An Integrated Assessment of Climate Change Impacts and Implications on Bonaire

Bonaire’s topographic and geographic characteristics, in combination with the island’s high dependency on economic sectors that are susceptible to the impacts of climate change, make this Caribbean island particularly vulnerable to climatic changes. In this study, biophysical and economic models are combined and complemented with stakeholder consultation to assess and quantify environmental effects and associated socio-economic impacts of climate change on Bonaire. We apply three climate scenarios of the 2021 IPCC report (SSP1-2.6, 2–4.5, and 5–8.5) and combine them with local conditions to conduct a site-specific integrated assessment. The results show that various buildings, critical infrastructure, and identified tangible cultural heritage, especially at the south of Bonaire, are at risk of climate change induced coastal inundation by 2050, even under the least severe climate projection. In addition, the overall health of coral reefs declines under the climate scenarios SSP2-4.5 and SSP5-8.5 due to sea level rise, acidification, and increasing temperatures. In the most pessimistic scenario, Bonaire could experience a reduction in dive tourist arrivals of 118,000, which can lead to an economic contraction of 174 USDm (25%) in Bonaire’s GDP. In the absence of timely planning and implementation of adaptation measures, the impacts of climate change may have serious implications for inhabitants’ lifestyles and wellbeing. These results are imperative for various stakeholders, and stress that decision-makers should focus on the development and implementation of effective and feasible adaptation strategies urgently. Moreover, future researchers confronted with data scarcity in comparable contexts can utilise the novel methodologies employed in this study.


Introduction
The Dutch Caribbean Island of Bonaire, located in the Caribbean Sea (Uyarra et al. 2005), as scuba diving is tourists' primary mainstay activity. It was estimated that in 2012, the direct GDP from tourism amounted to 16.3% of total GDP (Statistics Netherlands 2012), but it should be noted that Bonaire's tourism sector is a far-reaching system that serves growth in other vital industries . However, this also implies that Bonaire's local economy is heavily dependent on external rather than domestic resources and, since Bonaire is relatively limited in size and natural resources, the lack of a diversified local economy makes it inherently sensitive to exogenous shocks (Spies et al. 2015;World Bank 2021). As with other areas in the Caribbean, the reefs around Bonaire have suffered from increased pressures from climate change, such as rising coral mortalities due to bleaching events and diseases (Spalding et al. 2001), and local pressures, such as erosion and nutrient loading.
This loss of healthy coral reefs will not only affect the attractiveness of Bonaire as a travel destination and influence tourism demand, but may indirectly result in socio-economic impacts that affect the quality of life on the island, including health, natural resources, coastal protection, water and food security, and employment opportunities (Rhiney 2015;Taylor et al. 2018). Furthermore, other effects of climate change, such as gradual changes in SLR, the intensity of storm surges, and the occurrence of extreme events, may directly jeopardise public safety and the built environment. Due to the fact that large parts of the island are low-lying, Bonaire's built environment and tangible cultural heritage are specifically exposed to storm surges and coastal flooding.
Recent years have seen a spike in studies addressing the impacts of climatic stressors and assessing associated socio-economic impacts (Sesana et al. 2021). For example, various studies examine the direct inundation damage costs of SLR to infrastructure such as buildings and roads (e.g. Chinowsky and Helman 2021;Lin et al. 2014). Other studies, on the other hand, have addressed the threats posed by SLR on cultural heritage, of which some target islands specifically (e.g. Ravanelli et al. 2019;García Sánchez et al. 2020;Ezcurra and Rivera-Collazo 2018). Burke and Maidens (2004), for example, estimated that direct economic damage in the Caribbean region ranges from 350 to 870 USDm per year, while indirect losses resulting from losses in fisheries, dive tourism, and the need for additional shoreline protection services amount to 3 USDb to 4 USDb per year. However, most of these studies generally concentrate on one single phenomenon of climatic change, such as SLR, droughts, or coral reef degradation, and their socio-economic implications.
As opposed to existing literature, this study examines multiple socio-economic impacts of climatic change and the challenges of adaptation on a small island. By doing so, this study aims to assess various socio-economic impacts of anthropogenic warming on Bonaire by focusing on (I) economic losses due to coral reef degradation and (II) damages to buildings, critical infrastructure, and tangible cultural heritage due to coastal flooding by 2050. By means of a risk modelling framework, the environmental changes and associated consequences for the island of Bonaire are examined under various climate projections. First, the expected environmental changes associated with different climate projections, such as levels of coastal inundation, coral degradation, and temperature changes, are determined. Second, the expected socio-economic implications associated with the climatic changes are predicted, including impacts on the economy through reduced tourism demand, damages to the built environment, and the expected loss of cultural heritage. The results will indicate whether and what type of action is needed to protect the island of Bonaire and its inhabitants against the implications of climate change. In addition, this paper provides several policy recommendations for local and national authorities, including the necessity and location of coastal adaptation measures to safeguard the local community, infrastructure, and cultural heritage. Moreover, this study applies various innovative research methods and could therefore serve as a basis or inspiration for various other data-scarce islands being at risk of climate change.
The paper is structured as follows: Sect. 2 describes the study's methodological approach, including its strategy to data collection, modelling and analysis. The findings of the analyses are presented in Sect. 3. The discussion in Sect. 4 puts the findings of this paper in perspective to other studies and reports this study's limitations and opportunities for future research. Section 5 concludes.

Methods
To assess the impacts of climate change on Bonaire, a traditional risk approach was applied, in which risk is defined as the function of hazard -the probability of a flood event or coral reef degradation; exposure-the population and value of assets subject to flooding; and vulnerability-the capacity of a society to deal with the event (UNDRR 2015). Figure 2 shows the methodological strategy that has been applied for the integrated assessment of climate change impacts on Bonaire.

