spatio temporal landscape utilization using technological innovations addressing one health and sustainability case studies in the philippines

Land use/land-cover change and land surface temperature in Metropolitan Manila Philippines using Landsat imagery……….……..……….27 3.1.1.. LIST OF ABBREVIATIONS CVI Coastal Vulnerability Ind

Research Rationale

Technologies such as remote sensing and species distribution modeling have become increasingly important tools for addressing global challenges such as climate change and biodiversity conservation Remote sensing techniques, such as the utilization of satellite imagery, allow for the monitoring and mapping of the Earth’s surface through land cover change studies On the other hand, modeling techniques such as species distribution modeling, provides insights into the potential impact of the changing climate to a species which can then lead to effective management strategies and adaptive coastal management plans With these technologies, this study aims (i) to investigate the impact of of land-use/land-cover change on land surface temperature in Metropolitan Manila, Philippines using Landsat imagery, (ii) to predict the habitat suitability of the endangered Tamaraw (Bubalus mindorensis) in Mindoro, Philippines using species distribution modelling approach and (iii) to conduct a coastal vulnerability assessment using the (Integrated Valuation of Ecosystem Services and Tradeoffs) InVEST software in La Union, Philippines that target the United Nations Sustainable Development Goals and One Health Agenda

These three different approaches using remote sensing and modeling, are significant because they address pressing environmental issues in the Philippines For instance, Metropolitan Manila, a highly urbanized area, is facing rapid land-use/land- cover change throughout the years, which can have significant impacts on local climate, especially the land surface temperature The understanding of patterns on Metropolitan Manila’s landscape can aid the development of environmental policies and management plans to ensure the sustainability of urban areas for the betterment of the quality of urban life On the other hand, the Tamaraw is a critically endangered species that is endemic to the island of Mindoro, Philippines Habitat loss and fragmentation have led to a decline in the Tamaraw population that was once common in the island of Mindoro Species distribution models that integrate various environmental factors with known presence data can provide a comprehensive understanding of the factors affecting these species to aid the ongoing conservation and management efforts Finally, coastal vulnerability assessment using the InVEST software in La Union, Philippines that will provide information on the exposure, sensitivity, and adaptive capacity of coastal areas The results of this study can guide policymakers and stakeholders in implementing measures to reduce the vulnerability of coastal communities to climate change and promote sustainable coastal management.

Research Questions and Hypotheses

The use of remote sensing and modeling in these independent research topics can contribute to the One Health Approach and Sustainable Development Goals The One Health is a holistic approach that recognizes the interconnectedness of human, animal, and environmental health The land-use/land-cover change in Metropolitan Manila, habitat suitability of the Tamaraw in Mindoro, and coastal vulnerability assessment in

La Union, Philippines are all important research topics that have implications for human and animal health, as well as the environment These research topics are relevant to the following Sustainable Development Goals (SDGs): SDG 11 (Sustainable Cities and Communities), SDG 13 (Life on Land), and SDG 15 (Climate Actions) Addressing the impacts of land-use/land-cover change in Metropolitan Manila, predicting the habitat suitability of the Tamaraw, and assessing the coastal vulnerability in La Union contribute to achieving these SDGs by promoting sustainable urbanization, conserving biodiversity, and enhancing coastal resilience Overall, these research topics are crucial for addressing environmental challenges and promoting sustainable development in the Philippines

1.2.1 Assessing the land use/land-cover change and land surface temperature in Metro Manila, Philippines

Background/Hypothesis: Since vegetation influence the temperature of the land, higher temperature is expected in plant-free landscapes, thus, this research would like to answer the following questions:

Research questions: a) What will be the change of landscape and land surface temperature (LST) in Metro Manila, Philippines over the years? b) What will be the correlation of the land surface temperature to vegetation and urbanization indices to Metro Manila, Philippines?

1.2.2 Habitat suitability mapping of Tamaraw in Mindoro, Philippines

Background: Since modern technologies such as species distribution modeling is capable of providing baseline information and prediction of Tamaraw’s distribution in Mindoro, Philippines, hence, this proposed research study would like to answer the following question:

Research Questions: a) What are the possible statistically predicted areas where Tamaraw can potentially be found using a machine-learning approach? b) What will be the change of the distribution of Tamaraw under futuristic climate change scenarios?

1.2.3 Modeling of coastal vulnerability in La Union, Philippines

Background/Hypothesis: Since natural habitats such as mangroves provide protection in coastal areas from storm surges, and erosion, it is expected that the areas are relatively less exposed to natural hazards

Research Questions: a) Which areas in La Union are vulnerable to natural hazards using the InVEST modeling approach? b) What will be the change in vulnerability in the no habitats scenario?

This research has three independent case studies utilizing mapping and modeling approaches to address One Health Approach and sustainability in unique landscapes of the Philippines The first case study in Metropolitan Manila (MM), Philippines aims:

● to update the maps addressing urban heat islands LST, Normalized Difference Vegetation Index (NDVI), Normalized Difference Built-up Index (NDBI), and Land use/ Land-cover (LULC) maps over the last decade (from 2001 – 2019)

● to verify the possible relationship on the effect of decreasing green spaces and increasing impervious surfaces in MM with regards to the land surface temperature, the values for LST was correlated with the values of NDVI and NDBI

● To assess how the land use and land cover changes during the last 18 years have influenced the rising heat in MM, the LULC from the year 2001 and

2019 was compared with the LST maps of MM

Secondly, the following objectives were constructed to answer the research questions raised in the second case study:

● to identify suitable areas of Bubalus mindorensis or Tamaraw in Mindoro,

● to assess the which environmental factors that affect the distribution of

Bubalus mindorensis or Tamaraw in Mindoro, Philippines

Lastly, the final research topic aims to:

● to determine the vulnerability of the coastal areas of La Union, Philippines

● to compare the coastal vulnerability between two coastal scenarios (with vs without natural habitat)

This research has three different case studies in the Philippines which aimed to: (i) assess urban heat islands LST over the last decade in Metropolitan Manila, Philippines, (ii) predict the distribution of Tamaraw in Mindoro, Philippines using machine learning under the changing climate scenarios and lastly, (iii) to assess the vulnerability of of the coastal areas of La Union, Philippines using the InVEST software Moreover, here are some limitations for each case study First, most satellite images for

MM have a huge percentage of cloud cover which could hinder accuracy A preprocessing step was done to ensure the image quality for further analysis Secondly, the presence data for Tamaraw has more than 1000 records with Tamaraw being less than only 500 in the wild To address this issue but still retain the quality of the occurrence data, a spatial thinning was done to avoid bias in the SDM Lastly, the coastal vulnerability assessment using InVEST software does not measure the quality of the habitats and only the protection based on their presence Further on-site coastal evaluation should be conducted in La Union in order to verify and validate the inputs Due to the limitation of time and resources, that particular phase of the study was not done, however, a proposal for that next stage is underway

