Linking coral reef locations to seafloor geodiversity in the Dutch Caribbean

(1)

Linking coral reef locations to seafloor

geodiversity in the Dutch Caribbean

Bachelor Thesis

Daniël Peters

Supervisor: Mr. Dr. Harry Seijmonsbergen BSc Future planet studies

Major: Earth Sciences

(2)

(3)

Abstract

Geodiversity (the diversity of earth's abiotic features) plays a crucial role in supporting global ecosystem services and biodiversity (Zarnetske et al., 2019). While there is evidence that this also applies to the marine environment (Gordon 2020; Fisher et al., 2019), no previous studies examined the relations between geodiversity variables and coral reef locations. This study aims to develop a GIS-based method to quantify seafloor geodiversity for the Saba Bank and St. Eustatius. Links between geodiversity and coral reef locations were analyzed to gain a better understanding of the relationship between abiotic seafloor features and Caribbean coral reef ecosystems. An external grid overlay was used to create a geodiversity index, based on a bathymetric dataset in ArcGIS. The sum of the sub-indices (bathymetric diversity, roughness diversity, light penetration) resulted in the final geodiversity index map. The results show that geodiversity accurately predicts coral reef locations at the Saba Bank and coral reefs surrounding Saba Island. The majority of coral reefs overlap with moderate to high geodiversity scores. There are no significant observed differences in geodiversity between coral ecosystems and other marine habitats for the marine environments of St Eustatius. Future research should focus on improving the light penetration index by incorporating local turbidity factors into the seafloor geodiversity index. This study's findings demonstrate that there is great potential for implementing geodiversity assessments in coral reef conservation on a global scale.

(4)

1 Introduction

Global coral reef populations are declining at an alarming rate. One-third of global coral reefs face higher extinction risk through the impacts of climate change and additional local factors (Carpenter et al. 2008) such as but not limited too; coral bleaching, unsustainable fishing (Spalding et al. 2017), pollution from increased tourism and urban waste. There is currently ongoing debate on the effectiveness of the quantification and regulation of Marine Protected Areas in tropical coral reef environments (Rudolph et al., 2020). Coral reef conservation methods (Foo et al., 2019; Hedley et al., 2016) are developed that integrate geoinformatics and spatial data analyses. Such developments support holistic and integrative approaches by scientists and policymakers to protect reefs, which is necessary to safeguard these vital ecosystems.

The decline of coral reefs is negatively impacting the marine environment. Although coral reefs cover less than 0.35 percent of the ocean’s seafloor, more than 25 % of marine life relies on healthy coral ecosystems (Mulhal, 2008). In particular, coral reefs provide habitat and shelter for a wide range of marine organisms and play a buffering role in coastal protection through the absorption of wave energy (Hoegh-Guldberg, 2011). Approximately one billion people worldwide rely on coral reefs for food and income from fishing and tourism. Coral reefs provide essential ecosystem services to society (Woodhead et al. 2019) which should be managed carefully in the light of the sustainable development goals (Convention on Biological Diversity).

Marine biologists have traditionally dominated research on coral reefs. Geoscientists, however, claim that abiotic nature – or geodiversity, "the natural range (diversity) of geological (rocks, minerals, fossils), geomorphological (landforms, processes), and soil features. It includes their assemblages, relationships, properties, interpretations, and systems" (Gray, 2012) has been neglected in biodiversity- focused approaches on conservation of Marine Protected Areas (Crofts 2014; Gordon,2020). Abiotic seafloor features (e.g., bathymetry, geomorphology) must be considered during the qualification of Marine Protected Areas (MPAs), as they proved to be good predictors for biodiverse marine areas (Fisher et al.,2019; Harris et 2013). Diverse and resilient tropical coral reefs can only develop under a set of habitat-specific conditions, strongly related to the benthic environment's physical properties. Access to sunlight, temperature, and a suitable hard substrate to settle are vital abiotic factors, necessary for coral reef development (Chappel, 1980). Geodiversity assessments should, therefore, be integrated into coral reef research to support their conservation and management.