Hazard
To evaluate the effect of climate change on coastal flooding and coral reef degradation on the island of Bonaire, the following methodological steps were undertaken: First, different SLR projections for the year 2050 and the impact of storm events on coastal inundation were evaluated by dynamically incorporating storm tide and waves into a coastal inundation model. In addition, the influence of climate change-induced coral reef degradation on the inundation extent was assessed. A visual representation of the input data, consulted databases, applied research methods, and expected output can be found in Appendix A.

Sea Level Rise and Coastal Flooding
Coastal inundation was simulated using SLR projection, storm tides, wave setup, a Digital Elevation Model (DEM), and implications of coral reef deterioration.
The (NASA) IPCC AR6 Sea Level Projection Tool (2021) was consulted to acquire SLR projections for Bonaire in 2050. This tool forecasts global SLR from 2020 to 2150 based on the AR6 IPCC report's Shared Socioeconomic Pathways (SSP) scenarios, describing alternative socioeconomic developments (Garner et al. 2021;IPCC 2021). To investigate a wide range of possible futures for Bonaire, we look at three SSP-RCP scenarios: SSP1-2.6 is the applied low-end scenario, followed by SSP2-4.5 and SSP5-8.5 as the medium-and high-end scenarios, respectively. Whereas the first scenario depicts a drastic reduction of global CO2 and net-zero by 2050, in the second scenario CO2 emissions hover around current levels before starting to reduce mid-century, and in SSP5 at the upper end of the range, CO2 emissions levels roughly double by 2050. For every SSP SLR projection, the quantile was taken as the SLR value modelled in this study. Spatially, the projections of the grid cell where Bonaire is located are: 12°-13°N; 68°-69°W.
Storm tide data was retrieved from the COastal dAtaset of Storm Tide Return Periods (COAST-RP) dataset (Dullaart et al. 2021). This is a global dataset with storm tide levels corresponding to ten different return periods (RPs), ranging from 1 up to 1000 years. For the storm tide data of Tropical Cyclones (TCs), the Synthetic Tropical cyclOne geneRation Model (STORM) model by Bloemendaal et al. (2020) was applied to 38 years of historical TC data from the International Best Track Archive for Climate Stewardship (IBTrACS), to statistically extend this dataset to 10,000 years of TC activity. Subsequently, the synthetic tracks from the STORM dataset were utilised as forcing for the hydrodynamic Global Tide and Surge Model (GTSM) to simulate time series of TC storm surge levels (Dullaart et al. 2021). For all other types of storms, forcing data from the ECMWF Reanalysis v5 (ERA5) global climate reanalysis was used (Hersbach et al. 2022). In the last step, tidal levels were sampled randomly and combined with surge to obtain storm tide levels; the storm tide Return Periods (RPs) were then calculated using this information.
Wave setup was simulated using significant wave heights of wind waves and swell from the ERA5 global climate reanalysis (Hawker et al. 2022). Similar to Muis et al. (2020), extreme significant wave height was obtained by calculating the RPs from the annual maxima. Based on the study by Peachey (1986), the hydrological form of a wave was simplified to the shape of a sinusoidal with the formula: where: The DEM used in this research is the so-called FABDEM (Hawker et al. 2022). The FABDEM is referenced to the EGM2008 geoid model (Hawker et al. 2022) and has been corrected to the reference of the WGS84 geoid model for the storm tide hydrographs (Andersen and Knudsen 2009). This DEM is chosen because it is the first global DEM to exclude forests and structures, thereby expectedly providing more accurate elevation data of the land surface.
To incorporate the dynamic component of storm tides and waves into a simulation, the reduced-physics model Super-Fast Inundation of CoastS (SFINCS) was employed (Leijnse et al. 2021). SFINCS is a dynamic inundation model, indicating the possibility of incorporating the time dimension into the simulation. In this study, the SFINCS model is preferred over the use of other studies because other modelling approaches are considered too simple as they are static (e.g., the bathtub approach), or accurate but too slow (e.g., Delft3D, XBeach).
The most important topographical input files used by the model-besides the DEMinclude a bathymetry file (GEBCO_2021 Grid 2021), a land use file representing the roughness of the landscape (Buchhorn et al. 2020), and a global curve number file used for soil infiltration rates (Jaafar et al. 2019). Lastly, for the hydrography on Bonaire, the MERIT Hydro dataset was applied. This global dataset incorporates flow direction, flow accumulation, and river channel width (Yamazaki et al. 2019). The horizontal resolution of the model was set at 20 m to keep it computationally efficient for the size of the research area. The time span of the simulation was 7 days, with output calculations every 600 s for the output files. The model was adjusted to account for the constant water level of SLR prior to the storm, resulting in a more realistic future storm simulation (Leijnse 2022).

Ecological Module
The ocean near the coast of Bonaire contains coral reefs, which are a natural coastal protection against waves and storm surges (Ferrario et al. 2014). A case study from the Seychelles analysed the effects of coral reef degradation on the total wave energy that reaches the shore (Sheppard et al. 2005). Showing high variability, their model predicted that 20 years after the degradation, the wave energy reaching the shore could increase by 80% depending on the coral reef setup. In order to predict flood risks on Bonaire with deteriorated coral reefs, it was hypothesised that based on the coral reef type and profile the wave setup would also increase by 45%. This gives an indication on how coral reefs affect changes in flood risk and subsequently, the wave setup factor of 0.2 ) was multiplied by 1.8 for use in the model's wave setup calculations, yielding a wave setup factor of 0.36. Using this new factor, the wave height was recalculated, and a coral reef deterioration scenario was simulated for the 2050 middle-of-the-road SSP2-4.5 scenario.