For the sole purpose of understanding words, here are the list of terms used for this study and their respective meanings

Anthropogenic variables refers to the abiotic variables such as land-use/land cover, distance to village that significantly affect Tamaraw's distribution

Climatic variables refers to the 19 abiotic variables such as temperature and precipitation that may affect species distribution

Land-use/Land Cover refers to the classification of human activities and natural elements on a landscape

Land-use/Land Cover change refers to the change of the landscape in the given time period

Land Surface Temperature refers to the brightness temperature of the land surface Landsat Enhanced Thematic Mapper+ refers to the seventh satellite launched by the

Normalized Built-up Vegetation Index refers to another graphical indicator to emphasize the manufactured built-up areas

Maximum Likelihood refers to the algorithm that is based from the Bayes’ theorem of decision making and class sample being normally distributed

Normalized Difference Vegetation Index refers to the graphical indicator to assess if an area contains live green vegetation or not

Species distribution modeling refers to the process or technique that utilizes species occurrence data and environmental variables that are stacked and modeled in an algorithm to produce a statistically predicted distribution map

Topographic variables refers to the abiotic variables such as elevation, slope and aspect

Urban Heat Islands refers to the area in metropolitan that experiences higher temperature than in nearby rural areas due to anthropogenic activities

Urbanization refers to the population shift from rural to urban areas which often associated with bad effects on quality of life and the environment

Scope and Limitations

MM have a huge percentage of cloud cover which could hinder accuracy A preprocessing step was done to ensure the image quality for further analysis Secondly, the presence data for Tamaraw has more than 1000 records with Tamaraw being less than only 500 in the wild To address this issue but still retain the quality of the occurrence data, a spatial thinning was done to avoid bias in the SDM Lastly, the coastal vulnerability assessment using InVEST software does not measure the quality of the habitats and only the protection based on their presence Further on-site coastal evaluation should be conducted in La Union in order to verify and validate the inputs Due to the limitation of time and resources, that particular phase of the study was not done, however, a proposal for that next stage is underway.

Definition of terms

Literature Review

Land Use/land-cover change and land surface temperature

Satellite images can provide data to address these gaps in MM, specifically from Landsat led by the joint program of NASA and USGS The Landsat program has been running around since 1972 with the launch of the first Landsat 1 satellite with a Multi- spectral Scanner (MSS) In 1999, the Landsat 7 was launched with the Enhanced Thematic Mapper Plus (ETM+) with these new features such as (i) a panchromatic band with 15m spatial resolution (ii) on-board, full aperture, 5% absolute radiometric calibration (iii) a thermal infrared channel with 60m spatial resolution (iv) an on-board data recorder Due to the accuracy of the derived data from Landsat 7 to ground measurements, it is called the most stable observation instrument ever placed in orbit at the time The images captured by the Landsat satellites are widely available on their website (https://earthexplorer.usgs.gov/) for free Interestingly, there are many parameters that can be provided using this remote sensing imagery from Landsat

One of which is Land Surface Temperature (LST) which measures the temperature on earth’s surface by combining vegetation and bare soil temperature (Goward, 1981) The LST in urban areas are affected by several factors such as building materials, wind speed, anthropogenic activities and green spaces (Ren et al., 2016; Xia et al., 2016; Yang et al., 2017) LST is estimated using Top-of-Atmosphere brightness temperature from different satellites and weather balloons Estimating LST using satellite imagery are mainly utilizing two approaches: (i) Mono-Windows/Single Channel algorithms (Zhou et al., 2001; Zhi-hao et al., 2003) which utilizes satellites that have a single thermal band e.g Landsat 4-5 Thematic Mapper (TM) and Landsat 7 Enhanced Thematic Mapper + (ETM+), and (ii) Split-Window (SW) algorithms are used in sensors that have multiple thermal bands e.g., Landsat 8 TIRS, MODIS and ASTER (Dwivedi & Khire, 2018; Wang et al., 2019; Zhang et al., 2019) The estimation of LST has been widely used to assess how the growth of urban heat island takes place in some major cities in the world (Texas, USA, Streutker, 2003; Hong Kong Liu & Zhang, 2011;

Shijiazhuang, China, Liu et al., 2015) In addition, LST has also been linked to many several parameters, namely, Normalized Difference Vegetation Index (NDVI), Normalized Difference Built-up Index (NDBI), and land-use/land cover (LULC), that utilizes data gathered from the satellites Specifically, finding the relationship between LST with NDVI, NDBI, and LULC provide maps where green spaces can be allocated to combat the excess heat produced by the UHI effect (Grover & Singh, 2015)

NDVI is one of the most widely used indices for vegetation As satellites capture different light wavelengths, the density and health of vegetation can be calculated using mathematical formulas (Gandhi et al., 2015) Healthy and dense vegetation tends to reflect a huge portion of the near-infrared light The NDVI values usually range from +1 to -1 Forests, shrubs, grasslands, and crops mostly yield high NDVI values Snow, barren land, built-up, and water bodies yield the lowest NDVI values as they do not reflect NIR light waves back to the satellites Next, NDBI can be used to detect impervious surfaces such as roads, buildings, roofs, etc which reflect shortwave infrared (SWIR) light waves back to the satellites Similar to NDVI, the NDBI values range from +1 to -1 with impervious surfaces having the highest value Lastly, the land-use/land cover (LULC) mapping is a machine-learning approach that classifies the map in the study area that can monitor urban growth (Mohan et al., 2014; Zope et al., 2016; Deng et al., 2019) and more recently, associates the landscape changes with temperature (Palafox‐Juárez et al., 2021)

2.1.2 UHI studies in Metro Manila, Philippines

Assessment of the urban heat island (UHI) in Metro Manila utilizing remote sensing has been a growing field during the past decades In 2006, the paper of Tiangco et al., reported the derivation of land surface temperature that aids in quantifying the UHI effect in MM during the night using the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) In that spatial study the following were suggested: (i) thermal images revealed that UHI intensity is at highest in the central part of MM and (ii) an inverse pattern of relationship between NDVI and temperature A more recent study by Estoque et al (2017) that used an ensemble cross sectional design and numerous spatial metrics-based approach to evaluate the UHI phenomenon in three Southeast Asian megacities, including Metro Manila, revealed the strong correlation of impervious surface and availability of green space on the variability of land surface temperature Estoque et al (2020) has also used remotely sensed data together with socioecological indicators to provide health risk assessment among 139 Philippine major cities These research papers served then as the foundation of this research study that was conducted to examine and update in a much more detail several parameters that have been done by these previous studies such as vegetation and built-up indices in relationship with the land surface temperature using Metro Manila as a case study The additive value of this research is to correlate land use/land cover change with land surface temperature to specifically address if urban sprawl may influence the increase in warming in Metro Manila.

