pandey, R., Papes, M., 2019. Collinearity in ecological niche modeling: Confusions
and challenges. Ecology and Evolution. Volume9, Issue1, Pages 10365-10376. https://doi.org/10.1002/ece3.5555
Garcia, K., Lasco, R., Ines, A., Lyon, B., Pulhin, F. 2013. Predicting geographic distribution and habitat suitability
The 42nd Asian Conference on Remote Sensing (ACRS2021)
22-24th November, 2021 in Can Tho University, Can Tho city, Vietnam
due to climate change of selected threatened forest tree species in the Philippines. Applied Geography, 44, 12-22.
Krisnawati, H., Varis, E., Kallio, M., Kanninen, M., 2011. Paraserianthes falcataria (L.) Nielsen; Ecology,
silviculture and productivity. Center for International Forestry Research, Bogor, Indonesia.
Rojas-Briceño, N. B., Cotrina-Sánchez, D. A., Barboza-Castillo, E., Barrena-Gurbillón, M. Á., Sarmiento, F. O.,
Sotomayor, D. A., Salas-López, R., 2020. Current and Future Distribution of Five Timber Forest Species in
Amazonas, Northeast Peru: Contributions towards a Restoration Strategy. Diversity, 12(8), 305.
Santillan, J., Gagula, A., Santillan, M., 2021. Machine Learning Approach for Tree Plantation Suitability Mapping.
2021 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), pp. 6865-6868.
Zhang, K., Zhang, Y., Tao, J., 2019. Predicting the Potential Distribution of Paeonia veitchii (Paeoniaceae) in China
by Incorporating Climate Change into a Maxent Model. Forests, 10, 190; doi:10.3390/f10020190
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Forest and land degradation is a serious problem worldwide and the Peruvian National Map of Degraded Areas indicates that 13.78% (177,592.82 km2) of the country’s territory is degraded. Forest plantations can be a restoration strategy, while conserving economically important species affected by climate change and providing forestry material for markets. This study modelled the species distribution under current conditions and climate change scenarios of five Timber Forest Species (TFS) in the Amazonas Department, northeastern Peru. Modelling was conducted with Maximum Entropy (MaxEnt) using 26 environmental variables. Of the total distribution under current conditions of Cedrelinga cateniformis, Ceiba pentandra, Apuleia leiocarpa, Cariniana decandra and Cedrela montana, 34.64% (2985.51 km2), 37.96% (2155.86 km2), 35.34% (2132.57 km2), 33.30% (1848.51 km2), and 35.81% (6125.44 km2), respectively, correspond to degraded areas and, therefore, there is restoration potential with these species. By 2050 and 2070, all TFS are projected to change their distribution compared to their current ranges, regardless of whether it will be an expansion and/or a contraction. Consequently, this methodology is intended to guide the economic and ecological success of forest plantations in reducing areas degraded by deforestation or similar activities.
Article
Full-text available
Ecological niche models are widely used in ecology and biogeography. Maxent is one of the most frequently used niche modeling tools, and many studies have aimed to optimize its performance. However, scholars have conflicting views on the treatment of predictor collinearity in Maxent modeling. Despite this lack of consensus, quantitative examinations of the effects of collinearity on Maxent modeling, especially in model transfer scenarios, are lacking. To address this knowledge gap, here we quantify the effects of collinearity under different scenarios of Maxent model training and projection. We separately examine the effects of predictor collinearity, collinearity shifts between training and testing data, and environmental novelty on model performance. We demonstrate that excluding highly correlated predictor variables does not significantly influence model performance. However, we find that collinearity shift and environmental novelty have significant negative effects on the performance of model transfer. We thus conclude that (a) Maxent is robust to predictor collinearity in model training; (b) the strategy of excluding highly correlated variables has little impact because Maxent accounts for redundant variables; and (c) collinearity shift and environmental novelty can negatively affect Maxent model transferability. We therefore recommend to quantify and report collinearity shift and environmental novelty to better infer model accuracy when models are spatially and/or temporally transferred. Excluding highly correlated variables does not affect Maxent model performance. Model transfer may lead to novel environment and collinearity shift, while both can negatively affect model performance.
Article
Full-text available
A detailed understanding of species distribution is usually a prerequisite for the rehabilitation and utilization of species in an ecosystem. Paeonia veitchii (Paeoniaceae), which is an endemic species of China, is an ornamental and medicinal plant that features high economic and ecological values. With the decrease of its population in recent decades, it has become a locally endangered species. In present study, we modeled the potential distribution of P. veitchii under current and future conditions, and evaluated the importance of the factors that shape its distribution. The results revealed a highly and moderately suitable habitat for P. veitchii that encompassed ca. 605,114 km2. The central area lies in northwest Sichuan Province. Elevation, temperature seasonality, annual mean precipitation, and precipitation seasonality were identified as the most important factors shaping the distribution of P. veitchii. Under the scenario with a low concentration of greenhouse gas emissions (RCP 2.6), we predicted an overall expansion of the potential distribution by 2050, followed by a slight contraction in 2070. However, with the scenario featuring intense greenhouse gas emissions (RCP 8.5), the range of suitable habitat should increase with the increasing intensity of global warming. The information that was obtained in the present study can provide background information related to the long-term conservation of this species.
Article
Lumber recovery and grades, mechanical properties (static bending, hardness, toughness, shear, compression parallel and perpendicular-to-grain), and cost-benefit in utilizing falcata [Falcataria moluccana (Miq.) Barneby & J. W. Grimes] at different ages from different sites were determined and evaluated. The aim is to study the feasibility of harvesting falcata at a younger age (four to eight years old) instead of the current cutting age of 8 to 12 years old, and help widen the raw material base of the local wood-based industries. The materials were collected from three sites in Caraga region: Prosperidad, Agusan del Sur (Site 1), Nong-nong, Butuan City (Site 2) and Las Nieves, Agusan del Norte (Site 3). Standard procedures in sampling and properties testing were followed. Lumber quality or grade yield per log per species was evaluated based on the National Hardwood Lumber Association (NHLA) Standards. Lumber recovery and grades improved from four to eight years old, significantly highest in Site 1. The highest grade recovered was No.1Common (No.1C) mostly from the 8 and 6-year-old trees. Log samples in the younger age class and with small diameter (4-year-old trees) yielded mostly No.3 Common (No.3C) boards. The mechanical properties classification of the species was the same regardless of age, sites as well as diameter across ages (Class V - Low Strength). The wood can be used for purposes where strength is not a critical requirement. Instead of waiting until they are eight years old, falcata trees may be cut at four to six years old as far as mechanical properties are concerned. Cost analysis showed it is not viable to harvest falcata trees with diameter of 16 cm and below. It may be profitable to harvest falcata logs with 16 cm and above diameter (attained even by four-year-old trees from Site 1) if selling price is Php 2,000/m³. © 2016, Science and Technology Information Institute. All rights reserved.
Predicting geographic distribution and habitat suitability The 42 nd Asian Conference on Remote Sensing (ACRS2021) 22-24 th November, 2021 in Can Tho University, Can Tho city, Vietnam due to climate change of selected threatened forest tree species in the Philippines
  • K Garcia
  • R Lasco
  • A Ines
  • B Lyon
  • F Pulhin
Garcia, K., Lasco, R., Ines, A., Lyon, B., Pulhin, F. 2013. Predicting geographic distribution and habitat suitability The 42 nd Asian Conference on Remote Sensing (ACRS2021) 22-24 th November, 2021 in Can Tho University, Can Tho city, Vietnam due to climate change of selected threatened forest tree species in the Philippines. Applied Geography, 44, 12-22.