Exposure Built Environment
Microsoft open-source data was used in combination with Open Street Map (OSM) data to determine the locations of all exposed assets (Microsoft Bing Maps, 2022, OSM;Open-StreetMap, 2022); other databases containing information of local buildings and addresses were not publicly available. The value of each type of structure was derived from several sources. Rider Levett Bucknall (RLB 2021) and BCQS International (2020) provide realistic construction costs for the majority of building categories. Federal Emergency Management Agency (FEMA) (2009) was consulted for the categories "schools," "government buildings," "(movie) theatres and museums," and "religious buildings" (2009). A neighbourhood sampling method was used to estimate the category of value per square metre for residential structures. Neighbourhoods were classified into four categories, and the individual value category can be found in Appendix B. For sheds, a calculation of half the price per square metre of industrial buildings was taken, since, based on fieldwork, it can be assumed that most sheds are constructed of similar materials as industrial buildings.
Similar to buildings, and following Nirandjan et al. (2022), critical infrastructure was classified into different categories: energy, transportation, drinking water, waste, health and first responders, and industry. The transportation category was considered most important, as it includes crucial international trade facilities, such as the airport and the harbour, as well as the islands' road infrastructure network.

Exposure Cultural Heritage
UNESCO defines cultural heritage as tangible monuments, artefacts, and sites that have historical, symbolic, and social values, and the intangible cultural heritage (ICH) that is embedded within this tangible cultural heritage (TCH) (Pessoa et al. 2009). Four research methodologies were used to identify and validate the most valued tangible cultural heritage assets of Bonaire: a literature study, expert interviews, participatory mapping, and social media analysis.
First, various local policy documents were reviewed in order to locate tangible cultural heritage referenced in the literature. Second, experts from culturally engaged nongovernmental organisations (NGOs) and foundations on Bonaire were interviewed to further examine how climate change is anticipated to affect tangible cultural assets on Bonaire and to validate the cultural heritage identified by the literature research. Third, a participatory mapping (PM) approach was applied; cultural experts and Bonairian residents (who had lived on Bonaire for at least two and a half years) were shown a map of the island and asked to identify cultural heritage that they thought was valuable and significant to Bonairian culture. Fourth, the findings of the literature study, participatory mapping, and expert interviews were complemented and validated by a social media analysis. In the social media analysis, 1,137 photographs that have been uploaded to Flickr by tourists and locals, were examined and categorised to identify the most frequently visited Bonairian cultural sites.
After identification of Bonaire's tangible cultural heritage, the results were mapped and subsequently merged with flood maps based on different climate projections to determine the tangible cultural heritage at risk from coastal inundation by 2050.

Exposure Economy
To assess the associated direct and indirect economic consequences of tourism losses due to climate change, an input-output (IO) modelling framework was developed. Considering the simplicity and transparency of IO models, this paper applied an IO analysis to link external ecological effects to economic impacts. The IO table of 2017 for Bonaire was constructed and updated using the supply and use table (SUT) of 2004 for Bonaire and has been structured as an industry-industry table. To estimate how the different industries have developed from 2012 to 2017, data on gross domestic product (GDP) growth and average salary growth per industry was retrieved from the Statistics Netherlands (Statistics Netherlands 2018). This results in a closed-economy IO table for the island of Bonaire for the year 2017. While an island such as Bonaire is highly dependent upon the rest of the world, there was insufficient data on international trade to explicitly incorporate it into the modelling framework.
Due to the island's small size and remoteness, its industries require high levels of imports of agricultural and manufactured products to operate. Therefore, the total intermediate demand of each industry required for production was split into domestic products (locally produced) and external products (import). Next, to better align the tourist expenditure patterns with the industry categorization of the Statistics Netherlands, the "Hospitality" sector was split up into the "Hotels" sector and the "Food services" sector. This subdivision is essential to capture the differences in expenditure profiles between stay-over tourists and cruise tourists. Since the tourism satellite account (TSA) of Bonaire (Statistics Netherlands 2012) does not account for this subdivision, the relative share of the two industries was based on the TSA of Aruba (Steenge and Van De Steeg 2010), which is assumed to have a similar structure as Bonaire. Data on tourism expenditures was retrieved from the exit survey of Schep et al. (2013) and the 2017 Tourist Exit Survey'' of STMP Bonaire (Croes et al. 2019). Finally, the total tourism expenditures per category are matched with its related industry in the IO table. The economic structure of Bonaire for the current situation and for the year 2050 was predicted by updating the developed IO matrix of 2017 using a CRAS (cell-corrected RAS-algorithm) approach, as proposed by Lenzen et al. (2009). CRAS is capable of compromising between inconsistent matrix entries, which is appropriate given that the 'new' set of constraints for 2050 is derived by making assumptions regarding sectoral and GDP growth for 2050. More information about the application of the CRAS approach in this study is presented in Appendix C.
Moreover, we apply the concept of "social carrying capacity" to predict changes in dive tourism arrivals. This concept captures the restricting factor of the availability of healthy reefs when visiting a diving destination (Davis and Tisdell 1995). The social carrying capacity of the coral reef on Bonaire is calculated by multiplying the number of quality dive spots by the maximum number of divers that can be sustained at each dive site (Koks and Van Zanten 2015). By comparing the total number of dives that the island can "supply", with the total "demand" for dives by visitors, a "resource balance" is established that indicates whether the island is able to accommodate the expected number of dive tourists. The resource balance in 2017 amounted to a surplus of 629,000 dives, indicating that the current demand for dives is within the social carrying capacity of the island. It is assumed that the total number of dives cannot exceed the social carrying capacity.