Predicting Habitat Suitability of Tamaraw (Bubalus mindorensis) under

The Philippines, known as a biodiversity hotspot (Gather & Rocha, 2013), is home to numerous unique and threatened species facing various anthropogenic activities that pose significant challenges to their survival (Koh et al., 2013; Von et al., 2017) Among these species is the critically endangered Tamaraw, an endemic wild bovine found exclusively on the island of Mindoro (Boyles et al., 2016) A century ago, estimates reached 10,000 individuals of tamaraws inhabiting the island from its coastal lowlands to the highest peaks; traditional hunting was thought of as a probable cause of the decrease in numbers as these are source of food sustenance by local communities (Kuehn, 1986, Custodio et al., 1996) However, with the establishment of cattle ranches, hunting has become unprofitable, and habitat disturbance has emerged as a more prevalent cause (Erni, 2006) And in 1930, large plantations (Oliver and Heany, 1997) and cattle ranching grew rampant in the island (Postma, 1974) Thus introducing the rinderpest epidemic that infected many of the island’s bovines (Maala, 2001) The epidemic decimated the population, by 1969 it was estimated that there were less than

100 individuals of tamaraw left (UNDP-BioFin, 2019)

Recognizing the urgent need for conservation, the Philippine government took action to protect the remaining Tamaraw population by establishing the Tamaraw Conservation Program (TCP) under the Department of Environment and Natural Resources (DENR) It was initially established as a Presidential Committee for the Conservation of the Tamaraw in 1979 under Executive Order No 544 on July 9, 1979 Later, management was absorbed to the Protected Areas and Wildlife Bureau (now Biodiversity Management Bureau) of the Department of Environment and Natural Resources with the institution of the "Administrative Code of 197," through Executive Order No 292 on July 25, 1987; it is now known as the Tamaraw Conservation Program DENR-TCP started conducting annual surveys of point count in 2000 at the Mount Iglit- Baco Natural Park (MIBNP) to estimate the number of Tamaraw individuals in the wild as part of its comprehensive conservation program; the rest of the tamaraw sites are monitored by regular patrolling or specific research activities (e.g camera traps) Recent estimates of the tamaraw population are around 200 individuals recorded at the MIBNP (Schutz et al., 2023)

Currently, due to their small population size and restricted range, the ecological role of tamaraw is in question However, their survival in their natural habitat is crucial for maintaining the health and integrity of the ecosystem as Tamaraw grazing and wallowing behavior influences the distribution and abundance of other plant and animal species (Rook & Tallowin, 2003; Wood et al., 2014) Currently, the Tamaraw's distribution is limited to a few known locations, Aruyan-Malati Tamaraw Reserve, and

Mt Iglit-Baco Natural Park (Boyles et al., 2016) Most recent area confirmed the “re- discovery” of tamaraw population in the Upper Amnay Watershed Region at the border between Occidental and Oriental Mindoro (Schütz, 2019); estimates in the Mount Calavite Wildlife Sanctuary ranges in 4-6 individuals (Tabaranza et al, 2022) However, conducting field surveys to locate Tamaraw in the vast mountainous regions of Mindoro remains extremely challenging due to their elusive nature, harsh terrain and limited information on their distribution Nonetheless, ongoing conservation efforts and collaborative research initiatives aim to improve our understanding of Tamaraw ecology and develop effective strategies for their long-term survival and understanding of their distribution

Climate plays an important role in shaping the distribution patterns and assessing the influence of climatic variables over geographical areas (Thomas 2010; Sattar et al., 2021) For instance, temperature influences the physiological processes, metabolic rates and activity levels of organisms (Robinson et al., 1983; Amissah et al., 2014) Different species have specific temperature ranges within which they can thrive Temperature also determines the availability of suitable breeding or nesting sites, influencing the distribution patterns for many terrestrial species such as bears (Whiteman et al., 2015), tortoises (Demuth, 2001) and other bovine species such as cattles (Hansen, 2007) Next, precipitation has a significant impact on the distribution of species, particularly in terms of water availability (Guo, 2009) The availability of water directly affects the growth of vegetation, which in turn influences the availability of food resources for herbivorous species and the distribution of species at higher trophic levels (Bale et al., 2002) In addition to temperature and precipitation, other environmental factors such as elevation, slope, and topographic features play a crucial role in shaping species distribution (Hijmans et al., 2005; Ferrier et al., 2002) These factors influence microclimatic conditions, habitat structure, and resource availability, further influencing the spatial patterns of species' distributions Integrating these additional environmental variables into species distribution modeling enhances our understanding of the complex interactions between climate and habitat suitability, leading to more accurate predictions of species' potential ranges Understanding the intricate relationship between climate and species distribution is crucial for accurate species distribution modeling (SDM), which aims to predict and map the suitable habitat for a given species based on environmental variables (Araújo and Guisan, 2006; Elith and Leathwick, 2009)

SDM has promising applications in wildlife conservation These predictive models are crucial in the environmental management plans and development of policies (see studies of Zhang et al., 2012; de Queiroz et al., 2012) Commonly, the principle of species distribution modeling combines the occurrence data, where the species has been found or reported, with environmental layers that affect the distribution of a species There are two main types of SDM: correlative and mechanistic Correlative SDM utilizes the relationship between the occurrence data to the environmental data (see studies of Briscoe et al., 2016; Mod et al., 2016) On the other hand, mechanistic SDM requires physiological parameters that can be attributed to its distribution in a given environment (Buckley et al., 2010) SDMs utilize different types of machine-learning algorithms such as GARP, Bioclim, Domain, logistic regression and MaxEnt (maximum entropy) The maximum entropy approach in SDM was first proposed in the study of Dudik et al., 2004 to address the main challenges in machine-learning: small number of occurrence data and the absence of information Due to their predictive performance, SDM are often employed for conservation strategies There are numerous studies of SDM in the conservation of plants (see studies of Buebos-Esteve et al., 2023; Sharma et al., 2018; Kaky et al., 2020; Qin et al.,), animals (see studies of Lawler et al., 2011; Evans et al., 2015; Villero et al., 2017) and even microbial groups (see studies of Zarfl et al., 2008; Zhang et al 200) Among these algorithms, the MaxEnt performs better even though there is only limited data provided (Philips et al., 2008; Yackulic et al., 2013; Buebos-Esteve et al., 2023) The robustness of the maximum entropy algorithm is why it is chosen in this study as the primary approach to identify suitable locations of Tamaraw.