Seafloor geodiversity has not been quantified in Caribbean coral reef ecosystems. Novel applications to restore coral reefs are already being field-tested (Meesters et al., 2015), showcasing the importance of geoscientific research in marine environments. The construction of biodiversity indices for terrestrial habitats regularly incorporates geodiversity aspects (Hjort & Luoto,2010; Zarnetske et al., 2019), thus considering the role of microclimates for species diversity. To test for the merits of geodiversity assessments in marine conservation sciences requires a coral reef ecosystem, mostly unaffected by confounding anthropogenic factors. To this end, the Saba Bank marine environment provides a suitable case-study because it remains mostly free of direct pollution or degradation by human activities (de Baker et al., 2017). The present study will develop a seafloor geodiversity index of the Saba Bank and the marine environment of St. Eustatius. In order to analyze the links between geodiversity and coral reef cover, bathymetric seafloor diversity alongside roughness diversity and light penetration are incorporated into the geodiversity assessment. Seafloor geodiversity could serve as a suitable surrogate for coral reef cover trough the integration of bathymetric data layers in ArcGIS. The research aims to develop a GIS-based method for the quantification of seafloor geodiversity. The development of an index-based geodiversity assessment will open up opportunities to better understand the abiotic influence on coral reef ecosystems. To further assess the relations between seafloor geodiversity and coral reef cover the following research questions will be answered:

(7)

• What data layers are necessary to develop a seafloor geodiversity index?

• Is seafloor geodiversity a good indicator of the presence of coral reef

ecosystems?

(8)

1.2 Study Area

Figure 1. The location of the study area in the Dutch Caribbean, including Marine Protected Areas.

The study area (Figure1) is located in the northwestern part of the Caribbean Sea. The research area is divided into two neighboring areas. The box at the left (Figure1) covers the western part of The Saba Bank marine park. The other sub-research area is located approximately 20 km west from the Saba Bank and contains the marine environment of St Eustatius.

The Saba Bank and St. Eustatius are part of the Caribbean global biodiversity hotspot area (Hoeksema et al., 2017). Past Volcanic activity shaped the terrestrial and marine environments of the Dutch Caribbean. Most of the coral reefs in the research area developed at paleo-volcanic structures. Partly through the bathymetric complexity, a wide range of coral reefs are present (Debrot et al.,

2018; Trembanis et al., 2017).

The Saba Bank contains the largest Marine Protected Area in the Kingdom of the Netherlands and is considered as the biggest atoll in the Caribbean Sea. The submerged atoll is separated by the deep ocean and is, therefore, through its isolated location relatively unaffected by land induced processes (e.g., sediment influx, eutrophication, pollution) (de Bakker et al., 2017).

The volcanic island St. Eustatius is surrounded by a well-developed, biologically rich patch coral reef stretching up to 3 km from the shore. Many of the coral reefs are located on large volcanic boulders, which originate from the Quill Volcano's past eruptions. In contrast to the Saba Bank, coral reefs on St. Eustatius are affected by pollution, high sediment influx, and other human-induced local processes (Debrot et al., 2014).

Three MPA’s (The Saba National Marine Park, Saba National marine park, St. Eustatius Marine Park) are located within the research area (Figure 1). Despite conservation efforts, coral coverage declined by more than 50% over the past 40 years (Debrot et al., 2018). At present, geodiversity is unknown within these MPAs. Getting a better understanding of ocean geodiversity within MPAs thereby contributes to more effective marine conservation strategies.

(9)

2 Methodology

The methodology contains two sections; the first section elaborates on the methods used to calculate a seafloor geodiversity index. The seconds part describes how correlations are made between coral cover and the geodiversity index. For this research, ArcGIS Pro (Esri, Inc) was used to calculate the geodiversity index. Data management and statistical analyses were carried out in R Core Team (2018), Vienna. The geodiversity calculation research will follow the workflow (Figure 2). This workflow contains three distinct phases of processing, analysis, and final results, which are included in the deliverables. Links between geodiversity and coral reef cover, are examined. All ArcGIS executions are also carried out in the ArcGIS model builder. (A built-in visual programming language which enables to automate geoprocessing workflows)

Figure 2. Workflow diagram of the research.

Proc ess ing A nal ys is D el iv er abl es (map s & st at ist ic s)

Thematic & digital data collection

Pre-processing of input datasets

Design of Geodiversity index & Calculation

Reclassification & statistics Sub- indices Calculation

Visualize geodiversity maps/tables

Interpretations of Geodiversity & Coral reef locations (ARCGIS & Rstudio)

Grid definition

(10)

2.1 Thematic & digital data collection

Datasets regarding seafloor geodiversity, including coral reef locations, were collected for the research area. The Datasets were selected based on availability, quality, and relevance. All of the datasets were downloaded in the World Geodic system (WGS84). The collected datasets were retrieved from freely available sources (Table 1).