Vulnerability Dive Sites and Tourism Demand
To evaluate the decreasing supply of dives for tourists, the degradation and quality of local coral reefs were used to simulate vulnerability of dive sites. First, to estimate the effects of climate change on the quality of coral reefs around Bonaire, the Reef Health Index (RHI) (Kramer et al. 2015) was utilised as an indicator of coral reef quality that could be modelled under climate change scenarios. This index was developed to assess the state of the Mesoamerican Reef in the Caribbean (Díaz-Pérez et al. 2016). The coral reefs of Bonaire exhibit similar characteristics in coral species and depth as Mesoamerican reefs, making the RHI a valid indicator to proxy the state of Bonairian reefs (Meesters et al. 2019). The RHI consists of four key health indicators: (1) coral cover, (2) macroalgae cover, (3) biomass of key herbivorous fish, and (4) biomass of key commercial fish. An RHI ranging between 1.0-1.8 indicates a reef in "Critical" state, 1.8-2.6 indicates a "Poor" state, 2.6-3.4 equals a "Fair" state, 3.4-4.2 "Good," and 4.2 -5.0 is seen as "Very Good" (Kramer et al. 2015).
To predict future changes in the RHI of the 115 transect points by 2050, an ecological module was constructed based on the Green Economic Model (GEM) model by Koks and Van Zanten (2015), a macroeconomic IO model with an ecological module. Key reef health indicators are impacted by global climate stressors through the external effects of algal blooms, ocean acidification, and coral bleaching. The magnitude of these global stressors depends on the CO2eq concentrations, Sea Surface Temperature (SST), and Air Temperature (AT) levels of the climate scenarios. The ecological module directly links any changes in key health indicators in the reef ecosystem to the quality of the available dive spots on Bonaire. When a transect zone reaches a final RHI below 1.8 and its state is considered "Critical", it is assumed that the dive site becomes unattractive for scuba diving. As such, the degradation of local coral reefs reduces the available amount of dive sites, thus decreasing the supply of dives for tourists.

Vulnerability Built Environment
This part of the study aims to identify the extent to which Bonaire's buildings and critical infrastructure will be directly impacted by future climate change in 2050, focusing on floods and storms specifically. To identify the vulnerability per asset, a classification was made based on building height. The buildings were classified using the earlier described neighbourhood sampling, as no detailed data was available. A vulnerability curve was created for each building type, which shows how much an asset or network feature will be damaged by a specific hazard of a certain intensity (i.e. the level of damage for a certain flood depth). Furthermore, the road network's connectivity was evaluated. Roads that are cut off from the network by flooding are added to the already unusable roads as they are unreachable and therefore also unusable. To determine which roads are cut-off and, hence, unusable, a line polygon connectivity checking tool in a GIS was utilised. By combining the damages to both buildings and critical infrastructure, a more complete overview of the expected direct impacts is presented (Garschagen et al. 2016).

Vulnerability Economy
One of the key strengths of IO analysis is to identify how changes in one sector may influence changes in another sector through inter-industry linkages. These linkages allow us to determine the economy-wide impacts of environmental shocks from tourism losses to other industries in the economy of Bonaire. This concept is also known as the 'multiplier effect': the variances in output of other industries within an economy when the demand of a single industry changes. The magnitude of these interrelationships is estimated by determining the multiplier coefficients. The multiplier effects of tourism were estimated using the approach of Frechtling and Horváth (1999). The process of estimating total revenue losses was based on the number of stay-over and cruise tourist arrivals, the average number of nights spent on Bonaire, and the average tourist expenditures per tourist per day (Mayer and Vogt 2016), as shown in Appendix D. By coupling tourist expenditure with our macroeconomic framework, we can assess the macroeconomic impact of tourism losses. We define these losses as the changes in the GDP of Bonaire due to the induced effects of tourism demand. Figure 3 illustrates the inundated areas of Bonaire in 2050 under the climate projections SSP1-2.6, SSP2-4.5 and SSP5-8.5. It is noticeable that mainly the parts in the South of Bonaire are predicted to be inundated by 2050 under all climate scenarios and that differences in predicted inundation levels between climate scenarios appear to be minimal. This latter insight can be explained by the minor variations in SLR.

Coastal Inundation
To indicate the difference in inundation across the SSP scenarios, the total inundated surface area is calculated and presented in Fig. 4 and shows that the total inundated surface area is between 8.0 and 8.3 km 2 . The saliñas and mangroves are excluded from the total inundated surface area because these areas are already inundated. However, it must be noted that the characteristics and, hence, the services provided by these ecosystems may change due to permanent inundation and coastal flooding.  Figure 5 illustrates the projected inundation difference between a healthy and completely degraded coral reef for the SSP2-4.5 scenario in 2050. Assuming that reef degradation increases WSU by 45%, the flood extent increases. The total inundated surface area, excluding the saliñas, increases by 41%, from 8.0 to 11.3 km 2 . Inundation depth increases by an average of approximately 0.24 m, with a larger increase directly on the coastline.