Understanding coastal vulnerability patterns

Climate pressures on coastal areas had become unprecedented throughout the years as these areas face constant jeopardy from natural hazards such as storm surges, tropical cyclones, and floods (Al Ruheili & Boluwade, 2023) In fact, projections indicate that risks to coastal areas will escalate in the twenty-first century due to the combined effects of rising sea levels, expanding presence of human development in flood-prone coastal regions, and other climate change-induced events (Tiggeloven et al., 2020) This in turn, yields cascading effects ranging from flooding and coastal erosion, to the destruction of critical ecosystems, such as mangroves and coral reefs This intensification of risks to coastal communities resulted in increased coastal vulnerability, which, in a more precise definition, refers to the susceptibility of coastal communities to the detrimental impacts of climate change Anthropogenic activities and climate change have already significantly degraded approximately 60% of global ecosystems, with the continuation of this threat anticipated in the future as climate change advances (Pereira et al., 2010) As such, coastal regions around the world are becoming increasingly vulnerable to these effects, putting both human populations and valuable ecosystems at risk, especially in countries like the Philippines (Yusuf & Francisco, 2009)

In the Philippines alone, the extent of mangroves has witnessed a substantial decline from an estimated 400,000-500,000 hectares in the 1920s to approximately 120,000 hectares in 1994 (Primavera, 2000) This decline can be attributed to the overexploitation of mangroves and their conversion into agricultural lands, salt ponds, and commercial settlements (Garcia et al., 2014) As a result, the loss of a protective buffer provided by coastal ecosystems and natural habitats increases the vulnerability of coastal populations and infrastructure to natural hazards It has become essential to adopt an ecosystem-based management approach to address this issue (O’Higgins et al., 2019) Mitigating the impacts of climate change, monitoring and modeling the effects of human expansion and interference on ecological and social systems and implementing sustainable development practices are critical to tackle both the current and future challenges (Yang et al., 2020) All of which are covered by the targets of the Sustainable Development Goals (SDG) particularly SDG14 (Life on Water) by the United Nations thus promoting strong sustainability in such areas (Neumann et al., 2017)

The prevailing increase in coastal vulnerability emphasizes the roles of natural habitats in mitigating coastal risks and hazards and is among the major approaches in coastal sustainability and resiliency programs Being a precursor for this science-based approach, there is a developing research interest in the methods that assesses coastal vulnerability There have been a number of models that have been developed to evaluate the effects of sea level rise, erosion, and inundation (Gornitz et al 1990; Cooper and McLaughlin, 1998; Hammar-Klose and Thieler 2001; Marzouk et al., 2021; Parthasarathy, 2021) based solely either on geophysical characteristics or qualitative assessments of natural habitat However, relatively few models consider both the geophysical and natural habitat characteristics of a region to map the relative vulnerability of coastal areas to erosion and inundation This gap was filled by the Coastal Vulnerability Model of the Integrated Valuation of Environmental Services and Trade-offs (InVEST) software (Sharp et al., 2020) The model produces a qualitative index of coastal exposure to erosion and inundation and ranks them from low, to moderate, to high exposure For instance, the coastal vulnerability takes into account a range of variables including sea-level rise predictions, coastal elevation, natural habitats e.g mangroves, corals, etc., and human density to assess the risks associated with climate change for coastal regions Recently, the InVEST Coastal Vulnerability Mode has been widely utilized and applied in numerous studies due to the robust framework for evaluating the susceptibility of coastal regions to erosion and other hazards making it a valuable resource for coastal management, policy development, and decision-making processes For instance, Hopper & Meixler (2016) applied InVEST model to spatiotemporally evaluate coastal vulnerability and the role of natural habitats in attenuating coastal vulnerability in Jamaica Bay, New York in different time scenarios: past (1609), present (2015), and future (2080) Similarly, a study conducted by Sajjad et al (2018) demonstrated that about a quarter of mainland China's coastline faces high vulnerability to coastal hazards, impacting over 5 million residents The study projects the doubling of these numbers by 2100, thereby underscoring the imperative to establish corresponding measures It also implicates the importance of natural habitats as they have the potential to reduce this vulnerability by up to 45% thereby providing a critical baseline for developing strategies to address these challenges and fortify coastal resilience On a more recent note, a study by Ai et al (2022) also utilized the same model to establish high coastal vulnerability in the northern coast of Jiaozhou Bay, in contrast to its southern and eastern counterparts, which display relatively lower vulnerability levels The juxtaposition between the north-south coastal vulnerability in the Jiaozhou Bay coast can be attributed to the type of coastline, elevation, distance to continental shelf, and urban development

The geography of the province of La Union is predominantly a hilly terrain which rises from the western coastal area eastwards to the Cordillera Mountain Range Its coastline stretches through the province with an estimated length of 114.70 km to 155.4km (DENR-R1, 2019), encompassing 11 municipalities with 94 coastal barangays, estimating to 185, 038 inhabitants in these coastal communities (Salmo et al., 2015) The province also covers natural coastal habitats of mangroves, coral reefs, and seagrass As of 2021, La Union has 644.973 ha of areas assessed for corals (DENR-R1, 2021) within the jurisdiction of the following municipalities: Balaoan (58.647 ha), Bacnotan (59.309 ha), Luna (39.575 ha), San Fernando City (364.496 ha), San Juan (12.431), and Bauang (110.515 ha) The coral reefs of La Union is part of the greater Lingayen Gulf embayment, encompassing its eastern section known as the Sector III that is characterized with nearshore and fringing reefs (Deocadez et al., 2003) Additionally, mangrove covers in the province are estimated to be around 120 (DENR-R1, 2021) –

150 ha (PENRO, 2020) These lands are declared as protected areas or parks in the province as per Provincial Ordinance No 352-2021, also known as the Environmental Code, found in Article B, Section 3.3B.04, Chapter 3.3; this includes the Bauang Mangroves, City of San Fernando Mangrove Swamp, Immuki Island of Balaoan, and the Agoo-Damortis Protected Landscape and Seascape Furthermore, data from 2017–2018 show that there are around 64 – 148 ha of seagrass (DENR-R1, 2021) within the jurisdiction of La Union; some study reports presence along the Agoo-Damortis Protected Landscape and Seascape on its cove together with some mangroves (Mamhot et al., 2018) The La Union Provincial Government considers the development and protection of mangrove lands, along with other natural coastal habitats, as a priority due to their provision of various ecosystem services to coastal communities This is particularly crucial since a significant portion of the population depends on fishing as a livelihood (Aduana-Alcantara et al., 2023) Nevertheless, the increase in aquaculture activities contributes to the loss of mangrove lands, resulting in a decrease in fish availability This, in turn, leads to the adoption of destructive fishing practices as a means of adaptation, which further damages coral reefs and seagrass ecosystems (Philreefs, 2003)