Table 1. Sources of input datasets including year produced Dataset Source dataset Year Scale/ Cell

size Type Bathymetry Saba Bank Extracted from contour Saba Bank 2014 250m Raster Bathymetry St. Eustatius Bathymetry St. Eustatius, Kramer (2016) 2016 2,4 m Raster Contour map of the Saba Bank

DCBD, Dijkman (2012) 2012 Regional (Study area) Polygon Slope map Saba Bank Eustatius Calculated from a contour map 2012 250m Raster Slope map St Eustatius Calculated from Bathymetric dataset 2016 10m Raster Light penetration Saba Bank Calculated from Bathymetric dataset 2012 250 m Raster

MPA Marine park map of St. Eustatius 2013 Regional (Study Area) Polygon Marine Habitat map Marine habitat map St. Eustatius, van den Broek (2014) 2014 Regional (Study Area) Polygon Vulnerable Ares BES Vulnerable areas BES islands, van den Broek (2014)

2014 Regional (Study Area)

Polygon

Coral reef map Global distribution Coral reefs(UNEP)

(11)

2.2 Pre-processing of input datasets

Before the actual geodiversity can be calculated, the datasets must be pre-processed. The datasets (Table 1) are loaded into ArcGIS Pro. Since the Saba Bank and St. Eustatius have different input data sets, the pre-processing is slightly different for both areas. The bathymetric source datasets of Saba and St. Eustatius contained an unconventional cell size and were therefore changed to respectively 250m and 5 m by applying the resample tool. The bathymetry dataset for the Saba Bank was derived from a contour map using the Topo to Raster tool in ArcGIS pro. For St. Eustatius, the bathymetry was already available. The raster calculator was used to divide the bathymetry dataset in areas above sea level and below sea-level. After converging the raster to polygons using the raster to polygon tool, areas above sea-level were deleted from the research area using the delete features tool. Both input datasets were subsequently clipped into the research area. The slope tool in ArcGIS calculated the Slope map for both areas.

2.3 Grid definition

The create fishnet tool was used to compile a grid, required to calculate the geodiversity index scores. Since St. Eustatius's research area is relatively small with respect to The Saba Bank and considering the scale size of the input datasets, two separate grids were created. For the Saba Bank, a grid cell size of 1000 m and a 100 m grid cell size for St. Eustatius. The grids were both clipped within the research area. The bathymetric dataset of St. Eustatius contained a few empty cells which were deleted from the fishnet grid.

2.4 Design of Geodiversity index & Calculation

The Geodiversity index score is the sum of sub-indices which define together the total geodiversity of the seafloor. These variables were adapted for seafloor geodiversity but based on previously developed methods for terrestrial geodiversity quantification (Perreira et al., 2013; Seijmonsbergen et al., 2017), adjusted to the aim of this research. The following formula (1.1) was used to calculate the seafloor Geodiversity index:

Total GDi = Bdi + Rdi+ Lpi (1.1.)

Where: GDi = Geodiversity Index, Bdi = Bathymetric diversity index, Rdi – Roughness diversity index, Lpi = Light penetration index.

(12)

2.5 Sub- indices Calculation

The bathymetric diversity and roughness diversity were calculated using the zonal statistics tool. The Standard deviation of the bathymetric dataset resulted in the bathymetric diversity using the external grid overlay. The Roughness diversity index was calculated similarly by calculating the slope's standard deviation using the zonal statistics tool. Both sub-indices were set to 5 classes using the reclassify tool and were subsequently classified according to Jenks natural breaks.

Beer-Lambert law: 𝐼 = 𝐼𝑜 ∗ 𝑒𝑥𝑝 (−𝑑/𝑙) (1.2) Where I = Light intensity at seafloor, Io = Light intensity at the sea surface, d = Depth, l =

constant for seawater (Wozniak & Deera , 2007)

The light density index was calculated using the beer lambert law (1.2). First, the light radiation at the surface was calculated by the '’area solar radiation tool’’ in ArcGIS Pro. The output value raster containing the surface radiation was added to the raster calculator. The bathymetric dataset (ocean depth) was used as an input value for the Beer-Lambert law. The formula calculates the solar radiation at the bottom, considering the light absorption of the seawater. Thereafter the mean surface light penetration for every grid cell was calculated with the zonal statistics tool. The “MEAN”- option was chosen, which after reclassification resulted in the light density index.

2.6 Geodiversity calculation

The geodiversity calculation was executed in the raster calculator by summing al sub-indices

according to formula (1.1). Subsequently, the total geodiversity scores (table 2.1, table 2.1) were then reclassified, according to Jenks natural breaks classification method (Jenks, 1967). Five ascending classes., very low, low, moderate, high, very high, were created.

(13)

Table 2.1 Geodiversity index reclassification Saba.