Impact and Damage to Buildings
As presented in Table 1, the differences in monetary impacts of flood hazards across the various climate scenarios for 2050 are relatively small: expected damage values associated with climate scenario SSP 5-8.5 are estimated to be almost 3,6% higher than the expected costs related to climate scenario SSP 1-2.6. This is due to the fact that the flood risk does not affect additional buildings in the most extreme climate scenario compared to the least extreme climate scenario. However, although the number of buildings hit remains the same in the different climate projections, namely 54, the damage costs are expected to increase due to higher levels of inundation, which can cause additional damage to the houses hit. The difference in total damage between SSP 1-2.6 and SSP 5-8.5 is approximately 500,000 USD. Figure 6 indicates that in the SSP 1-2.6 scenario for the year 2050, 38 of the 54 buildings exposed to storm inundation are located in the neighbourhood of Belnem, which is significantly more compared to the other neighbourhoods, resulting in 3.58 USDm of the total damage of 13.8 USDm. The estimation that (at least) 38 out of the 876 (4.3%) buildings in Belnem will be hit in 2050, even in the least severe climate scenario, suggests that local flooding can cause significant property damage within a specific neighbourhood and will result in severe local impact, which can be disruptive for a whole neighbourhood on Bonaire. The damages found in Belnem demonstrate that, in order to prevent the damage and disruption caused by climate change, neighbourhoods along the coast, which are generally the most valuable, must be protected against flooding by 2050. Total inundated surface areaexcluding saliñas Figure 7 illustrates Bonaire's key infrastructure, such as essential buildings and roadways, in 2050 under two climate projections (SSP 1-2.6 and SSP 5-8.5). All of Bonaire's essential infrastructure that is vulnerable to climate-driven flood risks is located in the South and/or along the coast.  Local firefighters stated that it can be presumed that roads are still passable until they are flooded to a depth of 40 cm, as this is the largest depth through which a first responders' vehicle can still travel (Fire fighter, personal communication, May, 2022). When considering the impact on infrastructure such as roadways, even roads that are not immediately exposed to flood threats can be severed from the network, rendering the entire road inoperable. As shown in Fig. 7, all roads in the southern part of Bonaire will be unusable in 2050 under climate scenarios SSP 1-2.6 and SSP 5-8.5. Therefore, emergency services cannot reach the Cargill facilities and other buildings in these areas of Bonaire. In addition, the roads leading to the salt ponds are reported as impassable, posing a severe threat to Cargill's operations. Moreover, the inundation depth of salt ponds might fluctuate following a coastal flooding event, and it is unknown how this affects the services provided by the ecosystems, such as salt provision. Figure 8 presents the cultural heritage that has been identified by literature, cultural experts, Bonairian residents, and/or the social media analysis.

Impacts on Tangible Cultural Heritage
As shown in Fig. 8, some locations have been identified in all three analyses, while others have only been identified by one of the applied research methods. For example, the literature review identified multiple locations around Rincon (e.g., 10, 12, 13, & 14 on the map) and traditional buildings in Kralendijk (e.g.,21,22,& 23), while these were not mentioned during the fieldwork. The participatory mapping, on the other hand, revealed locations that are important for fishing (19 & 20) and for recreation (24 & 25). The social media analysis features cultural heritage around Washington Slagbaai and mostly Boka Slagbaai, Kralendijk, the salt ponds, the slave huts, Willemstoren, and Lac Bay, thereby validating the literature review and fieldwork results.
The majority of Bonaire's tangible cultural heritage is located in coastal, mostly low-lying areas and is vulnerable to inundation from SLR and storm surges. As shown in Fig. 9, besides the Gotomeer, in 2050 no identified tangible cultural heritage in the North is identified as vulnerable to permanent or storm flooding, but almost all identified tangible cultural heritage in the South is considered susceptible already under climate scenario SSP 1-2.6, including the white and red slave huts, Flamingo sanctuary, Willemstoren, and old salt pans.

Expected Changes in Tourism Demand
Climate change will have negative effects on the health of coral reefs on Bonaire, which could lead to changes in the quality of dive sites and the social carrying capacity of these dive sites to accommodate dive tourism. Table 2 presents an overview of the change in coral reef quality and the expected diving demand and supply of quality dive sites in 2050 under each IPCC AR6 scenario. It is estimated that Scenario SSP1-2.6 shows a positive resource balance, since the coral reefs are expected to have the capacity to carry all diver demand. However, SSP2-4.5 displays a deficit of 174 thousand dives in the resource balance. The decline in quality dive spots decreases the social carrying capacity to a point where the dive demand exceeds the actual supply of dives. Consequently, over 17 thousand dive tourists cannot be accommodated on Bonaire's coral reefs in 2050 under scenario SSP2-4.5. This deficit progressively increases in scenario SSP5-8.5 1 3 (117.6 thousand dive tourists). If it is assumed that the dive tourists that cannot be accommodated on Bonaire's coral reefs will not visit the island, this would have an effect on the tourism expenditures and thus on total tourism demand, as shown in Table 2. While in SSP1-2.6 no change in tourism demand is expected, a reduction of over 205 USDm is expected in SSP5-8.5. Table 3 summarises the macroeconomic effects of a change in tourism demand as a result of reef degradation on Bonaire. The indirect economic consequences are derived by feeding the declines in final tourism demand in each sector into the induced Leontief coefficients of the added value per sector. As determined above, scenario SSP 1-2.6 does not result in any decline in tourism demand. Hence, no indirect economic consequences of reef degradation are found. In scenario SSP 2-4.5, the decline of 30.3 USDm in final tourism expenditures is divided among the relevant industries according to the related expenditure categories. Subsequently, this study finds that GDP is expected to shrink by 25.7 USDm. Considering "high growth", "business-as-usual", and "low growth" scenarios for GDP development, Bonaire is expected to potentially experience an economic contraction of 2.4%, 2.7%, and 3.8%, respectively. As