In a prior study conducted by Berdin et al., (2004) in La Union, Philippines, coastal erosion vulnerability mapping was performed to investigate the factors contributing to coastal erosion The study utilized geophysical characteristics and socioeconomic data by employing mapping and surveys in involved coastal communities The Coastal Vulnerability Index (CVI) method, as outlined by McLaughlin et al (2002), was employed to map the vulnerability To update and enhance the understanding of coastal erosion vulnerability in La Union, it is our goal to integrate InVEST Coastal Vulnerability Model By incorporating this model, the study can benefit from its comprehensive assessment of vulnerability based on various factors such as wind and wave exposure, and sea-level rise projections The InVEST model provides a robust framework for evaluating the susceptibility of coastal regions to erosion and other hazards caused by extreme weather events By employing the InVEST Coastal Vulnerability Model, the study herein can further refine the understanding of coastal erosion dynamics and inform effective coastal management strategies.In hindsight,the Coastal Vulnerability Model can help inform decision-making processes for policymakers and resource managers seeking to mitigate and adapt to the effects of climate change Hence, the objectives for this study is to (i) determine the vulnerability of the coastal areas of La Union, Philippines and (ii) compare the coastal vulnerability between two coastal scenarios (with vs without natural habitat)

Methodology

Land use/land-cover change and land surface temperature in Metropolitan

This study is conducted over one region in the Philippines, Metro Manila Metro Manila (Fig 1) is located in the southwestern part of the Luzon Island of the Philippines which is geographically divided into 4 zones: (i) Coastal Margin, (ii) Guadalupe Plateau,

(iii) Marikina Valley and (iv) Laguna Lowlands According to the Koppen climate classification, most of the region has a tropical wet and dry climate MM is composed of

17 municipalities- Caloocan, Las Pinas, Makati, Malabon, Mandaluyong, Manila, Marikina, Muntinlupa, Navotas, Paranaque, Pasay, Pasig, Pateros, Quezon City, San Juan, Taguig, and Valenzuela MM has a total area of 620 square kilometers with 13 million residents as of 2020 according to the National Statistics Office of the Philippines This continuing rapid growth of population has brought significant changes in the landscapes of metropolitan areas along with serious health issues due to unsanitary conditions and expansion of squatters’ settlements (Malaque et al., 2007) The rapid growth from 3.97 million residents in 1970 to 13 million in 2020 is the result of natural birth and migration (Ortega, 2014) from rural areas As the 5th most populous urban area in the world, urban sprawl is undeniably a major factor in landscape changes of MM Key factors such as economic growth, opportunities and accessibility of MM are hypothesized to be influencing the patterns of LULC (Estoque, 2017) As a matter of fact, according to the Department of Environment and Natural Resources (DENR), only

12 million hectares of the 60 million hectares total land area of MM are green spaces (recreational parks, protected forests, etc.) like the La Mesa Ecopark and the Las Piủas-Paraủaque Critical Habitat and Ecotourism Area (LPPCHEA)

Figure 1 The study area of Metropolitan Manila, Philippines with its respective municipalities

3.1.2 Landsat Satellite Image and Metro Manila Boundary

In this study, satellite images of Landsat 7 Thematic Mapper + in the year of 2001 and 2019 (Path: 116 Row: 50) were downloaded (Table 1) from the USGS Earth Explorer Website (https://earthexplorer.usgs.gov/) Both scenes that were acquired in this study have less than 10% of cloud cover To focus on the region of interest, the downloaded images were subjected to masking using the boundary layer of MM that was downloaded from http://philgis.org/ in a shapefile (.shp) format The downloaded satellite images and boundary layer of MM were then imported to the ArcMap v10.3 software for geoprocessing analysis

Satellite Sensor Acquisiti on Date

3.1.3 Land Surface Temperature retrieval using Landsat 7 ETM+

Landsat 7 ETM+ thermal bands (Table 2) were used to calculate the LST for the year 2001 and 2019 using the algorithms provided by the National Aeronautics and Space Administration (NASA) First, the Digital Number (DN) was converted to Spectral Radiance (Lλ) using the Raster Calculator tool in ArcMap with the formula:

Lλ = ((LMAXλ – LMINλ) / (QCALMAX – QCALMIN)) * (QCAL –

QCAL = Quantized calibrated pixel value in DN

LMAXλ = Spectral radiance scaled to QCALMAX in (Watts / (m 2 * sr * àm)) LMINλ = Spectral radiance scaled to QCALMIN in (Watts / (m 2 * sr * àm)) QCALMAX = Maximum quantized calibrated pixel value in correspondence of LMAXλ in DN

QCALMIN = Minimum quantized calibrated pixel value in correspondence of LMINλ in DN

Lastly, the Spectral Radiance will be converted to temperature in Celsius to get the LST using the formula: b T = K2 / ln(K1 / Lλ + 1) – 273.15

3.1.4 Calculation of NDVI and NDBI using Landsat 7

In order to calculate the NDVI using Landsat 7, the bands 4 (near infrared) and 3 (red) were used in both 2001 and 2019 Using the ArcMap software, the Raster Calculator tool was utilized A similar approach was used to calculate the NDBI in MM however, the bands 5 (shortwave infrared) and bands 4 (near infrared) were used The NDBI retrieval was done similarly using the Raster Calculator tool in ArcMap software The relationship between LST with NDVI and NDBI was calculated using a Pearson R correlation analysis

3.1.5 Land-use / Land Cover (LULC) and Accuracy Assessment

The LULC procedure refers to the classification of anthropogenic activities and natural landscapes within a specific time Several studies show the importance of LULC in environmental planning (Bosso & Yunusa, 2021), urban heat mitigation (Karakuş, 2019), and ecosystem management (Ma et al., 2021) that can help policy-makers and decision makers to implement policies (Chamling & Bera, 2020) to save the environment The LULC of 2001 and 2019 in MM were done using supervised classification In this study, a total of 160 training samples (80 from 2001 and 80 from 2019) were randomly selected in the study area using the training sample manager in ArcGIS software The training samples were user-assigned into 4 different classes: (i) built-up areas (ii) water (iii) vegetation and (iv) barren land (see Table 3) Next, the training samples were classified using the Maximum Likelihood (ML) algorithm based on user-assigned selection criteria The ML algorithm was chosen in the supervised classification method in this study as it outperforms other algorithms in LULC procedures (Jog & Dixit, 2016, Chughtai et al., 2021, ) However, supervised classifications are also subjected to user errors such as mislabelled classes, unidentified classification etc which is why a statistical approach is needed to quantify the errors of the generated LULC maps (Akpoti et al., 2016, Rwanga & Ndambuki, 2017) Finally, for validation of generated LULC maps in 2001 and 2019, accuracy assessment matrices were generated to calculate the producer accuracy, user accuracy, and overall accuracy The classified maps were cross-checked to a high-quality basemap such as Google Earth Pro as a reference The producer, user and overall accuracy of the LULC maps were calculated using the following formula:

Producer’s Accuracy (PA) = Number of Correctly Classified Pixels in each Category/Total number of Reference Pixels in that category (Column total) * 100