Geodiversity 1 2 3 4 5

Value 6 – 7,4 7,4 – 8,8 8,8 – 10,2 10,2 – 11,6 11,6 - 13

Table 2.2 Geodiversity index reclassification St. Eustatius.

Geodiversity 1 2 3 4 5

Value 3 – 5,2 5,2 – 7,4 7,4 – 9,6 9,6 – 11,8 11,8 - 14

2.7 Analysis geodiversity

Correlation matrices for Saba and St. Eustatius were created using the band collection statistics tool in ArcGIS Pro. The correlation matrices (Table 4 & 5) contain a number between zero and one,

indicating the linear dependency between the Gdi and the sub-indices. Furthermore, the frequency tables (6.1 & 6.2) of the geodiversity scores were created.

2.8 Coral cover and geodiversity analyses

For the analyses of coral reef cover and the seafloor geodiversity index, a combination of ArcGIS and R Core Team (2018), was used. Since we are dealing with different input datasets, the coral –

geodiversity analysis of Saba and St. Eustatius require a slightly different approach.

The Available coral data for the Saba study area was merged into a point feature class layer containing 23 coral reef points. For the Saba Bank, 30 random points without coral reefs were generated using the generate random points tool. For St. Eustatius, a random sample was generated of 30 points with coral reefs and 30 points without coral reefs using Rstudio. Different types of coral (diffuse patch reef, gorgonian reef) where all dived in binary data column indicating a 1 for coral and a 0 for non-coral habitat.

The extract multi values to points tool in ArcGIS was used to extracts cell values at locations specified in the coral point feature class from the geodiversity index and sub-indices rasters. The tool adds the values of Gdi, Bdi, Rdi to the attribute table of the corresponding point feature class containing coral reef data. The attribute table of coral reef points, including the Gdi and the sub-indices score, was subsequently loaded into Rstudio for both The Saba Bank and St. Eustatius.

After rearranging the data in Rstudio, boxplots were created to visualize the geodiversity scores of locations with and without coral for both areas. An unpaired t-test was carried out to compare the means of geodiversity scores and sub-indices of points situated at coral reefs and outside coral reefs. A significance level of 0.05 was used. In order to determine the predictive power of coral on geodiversity scores, a logistic regression model was carried out in Rstudio for both study areas. The script for the coral reef - geodiversity analysis is available via this link.

(14)

3 Results

In this section, the results of the research are presented. The first paragraph presents a general overview of the seafloor geodiversity patterns of the research areas. The geodiversity maps of both Saba (Figure 4) and St. Eustatius (Figure 5) are presented and briefly explained. The maps of the sub-indices can be found in appendix A. The second paragraph summarizes the outcomes of the coral reef- geodiversity analysis.

3.1 Geodiversity scores

In Figure 4, the geodiversity map of the Saba Bank and near surroundings are visualized. A clear pattern in geodiversity scores can be distinguished. Moderate-, high-, very high geodiversity scores are clustered in a half-circle shaped belt following the edge of the Saba Bank. A sharp transition towards a very low score is visible in the east. A small circle shape cluster of moderate-, high-, very high

geodiversity scores surround Saba island. Low geodiversity scores account for more than 90 % of the total area. Areas with a very high, high, or moderate geodiversity cover a marginal part of respectively 1% and 2,5% of the research area.

(15)

Seafloor geodiversity scores for St. Eustatius (Figure 5) have a completely different distribution compared to the Saba Bank. The vast majority (61%) of the area contains a moderate geodiversity. This is also shown in the frequency table (table 5) and the histogram (Figure 6.2) of St. Eustatius. Equal geodiversity scores are clustered and have a circle shape surrounding the island. The

predominant pattern is an increase from low geodiversity scores near the coast to higher geodiversity offshore.

Regarding the statistical results (table 3) for the Saba Bank, most of the geodiversity variation is explained by the Bdi and Lpi. Both sub-indices contain a strong positive correlation with the geodiversity score (Respectively 0.69 and 0.60). Gdi and Rdi have a moderate to low correlation; a correlation coefficient of 0.35 was calculated. Bdi and Lpi have a correlation of approximately zero (-0.08).

For St. Eustatius (table 4), both the Bdi and Rdi have a strong positive correlation with the Gdi (respectively 0.78 and 0.80). The Lpi is negatively correlated with the Gdi (-0.68). Since the

calculations are based on two different grid sizes for both areas, the frequency tables (table 5 & 6) and the histograms (Figure 6.1 & 6.2) have different magnitudes.