GDP contracƟon (USDm)
Climate scenario % of diver tourists whose dive demand is not met and will not visit Bonaire All (100%) -study's assumpƟon 50% 75% 125% Fig. 10 Bar chart presenting the results of a sensitivity analysis on the effects of different tourism responses to coral reef deterioration on Bonaire's GDP   shows the most severe indirect economic losses. Total indirect economic losses amount to 173.9 USDm, or an economic contraction of between -16.3% and -25.4%, depending on the economic growth scenario. The tourism sectors such as "Hotels", "Food services", and "Other real estate activities" are impacted most in each scenario. For instance, in scenario SSP 5-8.5, the added value of the "Hotels" sector is expected to decrease with 28.7% compared to the baseline (Table 3). The indirect consequences of tourism spending also become evident in other related industries such as the "Trade", "Financial intermediation", and "Other services" industries, where the latter is expected to experience a downfall in gross value added of 24.0 USDm in the worst-case scenario. All taken into account, the GDP multiplier of tourism spending amounts to 0.85x, indicating that a reduction in tourism demand of 1 USDm leads to a reduction of 0.85 USDm in Bonaire's GDP.

Sensitivity Analysis on Annual Tourism Demand
The data in Table 2 are based on the assumption that all scuba divers whose demand for diving cannot be met due to degraded dive sites will not visit the island. This assumption is founded on the fact that Bonaire is often referred to as "diver's paradise" and the majority of tourists visit the island for diving (Tourism Corporation Bonaire 2008; Wolfs et al. 2015). In addition, Sookram (2009) states that reef deterioration in the Caribbean has a substantial impact on the number of tourist arrivals, confirming the strong dependence of tourism revenues on the quality of dive sites. However, a multitude of other factors influence tourism demand, such as altering policies, the availability of other tourist attractions and activities on Bonaire and alternative holiday destinations, and the island's reputation, among various other factors. To test the impact of this assumption, we conducted a sensitivity analysis with three alternative potential tourism demand effects on coral reef degradation and associated GDP implications: 50%, 75%, and 125% of the dive tourists who cannot be accommodated on Bonaire's coral reefs will not visit the island (see Fig. 10). As expected, although the changes in GDP contractions under climate scenario SSP 2-4.5 are modest-ranging from 13 to 32 USDm depending on low to high tourism demand responses on degraded coral reefs-the difference under climate scenario SSP 8-5.8 is substantial and accounts for 130 USDm-contractions in GDP range from 87 to 217 USDm, depending on the effects on tourism demand (see Appendix E for more detailed results).