User’s Accuracy (UA) = Number of Correctly Classified Pixels in each Category/ Total number of Reference Pixels in that category (Column total) * 100

Overall Accuracy = Total Number of Correctly Classified Pixels in each Category/ Total number of Reference pixels* 100

Table 3 Classification Schema for LULC (Source: https://www.arcgis.com)

Built-up Areas Human-made structures; industrial, residential, commercial infrastructure Examples: houses, dense villages/towns/cities, downtowns, paved roads, asphalt

Water Areas where water was predominantly present

Vegetation Areas which are covered with sparse grasses, small shrubs and dense forest Examples: parks and grasslands

Barren Land Areas of rock or soil with very sparse to no vegetation; fallow land and vacant land

Predicting habitat suitability of Tamaraw

The study was conducted in the island of Mindoro (see Fig 1), situated along the southwestern coast of Luzon, Philippines With a total land area of 10,671 km2, Mindoro Island is characterized by a mountainous core terrain that extends throughout its entire length, encompassing the Mindoro Mountain Range, the longest mountain range in the region Mount Halcon, standing at an elevation of 8,583 feet, marks the highest point on the island Despite its remarkable geographical features, the biodiversity of Mindoro Island remains understudied, even though it is recognized as one of the world's biodiversity hotspots The island harbors a rich diversity of endemic birds: Mindoro Bleeding-heart (Gallicolumba platenae), Mindoro Racquet-tail (Prioniturus mindorensis), mammals: Mindoro Forest Mouse (Apomys mindorensis), Mindoro Stripe- faced Fruit Bat (Styloctenium mindorensis), and reptile species: Mindoro Bronzeback Snake (Dendrelaphis mindorensis), Mindoro Crocodile Skink (Tropidophorus grayi), which are facing threats from anthropogenic factors and habitat loss (Gonzalez et al., 1998) Mindoro is currently divided into two main provinces, Occidental Mindoro and

Oriental Mindoro, each with its own unique geographical and ecological characteristics Due to this, there are two main seasons in Mindoro, dry and rainy seasons The rainy season in Occidental Mindoro begins in May and traverses until November, whereas the dry season, this starts as the former season ends and spans until April (Macabago et al., 2012) On the other hand, Oriental Mindoro has no distinct dry or rainy seasons Although, this still resembles a similar climate type with Occidental Mindoro where maximum rainfall with increasing intensities occurs during the months of June to September (Dagamac et al., 2015)

Figure 2 The study area, Mindoro, along with its administrative boundaries Mindoro is located at the southwestern part of Luzon

The study utilized occurrence records of Tamaraw from the data provided by the d'ABOVILLE Foundation and Demo Farm Inc (DAF), a non-profit organization working alongside the DENR-TCP with the purpose of protecting the environment and biodiversity of Mindoro The DAF and DENR_TCP conducted field surveys and assessments, resulting in a total of 1,552 occurrence records of Tamaraw These records were collected through indirect monitoring methods by detecting the indirect signs of presence such as dungs and hoof marks and by camera traps survey Among the occurrence records, there were 8 records from Amnay, 6 records from Aruyan-Malati, 1 record from Mt Calavite Wildlife Sanctuary (MCWS), and a majority of 1,537 records from Mount Iglit Baco National Park (MIBNP) Notably, there are higher occurrence records in some areas due to the frequency of assessments done in the field To address the potential bias towards areas with a higher number of occurrences, a data thinning approach was employed The spThin package in R Studio was utilized to thin the occurrence data (Fig 2), reducing spatial clustering and ensuring a more representative distribution across the study area in 10 km settings Lastly, the thinned occurrence records were then transformed into a comma delimited (CSV) file format, as required by the MaxEnt software

Figure 3 Direct and indirect presence records of Tamaraw based on surveys (left) and the final occurrence points used in the MaxEnt modeling after data thinning with spThin package (right)

The study incorporated various environmental variables (Table 1) to comprehensively assess their influence on the distribution patterns of Tamaraw The current bioclimatic variables and elevation data (SRTM) were obtained from the WorldClim website (https://www.worldclim.org/), providing global climate data at a resolution of 30 seconds This study also considered two Shared Socioeonomic Pathways (SSP) scenarios, SSP1-2.6 (optimistic) and SSP3-7.0 (pessimistic), for the time period 2081-2100 These hypothetical scenarios were developed by the Intergovernmental Panel on Climate Change (IPCC) that outline potential future socioeconomic and environmental conditions on different assumptions on population growth, economic development, energy use, land use, and many other factors SSP1-2.6 represents a pathway characterized by sustainable development and low greenhouse gas emissions

It assumes strong international cooperation, rapid technological advancements, and effective climate change mitigation strategies In this scenario, global greenhouse gas emissions are significantly reduced, leading to a stabilization of radiative forcing levels by 2100 The SSP1-2.6 pathway envisions a future with a focus on sustainability, social equity, and environmental protection On the other hand, SSP3-7.0 represents a pathway characterized by high population growth, slow economic development, and fragmented efforts to address climate change It assumes limited global cooperation, regional conflicts, and challenges in implementing climate change mitigation measures This scenario leads to high greenhouse gas emissions and a continuation of business-as-usual practices The SSP3-7.0 pathway envisions a future with increased inequality, limited environmental policies, and higher vulnerability to climate change impacts The projected climate data were downloaded from the EC-Earth3-Veg-LR model Additionally, the slope variable was derived from the elevation data using the slope tool in ArcMap 10.7 software, providing information about the steepness of the terrain To incorporate information about land-use/land cover, the study utilized the current land- use/land cover (LULC) data for Mindoro, which was downloaded from the ESRI Living Atlas website (Karra et al., 2021) This dataset, generated through a machine learning approach with an average accuracy of 85%, offers detailed information about the land cover classes present in the study area Furthermore, the Euclidean distance tool in ArcMap 10.7 was employed to calculate the distance values from specific land cover classes such as crops, settlements, and water In addition, the study integrated the Normalized Difference Vegetation Index (NDVI) values derived from Landsat 8 satellite imagery, accessed from the USGS EarthExplorer website NDVI, a widely used vegetation index, provides essential information about the density and vigor of vegetation Lastly, all environmental variables were carefully matched in terms of projection, extent, and resolution and were then converted from raster to ASCII format as the input data in the MaxEnt software

Table 4 List of all the environmental variables used in MaxEnt modeling

BIO2 Mean Diurnal Range WorldClim

BIO7 Temperature Annual Change WorldClim

BIO8 Mean Temperature of the Wettest

BIO16 Precipitation of Wettest Quarter WorldClim

BIO17 Precipitation of Driest Quarter WorldClim

BIO18 Precipitation of Warmest Quarter WorldClim

BIO19 Precipitation of Coldest Quarter WorldClim

Distance to settlements Derived from ESRI 2020 LULC Data Distance to water Derived from ESRI 2020 LULC Data Distance to crops Derived from ESRI 2020 LULC Data