(16)

Table 3. Correlation matrix Saba Bank

Index Geodiversiy Roughness Bathymetric diversity Light Penetration Geodiversiy 1,00 Roughness 0,35806 1,00 Bathymetric diversity 0,69090 0,58877 1,00 Light Penetration 0,60567 -0,12485 -0,08854 1,00

Table 4. Correlation matrix St. Eustatius

Index Geodiversity Roughness Bathymetric diversity Light Penetration Geodiversity 1,00 Roughness diversity 0,79772 1,00000 Bathymetric diversity 0,78257 0,55660 1,00000 Light Penetration -0,63858 -0,66406 -0,85199 1,00000

Table 5. Frequency table Saba Bank & St Eustatius

Geodiversity Saba Frequency Saba Bank Relative Frequency St Eustatius St Eustatius Relative frequency Very low 18128 0,498968 30753 0,045746 Low 15611 0,429688 89386 0,132965 Moderate 1370 0,037709 410508 0,610645 High 904 0,024882 140091 0,20839 Very high 318 0,008753 1515 0,002254

(17)

Figure 6.1 Gdi frequency distribution Saba Bank Figure 6.2 Gdi Frequency distribution St Eustatius 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000

Very Low Low Medium High Very

High

Geodiversity Saba Bank

0 50000 100000 150000 200000 250000 300000 350000 400000 450000 Very Low

Low Moderate High Very High

(18)

3.2 Coral reef & geodiversity statistics

Saba Bank

The Boxplots below (Figure 7) demonstrate that the Saba Bank's geodiversity scores at coral reef locations are significantly higher. For locations with coral reefs (Figure 7A), 50% of the geodiversity scores lay between low and 4. In contrast to locations without coral cover where 50 % of Gdi scores are between 2 and 4. The Lpi for coral locations is more centered with respect to the Lpi of no coral locations. However, the mean of both Lpi scores are almost equal (Figure 7). Respectively 3.12 (coral area) and 3.15 (area without coral). The logistic regression model's output indicates that geodiversity is significantly associated with the probability of coral reef cover (p=0.0009). Also, the output of the unpaired t-test for geodiversity at coral locations and locations without coral resulted in a significant p-value (p= 6.792e-05).

Figure 7. Boxplots showing the geodiversity and sub-indices scores of Saba Bank for coral reef locations (A), locations without coral reefs (B), and the geodiversity of coral reef next to no coral reef locations(C).

C

(19)

St Eustatius

For St. Eustatius, there is no significant difference in geodiversity scores between coral reef locations and locations without coral reefs. The output of the unpaired t-test (p-value =0.58) is not significant. The logistic regression model's output indicates that geodiversity is not significantly associated with the probability of coral reef cover. The boxplots (Figure 9) below visualize the geodiversity and sub-indices of coral reef sites and sites without a coral reef.

Figure 8. Boxplots showing the geodiversity and sub-indices scores of St. Eustatius for coral reef locations (A), locations without coral reefs (B), and the geodiversity of coral reef next to no coral reef sites (C).

A B

(20)

4 Discussion

4.1 Interpretation of the results

We found for the Saba Bank that geodiversity is strongly correlated to coral reef cover for nearly all locations. This assessment demonstrates that coral reefs are predominantly situated in high

geodiversity regions. A significant p-value (p=0.0009) for the logistic regression model implies that geodiversity is a good predictor for coral reefs. The light penetration index of both coral and no coral sites are almost equal in areas with and without coral reefs. However, the variances differ significantly from 1.4 (Coral) to 3.4 (No coral). This results from the wide range of ocean depths, from relatively shallow (20m - 50 m) flat plateau of the Saba Bank to deep ocean, which reaches depths up to 600m. While a significant part of the Saba Bank has high to very high Lpi scores (Figure 10), only a select area is inhabited by coral. Stony and soft coral reef cover are most extensive at narrow southeastern belts where the Saba Bank is (10m- 15m) (van Haren et al., 2019). Furthermore, Lpi scores are

probably underestimating the real light penetration index as the Lpi does not distinguish the shallowest parts of the Bank. The symbiotic zooxanthellae can optimally use the light for photosynthesis essential for coral reef growth at this shallow belt.

Many coral reefs occur on complex irregular surfaces, resulting in coral reef species diversity. The resilience of dynamic coral reef ecosystems grows significantly (Chappel, 1980). This corresponds to the Saba bank's eastern rim, which contains a heterogeneous and irregular surface morphology (van der Land et al., 1977). The moderate to very high Rdi (Figure 11) and Bdi (Figure 9) scores at eastern flank demonstrate that complex surface topography overlap with the majority of coral reefs (Figure 12).