Discussion
The projections show that, similar to various other (Caribbean) islands (Carabine and Dupar 2014;Macpherson and Akpinar-Elci 2013;Robinson 2018;Scandurra et al. 2018), Bonaire will face several severe risks and disruptions under climate change. By 2050, even under climate projection SSP 1-2.6, which corresponds with a temperature increase of less than 2 degrees by the end of this century, as agreed in the Paris climate agreement, critical assets such as buildings, roads, and tangible cultural heritage are at risk of coastal inundation. Zellentin (2015) states that the citizens of SIDS will not only lose their physical homeland but also their social structure and cultural community due to anthropogenic climate change. Our study finds that various cultural assets, such as the slave huts, on Bonaire are at risk. The expected loss of this cultural heritage may have severe impacts on society, as it may lead to a decline in cultural identity and social cohesion. As opposed to findings presented in studies on other Caribbean islands (e.g., Monioudi et al. 2018;Wang 2020), Bonaire's international transportation airport and seaport, which  Schep et al. 2013). Similar to a study by Monioudi et al. (2018), it was found that SLR is the largest contributor to coastal inundation. While the expected differences in SLR between the selected climate scenarios are minimal, the differences in inundation, and consequently the accompanying negative effects, will be much more evident later in the century since the SLR variances between the various climate scenarios are much greater (IPCC 2021). Although coral reef degradation and coastal inundation are dependent on the selection of wave setup and loss of energy, the inundation results show that sensitivity is low.
The overall reef health index declines under climate scenarios SSP2-4.5 and SSP5-8.5 by 2050. Although these results are in line with previously published studies stating that climate change threatens coral reefs (Hoegh-Guldberg and Bruno 2010;Hoegh-Guldberg et al. 2017;Pandolfi et al. 2011;Crabbe 2008), it must be noted that the coral reef's health is expected to remain stable under SSP1-2.6 by 2050, indicating the importance of (global) climate change mitigation efforts. This loss of coral reefs will not only further amplify storm inundation and thus lead to an increase in the inundated areas and inundation depth, but in the most pessimistic scenario, Bonaire could also experience an economic contraction of 25%, which would have significant socio-economic implications for the island's inhabitants. However, the predicted economic contraction is considered susceptible to various factors, including the selected climate scenario and the assumed tourism responses to coral reef degradation, among others.
Nevertheless, it is evident that in the absence of timely planning and implementation of adaptation measures, the impacts of climate change may have serious implications for inhabitants' lifestyles and wellbeing. This study can serve as a foundation for stakeholders to develop and implement effective responses. We strongly recommend that further steps be taken to protect the Bonairians against the negative effects associated with climate change. For example, the identified inundation areas need to be taken into account in developing future zoning and construction plans. Moreover, the local authorities should implement coastal adaptation measures to protect at-risk communities and tangible cultural heritage from coastal flooding hazards. In addition, we recommend that future research concentrate on other ecological climate change-induced alterations, such as erosion and drought, as well as appropriate adaptation strategies against those changes.
While our study provides a holistic and unique view of the effects and impacts of climate change on a small Caribbean island, this study has limitations that are typical for modelling environmental, economic, and social interrelationships and integrated climate assessments. To start, although research on anthropogenic climate change is now well established (IPCC 2021), climate change modelling comes with various uncertainties. Ultimately, any model that uses numerical modelling comes with different sources of uncertainty (Foley 2010). Since climate systems and the models used to determine their direct risks to communities and the economy are composed of numerous complex processes and interactions, no model can ever be expected to perfectly simulate this. Consequently, uncertainties are inherent to these integrated modelling methods, and additional future sensitivity assessments will be necessary to determine how this may impact the conclusions of this study. Furthermore, as is common for most small islands (Van Beukering et al. 2007), the absence of certain types of (high-quality) data is a limitation that contributed to the levels of uncertainty in all the components of the study. For example, in the current study, modelled storm tracks were applied due to a lack of historical data. Moreover, the FABDEM dataset, based on satellite data, applied in this study has a spatial resolution of 30 m. Accurate elevation data are crucial inputs for geosciences and other disciplines (Hawker et al. 2022). For example, Vousdoukas et al. (2018) showed that a change in resolution from 10 to 100 m could change the estimated expected annual damages by 200%. A lack of data affects the IO model as well, since Bonaire fundamentally lacks publicly available and contemporary economic data .
The year 2017 was used as the base year in the model, as it was found to have the most complete dataset in terms of economic and ecological information of Bonaire. However, multiple variables had to be retrieved from earlier years and corrected or estimated for 2017 to fill in gaps in the data. Diverse and up to date datasets can be vital to further strengthening the predictive power of the ecological module and the IO model that was developed in this study. Moreover, assumptions had to be made concerning tourism responses to changing (quality of) diving sites, which appeared to have significant implications for projected GDP contractions.
It is furthermore important to acknowledge that the study utilised a closed-economy IO table for Bonaire, although the island's economy is open in practise. Data scarcity also prevented the development of an open-economy IO modelling framework, which implies that certain trends with diverse effects on Bonaire's economy could not be included in the model. For instance, the development of the tourism industry on Bonaire is dependent on trends in international tourism demand. On the one hand, factors such as population growth and income growth could lead to an increase in international tourism demand. In theory, an increase in international demand would, in an open economy, mitigate the anticipated GDP contraction in our results. On the other hand, altered climate change policies may increase future cruise-and/or airfare prices, which may reduce demand for tourism destinations that can solely be reached through air or sea transportation. Such policies are anticipated to exacerbate the predicted GDP contraction on Bonaire. In addition, when considering an open economy, the attractiveness of competing tourist destinations should also be considered, since the effect of a change in coral reef quality on the number of dive visitors also depends on supply of quality dive sites in competitive tourism destinations. As dive sites in other destinations are also expected to degrade, Bonaire's competitive advantage as a dive destination may not be affected. However, there's also a possibility that a global decline in coral reef quality will have a negative impact on the global demand for dive tourism and, consequently, also on tourism demand on Bonaire. Overall, it is anticipated that the various influences of an open economy framework will have an effect on tourism demand and related economic growth, but it remains uncertain how the trends will manifest in practise.
The various information gaps necessitated improvisation in obtaining credible evidence by employing alternate data collection techniques. Such methods included participatory mapping to identify cultural heritage on Bonaire, neighbourhood field sampling to validate infrastructural information and the analysis of social media information to identify ecosystem services hotspots on the island. As these methods are generally less costly to implement, they are suitable for application in a small island setting which is characterised by limited funds and capacity. When conducting research in data-scarce regions, such as various Caribbean islands, researchers could employ similar innovative data collection methods and assumption approaches to fill data and knowledge gaps.

Conclusions
This study aimed to explore the socio-economic impacts of climate change on Bonaire. First, we analysed the biophysical effects of climate change under different climate scenarios in 2050, namely coral reef degradation and inundation. This information was subsequently applied to determine socio-economic impacts associated with climate change, such as damage to the built environment and tangible cultural heritage, as well as to economic development as a consequence of a reduction in tourism. These impacts have all been studied in previous literature and were found to be impacted by climate change. However, this is the first study that analyses all of the aforementioned components in one research endeavour to create a more holistic picture of the impacts of global warming.
On Bonaire, coastal inundation is predicted to damage the built environment, infrastructure, and tangible cultural heritage. Moreover, the degradation of coral reefs will not only exacerbate storm flooding but will also have detrimental effects on Bonaire's economic growth. The findings of this study are not only applicable and relevant for Bonaire but may also be transferred to other Dutch and non-Dutch Caribbean islands. Nevertheless, recognizing geographical and cultural diversity, it is recommended that other low-lying SIDS also evaluate the impacts, vulnerabilities, and adaptation options against climate change hazards through primary research. The type and variation of research methods employed in this research could serve as a basis or inspiration for various other data scarce islands threatened by the effects of climate change.
Our results suggest that direct action is needed to protect the island of Bonaire and its inhabitants against the implications of climate change. In the absence of timely planning and implementation of adaptation measures, the impacts of climate change may have serious implications for inhabitants' lifestyle and wellbeing. Based on these results, it is evident that further research on potential adaptation measures, as well as immediate action to develop and implement these types of measures, are urgently needed.