3.2.4 Removal of highly correlated variables and model tuning using ENMeval

Collinearity among environmental predictors decreases the efficiency and increases the uncertainty of SDMs (De Marco & Nóbrega, 2018) In the study, the 19 bioclimatic variables were assessed using the SDMToolbox that was downloaded from https://www.sdmtoolbox.org/ SDMToolbox is a python-based toolkit developed by Brown (2014) specifically for ArcMap software The toolbox includes a number of tools to improve the performance and complement MaxEnt models In SDMToolbox, the

“remove highly correlated variables' ' tool was utilized and variables with correlation coefficients of >0.8 were chosen Furthermore, to avoid overfitting of ecological niche models, the ENMeval 2.0 (Kass et al., 2021) package in R studio was used to evaluate the environmental variables and occurrence records using the random k-fold method to analyze the best values of regularization multiplier (RM) and feature type (FT) The regularization multiplier controls the complexity of the model and helps prevent overfitting by penalizing overly complex models The feature type refers to the transformation applied to the environmental variables, such as linear, quadratic, or product terms By exploring different combinations of RM and FT, the study aimed to identify the most suitable configuration that maximizes model performance and generalizability to achieve better predictive accuracy and effectively represent the species' habitat preferences (Muscarella et al., 2014; Muscarella et a., 2017; Angelov, 2019)

Finally, the last step in the modeling procedure is the usage of MaxEnt software, an open-source software for species distribution modeling utilizing a machine-learning approach known as maximum entropy which can be downloaded from (https://biodiversityinformatics.amnh.org/open_source/maxent/) MaxEnt employs machine-learning techniques to predict species distribution based on the most influential environmental conditions The generated occurrence points in CSV file format and environmental variables in ASCII format were used as input files in the MaxEnt software First, the option "Create a response curve" was chosen to generate a curve illustrating the relationship between the species occurrence and the environmental variables Additionally, a "jackknife test" was performed to measure the importance of each environmental variable in the modeling process, identifying the variables that significantly influence the species distribution Moreover, The study followed the settings previously suggested by the ENMeval package, including a regularization multiplier (RM) value of 3 and a feature type (FT) of Linear These settings were applied with 10 replications, and the output format was set as cloglog The resulting MaxEnt model would produce a raster ASCII output, with values ranging from 0 (indicating unsuitability) to 1 (indicating suitability) across the study area Additionally, the area under the curve (AUC) of the receiver operating characteristic (ROC) curve was used as an evaluation metric Further, the resulting file was then imported into ArcMap 10.7 software to get a detailed visualization using a defined interval.

Applying InVEST model to assess vulnerability of the coastal areas of La Union, Philippines based on multiple bio-geophysical hazards…

La Union (Fig.1) is a province located in the Ilocos Region of the Philippines, situated along the country's western coastline with a population of over 820,000 people

It is bordered by the Cordillera Mountain Range to the east and the South China Sea to the west There are a total of 20 municipalities in La Union, where 12 of them are coastal areas facing the West Philippine Sea Within its boundaries, La Union encompasses several protected areas and natural reserves that are vital for biodiversity conservation and sustainable reef to ridge management (Balaoro-Banzuela et al., 2023) Mangroves, coral reefs, and seagrass beds are among the significant natural habitats found along the coastal areas of La Union.Despite its thriving economy and tourism industry, La Union is not immune to environmental challenges, including those posed by climate change (Capili et al., 2005; Ngilangil et al., 2013; Ngilangil et al., 2018) Over the last decade, the province has experienced significant impacts from typhoons, flooding, and landslides, thus highlighting the need for effective risk management and adaptation strategies

Figure 4 Study Area of La Union Province, Philippines with its respective municipalities

The study compiled several data requirements for InVEST Coastal Vulnerability Model (Table 1) including: Area of Interest (AOI) of La Union, Landmass, WaveWatchIII, Bathymetry, Digital Elevation Model (DEM), Continental Shelf Contour, Habitats (mangroves, coral reefs and seagrass) table, and population density (Fig 2) The AOI and DEM (Shuttle Radar Topography Mission or SRTM at 30 m resolution) were downloaded from the DIVA-GIS country level data (https://diva- gis.org/gdata) which were clipped to the extent of La Union to focus on the study area The landmass data was derived from the GADM (https://gadm.org/data.html) and the WaveWatch III and Continental Shelf Contour were sourced from the InVEST sample data The Bathymetry was downloaded from GEBCO’s global gridded dataset that has digitized depth points Further, the habitat data, specifically mangroves, coral and seagrasses,were downloaded from Allen Coral Atlas (https://www.allencoralatlas.org/)

All of these datasets were projected in the same coordinate system (WGS84 UTM57N) as preparation for input in the InVEST software

Table 5 Coastal vulnerability model data inputs and sources

Input Year Extent Resolution Source

La Union Vector DIVA-GIS Country Level

Mangrove 2020 The Global Mangrove Watch

National Oceanographic and Atmospheric Administration WaveWatch III

Continental Margins Ecosystem (COMARGE) effort in conjunction with the Census of Marine Life

3.3.3 Coastal Vulnerability Assessment using InVEST

The coastal risk assessment was done with InVEST 3.11.0 Workbench, an open- source software that was downloaded from Natural Capital Project website of Stanford University (https://naturalcapitalproject.stanford.edu/software/invest) The InVEST Coastal Vulnerability model is an index-based tool used to evaluate the relative risk of coastlines to erosion and inundation and acknowledges the role of natural habitats such as mangroves in providing coastal protection The coastal vulnerability index is constructed using bio-geophysical variables including geomorphic characteristics, sea- level rise, bathymetry, topography, and relative wind and wave forces which are qualitatively ranked based on the methods of Gornitz (1990) and Hammar-Klose and Thieler (2001) The computation of exposure index EI for each shoreline segment was done using the following formula:

Formula derived from InVEST Coastal Vulnerability Workbook

Moreover, the Coastal Vulnerability was done with the following settings in InVEST Workbench: model resolution= 500, maximum fetch distance00, elevation averaging radius= 5000 The output file (geopackage) was then exported to ArcMap 10.7 for a detailed classification using a defined interval with the following classification scheme:

1 (very low exposure), 2 (low exposure), 3 (moderate exposure), 4 (high exposure), and

This study also analyzed the role of habitats in safeguarding the coastal areas by generating two model scenarios, one considering the presence of natural habitats and without natural habitats By comparing the outputs of the InVEST models under these two scenarios, the study aimed to quantify and evaluate the influence of habitats on coastal protection Under the "with habitats" scenario, the InVEST model incorporated the presence and characteristics of various habitats, such as mangroves, coral reefs, and seagrass beds, which are known to provide natural protection against coastal hazards.