Another explanation for the coral reef cover, apart from previously mentioned light penetration, bathymetric diversity, and roughness diversity, might be the abundance of nutrients supplied by predominant south ocean eastern currents (van Haren et al.,2019). Furthermore, the geodiversity assessment, which was initially focused on the Saba Bank, also forms a good predictor for coral reefs surrounding Saba island. The submerged pinnacles 1.5 km west from the coast form ideal conditions for coral development (Debrot et al., 2018), as the geodiversity index map (Figure 7) and coral map (Figure 11) illustrate. For the Saba Bank area, a combination of bathymetric, roughness, and light penetration index are good predictors for coral reef areas. The results are in line with the hypothesis that areas with high geodiversity overlap with coral reefs.

While geodiversity serves as a good predictor for coral reefs at the Saba Bank, this is not the case for St Eustatius. The output of the logistic regression model (p = 0.38) is not significant. Most likely, due to a combination of methodological and data limitations, Gdi and sub-indices scores of coral and non-coral locations are nearly equal. In addition, the difference in the non-coral reef structure could be an explanation of varying results. St. Eustatius coral reef ecosystem exists predominantly of diffuse patch reefs (Debrot et al., 2014), which generally occur on a less rough substrate, surrounded by sand or rubble, making the calculations more prone to inaccurately pairing geodiversity scores to

corresponding coral points. Furthermore, a significant part (61%) contains moderate geodiversity index scores creating a more significant probability that coral and no coral reef habitat fall into the same geodiversity class.

(21)

4.2 Methodological discussion

Geodiversity quantification is strongly dependent on the quality and presence of input datasets. All used input datasets were derived from the bathymetric data of the research area. Due to limited seafloor data, it was impossible to include other abiotic seafloor datasets, which generally form the foundation for geodiversity quantifications. Most of the coral reefs' locations can be linked to past volcanic activity (de Bakker et al., 2017). Hence, analyzing coral cover, including seafloor geology, and geomorphology could be relevant. While the bathymetric dataset will not give an exact reflection on other input datasets, various bathymetric surface features often form an accurate proxy for different geological or geomorphological (Zaweda et al., 2010) units. The extent to which the absence of input datasets (Geomorphology, Geology) negatively influences the results of the geodiversity score is therefore limited.

Outcomes vary significantly between the Saba Bank and St. Eustatius. As two different grid sizes were used for the research areas' geodiversity calculations, comparisons are difficult to make. Determining the right grid size can lead to varying results in the geodiversity-coral analysis. For the Saba Bank, an external grid size of 1000m was used, and a 100 m grid size was used for St. Eustatius. This brings us to the first implication of the model. Coral point features were assigned a geodiversity score according to the corresponding grid cell. Whether a coral reef point feature is situated in Saba or St. Eustatius determines if the coral reef point is assigned a geodiversity value of 1000m x1000m or a 100x100m grid cell. Indicating that errors occur more often while assigning a geodiversity value in St Eustatius. A way to reduce this error could be through measuring the mean geodiversity of the neighboring cell values.

Furthermore, a gap in the bathymetric data possibly led to inaccurate results for St. Eustatius. In the western part of the St. Eustatius marine area, where diverse patch reef is abundant (de Graaf et al., 2015), no Gdi scores could be calculated due to a lack of bathymetric data. This imperfection could be partly the reason for varying geodiversity scores at coral and no coral locations for St. Eustatius.

4.3 Future Research

The results of this research in the Dutch Caribbean show an excellent opportunity for upscaling to global seafloor geodiversity applications. Due to improved computing and geoinformatics software, future assessments using a similar methodology can be made on a global scale in the future. Further research is suggested on analyzing global seafloor geodiversity using one global bathymetric dataset, including the same sub-indices. This could reveal interesting insights into relations between global tropical coral reefs and geodiversity scores globally. In addition, this approach will avoid

methodological issues of the present research, where two bathymetric datasets and two separate external grid cells were used. However, before upscaling similar assessments, it is essential to acknowledge that such studies require high-resolution bathymetric datasets, which are currently limited. Only if less cost-intensive methods based on remote sensing are developing at a fast pace more accurate bathymetry datasets will be available in the near future (Pike et al., 2019).