Appendix C Elaboration on the CRAS Approach
The economic structure of Bonaire in 2050 is predicted by updating the developed IO matrix of 2017 using a CRAS approach, as proposed by Lenzen et al. (2009). The CRASalgorithm (cell-corrected RAS-algorithm) is an adoption of the original RAS-algorithm that was developed to update IO tables by Stone (1961). In his method, an old matrix A, with row sum i and column sum i is transformed to a 'new' matrix Y that satisfies a 'new' set of given row sums and column sums with minimum loss of information (Stone 1961). However, the original RAS method only works when the start matrix A solely has non-negative entries (Bacharach 1970). Hence, Junius and Oosterhaven (2003) developed a general mathematical device that can be applied to original matrices with both negative and positive entries, which was later corrected by Lenzen et al. (2009) for additional issues, for example, when handling zeros as initial entries. In addition, CRAS is also able to compromise between inconsistent matrix entries. This is especially helpful since the 'new' set of constraints for 2050 are obtained by making assumptions on sectoral and GDP growth for 2050. Three economic scenarios are developed for Bonaire: a (1) high growth scenario, a (2) business-as-usual (BAU) scenario, and a (3) low growth scenario. Again, the economic scenarios start in the base year 2017 and run until 2050. The GDP growth rates in every scenario are corrected for inflation to translate the economic impact of reef degradation on output to 2017 prices. The problem that needs to be resolved is to find a 'new' matrix Y that deviates least from the original matrix A, given the set of new row and column constraints, row vector n and column vector n . The matrix entries should be chosen in order to satisfy the following target function (Wiedmann et al. 2007): where target function t' is minimized, given the constraints that the sums of all sector outputs x i,j are equal to the column and row totals of the IO matrix.

Appendix D Elaboration on the Process of Estimating Total Revenue Losses
First, the annual number of visitors to Bonaire was proxied using the resource balance. This paper assumes that divers will not visit the island if their dive demand cannot be met. In the case of a deficit in the resource balance, the demand for dives exceeds the supply. Consequently, not all divers that are willing to visit Bonaire can be facilitated. Assuming that every diver demands ten dives per stay (Council Underwater Resort Operators, personal communication, May 2022), the number of divers that will not visit Bonaire could be calculated. For instance, when the resource balance equals -100 thousand dives in 2050, the demand of 10 thousand divers cannot be satisfied. Therefore, total tourism arrivals decline by 10 thousand on the island. However, the type of tourist arrivals should also be differentiated since stay-over tourists and cruise tourists have different expenditure patterns. (1) According to Mercera, CEO of the Tourism Corporation Bonaire (TCB), Bonaire is planning to limit stay-over arrivals to a maximum of 250 thousand a year to prevent mass tourism on the island and to remain a unique destination (Nederhof 2022). Assuming a similar growth rate of stay-over-arrivals as in the period of 2001 until 2019, it is expected that Bonaire will have reached the maximum amount of 250 thousand stay-over tourists by 2029. Simultaneously, cruise tourism is expected to be restricted to focus more on highend tourism (Croes et al. 2019). Cruise ships will most likely be limited to one ship a day, as suggested by Nederhof (2022). Since cruise ships generally only visit Bonaire between October and April, and assuming an average capacity per ship of 1224 passengers (Wolfs et al. 2015), this paper predicts a maximum of 259.5 thousand cruise tourist arrivals by 2050. As mentioned in Sect. 3.3, it is assumed that 51.26% of total stay-over tourists and 4.35% of cruise tourists are divers. Employing these rates to the preceding example of a shortage of 10 thousand divers, we would find a decline of 9.2 thousand stay-over tourists arrivals and 0.8 thousand cruise tourist arrivals.
The decline in stay-over and cruise tourist arrivals affects the IO table through the final tourism demand. In turn, final tourism demand is based on the average tourist spending, "the total consumption expenditure made by a visitor on behalf of a visitor for and during his/her trip and stay at destination" (UN 1994, p. 21). A decline in X amount of stay-over tourists on Bonaire would result in total tourist expenditures that are lowered by X multiplied by the average expenditures of a stay-over tourist (i.e.1,886 USD per stay). Similarly, a decline in Y amount of cruise tourists would result in a final demand that is lowered by Y multiplied by the average added value of a cruise tourist (i.e., 122.13 USD per stay). The tourism revenue loss per category is matched to the impacted industry in the IO table. Due to the multiplier effect, variances in the final demand of a specific industry will result in 'spill-over' effects to other industries. To calculate the change in the total output of each sector, a Leontief matrix needs to be constructed. This matrix provides the technical coefficients necessary to estimate changes in intermediate demand between industries when the final demand changes. Subsequently, the sum of all aggregated changes in intermediate demand reflects the level of changes in the total output of Bonaire. The last step to finding the total macroeconomic impact of reef degradation is to estimate changes in the GDP of Bonaire by 2050.
where GDP is the gross domestic product, GVA denotes the aggregated gross value added of all industries in Bonaire, TP describes the taxes on products, and SP defines the subsidies on products. Similar to the changes in intermediate demand, the changes in GVA, TP, and SP are estimated using the Leontief matrix. The coefficients of a sector's GVA (CVAi) are obtained by dividing the value added per sector (VAi) by the total output of the sector (Xi) (Furtuoso and Guilhoto 2000). Subsequently, the CVAi are multiplied by the Leontief matrix (I -A)-1 to obtain the technical coefficients TVAi. Since total output Xi decreases when the total intermediate demand decreases, changes in the GVA can be calculated by taking the sum of TVAi multiplied by the changes in total output Xi of each sector. Hence, Eq. (4) can be developed.
Likewise, changes in TP and SP are determined by calculating the corresponding technical coefficients TTPi and TSPi and multiplying them by the change in total output Xi of each industry: Hence, if the coral reefs around Bonaire degrade, causing direct tourism demand in some sectors to decline, both direct and indirect economic consequences are expected to occur. Using the Leontief multipliers, the induced changes in intermediate demand between all sectors can be determined. The resulting decrease in total output per sector can subsequently be used to assess the changes in GDP, using Eqs. (3), (4), and (5).
Author Contributions All authors contributed to the article and approved the submitted version.

Data Availability
The data that support the findings of this study are available from the authors upon reasonable request.

Competing interests
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.