Results

Land use/land cover change and land surface temperature in Metropolitan

4.1.1 Temporal variation and correlation of LST, NDVI, and NDBI

The LST values (Table 4) from Landsat 7 ETM+ satellite yielded values ranging from 13.27° to 36.84°C with a mean temperature of 27.90°C in 2001 In 2019, the values are ranging from 11.75 to 42.64°C with a mean temperature of 31.92°C This is visualized in Fig 2, that showed the spread of hot areas in MM Next, the NDVI values for MM were ranging from -0.57 to 0.55 in 2001 and -0.71 to 0.60 in 2019 and were seen to have a decrease in areas with green vegetation (see Fig 2) Lastly, the NDBI values retrieved ranged from -0.68 to 0.48 in 2001 and -0.63 to 0.72 in 2019 that clearly showed the increase on the areas with impervious surfaces (Fig 2) Furthermore, The LST in

2001 and 2019 were both correlated with the NDBI and NDVI The relationship between LST and vegetation index represented as NDVI showed an inversely proportional relationship (Fig 3) as indicated by negative correlation with Pearson r values of -0.60 in 2001 and -0.050 in 2019 On the other hand, the LST and urbanization index represented as NDBI displayed a directly proportional relationship (Fig 3) as indicated by the positive correlation with Pearson r values of 0.80 and 0.68 in 2001 and 2019, respectively (Fig 3)

Table 6 Retrieved Land Surface Temperature using Landsat 7 ETM+

LST Min Max Mean SD

4.1.2 Comparison of LULC of 2001 and 2019

For the LULC, the built-up areas which consist of residential, commercial and roads have the highest area with a total of 424.21 square kilometers in 2001 and 445.16 square kilometers in 2019 Second, the water has a total area of 18.76 in 2001 and 16.42 square kilometers in 2019 Third, the vegetation has a total area of 106.18 square kilometers in 2001 and 88.32 square kilometers in 2019 Lastly, the barren land has a total area of 9.36 in 2001 and 8.60 in 2019 (Table 5) The accuracy assessment matrices were generated based on Google Earth Pro for validation For the LULC in 2001, barren land had the highest user accuracy (87.50%) On the other hand, vegetation had the highest user accuracy (96.29%) in 2019 In addition, the overall accuracy is 83% and 94% for 2001 and 2019, respectively

Figure 5 The LST, NDVI, NDBI, and LULC maps of Metro Manila in 2001 and 2019

Figure 6 Graphs of LST between NDVI and NDBI which showed directly proportional and indirectly proportional relationship, respectively

Applying InVEST model to assess the vulnerability of the coastal areas of La Union, Philippines

Figure 1 The study area of Metropolitan Manila, Philippines with its respective municipalities

Figure 2 The study area, Mindoro, along with its administrative boundaries Mindoro is located at the southwest part of Luzon

Figure 3 Direct and indirect presence records of Tamaraw based on surveys (left) and the final occurrence points used in the MaxEnt modeling after data thinning with spThin package (right)

Figure 4 Study Area of La Union Province, Philippines with its respective municipalities Figure 5 The LST, NDVI, NDBI and LULC maps of Metro Manila in 2001 and 2019

Figure 6 Graphs of LST between NDVI and NDBI which showed directly proportional and indirectly proportional relationship, respectively

Figure 7 Response curves for the highest contributing environmental variables for Tamaraw for current AB, optimistic CD, pessimistic EF climate scenarios

Figure 8 Habitat suitability models of Tamaraw under the current (1970 - 2000) and future climate change (2081-2100) scenarios

Figure 9 Exposure index map of La Union “with natural habitats” (A) and “without natural habitats” (B) scenarios with their distribution as depicted by the pie charts found on the upper left

Figure 6 Habitat suitability models of Tamaraw under the current (1970-2000) and future climate change scenarios (2081-2100)

Figure 7 The study area of La Union, Philippines with its respective municipalities Figure 8 Exposure index along the coast of La Union with habitats (left) and without habitats (right) scenarios

Table 3 Classification of Schema for LULC

Table 4 List of all the environmental variables used in MaxEnt modeling

Table 5 Coastal vulnerability model data inputs and sources

Table 6 Retrieved Land Surface Temperature using Landsat 7 ETM+

Table 7 Environmental variables that contribute to the suitability model (Percent Contribution) of Tamaraw in current and future climate scenarios

Table 8 LULC classification area size comparison in square kilometers in 2001 and 2019

CSV Comma delimited file format

InVEST Integrated Valuation of Environmental Services and Trade-offs LST Land Surface Temperature

LULC Land-use/Land cover

MCWS Mt Calavite Wildlife Sanctuary

MIBNP Mt Iglit-Baco National Park

NDVI Normalized Difference Vegetation Index

NDBI Normalized Difference Built-up Index

Technologies such as remote sensing and species distribution modeling have become increasingly important tools for addressing global challenges such as climate change and biodiversity conservation Remote sensing techniques, such as the utilization of satellite imagery, allow for the monitoring and mapping of the Earth’s surface through land cover change studies On the other hand, modeling techniques such as species distribution modeling, provides insights into the potential impact of the changing climate to a species which can then lead to effective management strategies and adaptive coastal management plans With these technologies, this study aims (i) to investigate the impact of of land-use/land-cover change on land surface temperature in Metropolitan Manila, Philippines using Landsat imagery, (ii) to predict the habitat suitability of the endangered Tamaraw (Bubalus mindorensis) in Mindoro, Philippines using species distribution modelling approach and (iii) to conduct a coastal vulnerability assessment using the (Integrated Valuation of Ecosystem Services and Tradeoffs) InVEST software in La Union, Philippines that target the United Nations Sustainable Development Goals and One Health Agenda

These three different approaches using remote sensing and modeling, are significant because they address pressing environmental issues in the Philippines For instance, Metropolitan Manila, a highly urbanized area, is facing rapid land-use/land- cover change throughout the years, which can have significant impacts on local climate, especially the land surface temperature The understanding of patterns on Metropolitan Manila’s landscape can aid the development of environmental policies and management plans to ensure the sustainability of urban areas for the betterment of the quality of urban life On the other hand, the Tamaraw is a critically endangered species that is endemic to the island of Mindoro, Philippines Habitat loss and fragmentation have led to a decline in the Tamaraw population that was once common in the island of Mindoro Species distribution models that integrate various environmental factors with known presence data can provide a comprehensive understanding of the factors affecting these species to aid the ongoing conservation and management efforts Finally, coastal vulnerability assessment using the InVEST software in La Union, Philippines that will provide information on the exposure, sensitivity, and adaptive capacity of coastal areas The results of this study can guide policymakers and stakeholders in implementing measures to reduce the vulnerability of coastal communities to climate change and promote sustainable coastal management