As mentioned before, global coral reefs face many local treats. Coastal reefs of St. Eustatius are highly susceptible to increased sediment influx, and other land sourced pollution (Debrot et al., 2018). These factors generally increase ocean turbidity and limit solar light penetration (Carpenter et al., 2008). For this research, the light penetration index was based on the beer Lambert-law (1.2), which only

incorporates bathymetry. As a result, the light penetration index overestimates the light penetration. To enhance the accuracy of the Lpi, a remote sensing approach using satellite or air born sensors could be

(22)

valuable in coastal waters, including local turbidity factors (Wei et al., 2018). Landsat-8 satellite images based on radar technology can capture high-resolution light penetration images in coastal waters, including turbidity and light availability. In this manner, point source pollution or sediment influxes could be incorporated into the calculations, improving the light penetration index accuracy. Furthermore, the addition of an ocean temperature dataset could improve the geodiversity index, since ocean temperature forms another critical habitat condition for coral reef growth (Chappel, 1980). During the past decades, various coral bleaching events occurred in Caribbean coral reef ecosystems due to climate change (de Bakker et al., 2017). Considering coral reefs susceptibility to temperature fluctuations, an additional seawater temperature data layer in future research could improve the geodiversity index.

(23)

5 Conclusion

In this study, a seafloor geodiversity assessment was made for the Saba Bank and St Eustatius. The aim of the research was to develop a GIS-based method for the quantification of seafloor geodiversity. The correlation between coral cover and geodiversity was calculated to explore the potential for geodiversity-based management of coral reefs in the Dutch Caribbean. Areas with high geodiversity scores were hypothesized to contain high coral reef cover. The seafloor geodiversity score was calculated based on the following sub-indices: Bathymetric diversity, Roughness diversity, and light penetration. The results demonstrate that only a bathymetric data layer is needed to create an adequate seafloor geodiversity assessment for the nature of this study. For the Saba Bank and Saba island, we may conclude that a combination of bathymetric diversity index, roughness diversity index, and light penetration index are a good indicator for coral reef cover. St. Eustatius does not show significant differences in geodiversity and coral reef cover. Most likely, this was due to a limitation of input datasets. In order to produce a more accurate light penetration index, crucial for coral reef

management, it is recommended to investigate the potential to incorporate ocean turbidity in the light penetration index. In addition, including ocean temperature datasets into the geodiversity assessment could contribute to improving the prediction of coral reef cover and map effects of temperature fluctuations on coral reefs. Linking seafloor geodiversity to coral reef locations is a pioneering field of study. This study provides novel insights into patterns and relationships between seafloor geodiversity and the coral reefs in the Dutch Caribbean. The incorporation of geodiversity assessments could markedly improve marine conservation practices. However, further research is needed on optimizing the geodiversity index trough the incorporation of local turbidity factors and temperature datasets.

(24)

6 Acknowledgments

I would like to thank The Dutch Caribbean nature alliance for providing the available datasets. I am indebted to my supervisor Harry Seijmonsbergen for his support during this Bachelor thesis. Also, I would like to thank Bart Hoekstra for his assistance in data management and Rstudio.

(25)

7 References

De Bakker, D. M., Van Duyl, F. C., Bak, R. P., Nugues, M. M., Nieuwland, G., & Meesters, E. H. (2017). 40 Years of benthic community change on the Caribbean reefs of Curaçao and Bonaire: the rise of slimy cyanobacterial mats. Coral Reefs, 36(2), 355-367.

Bindoff, N. L., Cheung, W. W., Kairo, J. G., Arstegui, J., Guinder, V. A., Hallberg, R., ... &

ODonoghue, S. (2019). Changing ocean, marine ecosystems, and dependent communities. In IPCC Special Report on the Ocean and Cryosphere in a Changing Climate.

Carpenter, K. E., Abrar, M., Aeby, G., Aronson, R. B., Banks, S., Bruckner, A., ... & Edgar, G. J. (2008). One-third of reef-building corals face elevated extinction risk from climate change and local impacts. Science, 321(5888), 560-563.

Chappell, J. (1980). Coral morphology, diversity, and reef growth. Nature, 286(5770), 249-252.

Comer, P. J., Pressey, R. L., Hunter Jr, M. L., Schloss, C. A., Buttrick, S. C., Heller, N. E., ... & Shaffer, M. L. (2015). Incorporating geodiversity into conservation decisions. Conservation Biology, 29(3), 692-701.

Debrot, A. O., Henkens, R. J. H. G., & Verweij, P. J. F. M. (2018). Staat van de natuur van Caribisch Nederland 2017: Een eerste beoordeling van de staat (van instandhouding), bedreigingen en managementimplicaties van habitats en soorten in Caribisch Nederland (No. C086/17). Wageningen Marine Research.

Debrot, A. O., Houtepen, E., Meesters, H. W. G., van Beek, I. J. M., Timmer, T., Boman, E., ... & Ballantine, D. L. (2014). Habitat diversity and biodiversity of the benthic seascapes of St. Eustatius (No. C078/14). IMARES.

EdgarGJ, Stuart-Smith RD, Willis TJ, Kininmonth S, Baker SC et al. (2014) Global conservation outcomes depend on Marine Protected Areas with five key features. Nature 506:216-220. Foo, S. A., & Asner, G. P. (2019). Scaling up coral reef restoration using remote sensing technology.

Frontiers in Marine Science, 6, 79.

Goldberg, J., & Wilkinson, C. (2004). Global threats to coral reefs: coral bleaching, global climate change, disease, predator plagues, and invasive species. Status of coral reefs of the world, 2004, 67-92.

Gordon, J. E. (2020). Geoconservation principles and protected area management. International Journal of Geoheritage and Parks.

de Graaf, M., Piontek, S., Miller, D. C. M., Brunel, T. P. A., & Nagelkerke, L. A. J. (2015). Status and trends of St. Eustatius coral reef ecosystem and fisheries: 2015 report card (No. C167/15). Imares. Gray, M. (2012). Valuing geodiversity in an ‘ecosystem services’ context. Scottish Geographical

Journal, 128(3-4), 177-194.

Van Haren, H., Duineveld, G., & Mienis, F. (2019). Internal wave observations off Saba Bank. Frontiers in Marine Science, 5, 528.

(26)

Harris, P. T., Bridge, T. C., Beaman, R. J., Webster, J. M., Nichol, S. L., & Brooke, B. P. (2013). Submerged banks in the Great Barrier Reef, Australia, greatly increase available coral reef habitat. ICES Journal of Marine Science, 70(2), 284-293.

Hedley, J. D., Roelfsema, C. M., Chollett, I., Harborne, A. R., Heron, S. F., Weeks, S., ... & Ticzon, V. (2016). Remote sensing of coral reefs for monitoring and management: a review. Remote Sensing, 8(2), 118.

Hengl, T. (2006) Finding the right pixel size. Computers & Geosciences 32(9), 1283-1298.

Hoeksema, B. W., Reimer, J. D., & Vonk, R. (2017). biodiversity of Caribbean coral reefs (with a focus on the Dutch Caribbean).

Hoegh-Guldberg, O. (2011). Coral reef ecosystems and anthropogenic climate change. Regional Environmental Change, 11(1), 215-227.

Jenks, G. F. (1967). The data model concept in statistical mapping. International yearbook of cartography, 7, 186-190.

Kroon, F. J., Schaffelke, B., & Bartley, R. (2014). Informing policy to protect coastal coral reefs: Insight from a global review of reducing agricultural pollution to coastal ecosystems. Marine pollution bulletin, 85(1), 33-41.

van der Land, J. (1977). The Saba Bank – A large atoll in the Northeastern Caribbean. FAO Fish. Rep. 200, 469–481. doi: 10.1371/journal.pone.0009207

Lundvall, S. (2008). Saba Bank Special Marine Area Management Plan 2008.

Meesters, H. W. G., Smith, S. R., & Becking, L. E. (2015). A review of coral reef restoration techniques (No. C028/14). IMARES.

Mulhall, M. (2008). Saving the rainforest of the sea: An analysis of international efforts to conserve coral reefs. Duke Envtl. L. & Pol'y F., 19, 321.

Moffitt, E. A., White, J. W., & Botsford, L. W. (2011). The utility and limitations of size and spacing guidelines for designing Marine protected area (MPA) networks. Biological Conservation, 144(1), 306-318.

Pereira, D. I., Pereira, P., Brilha, J., & Santos, L. (2013). Geodiversity assessment of paraná state (brazil): An innovative approach. Environmental Management, 52(3), 541-552.

Rudolph, T. B., Ruckelshaus, M., Swilling, M., Allison, E. H., Österblom, H., Gelcich, S., & Mbatha, P. (2020). A transition to sustainable ocean governance. Nature Communications, 11(1), 1-14.

Seijmonsbergen, A. C., Guldenaar, J., & Rijsdijk, K. F. (2017). Exploring Hawaiian long-term insular geodiversity dynamics. Landform Analysis, 35.

Snyder, John P., 1987, Map Projections, A Working Manual, U.S. Geological Survey Professional Spalding, M., Burke, L., Wood, S. A., Ashpole, J., Hutchison, J., & Zu Ermgassen, P. (2017). Mapping

the global value and distribution of coral reef tourism. Marine Policy, 82, 104-113. Trembanis, A. C., Forrest, A. L., Keller, B. M., & Patterson, M. R. (2017). Mesophotic coral

ecosystems: a geoacoustically derived proxy for habitat and relative diversity for the leeward shelf of Bonaire, Dutch Caribbean. Frontiers in Marine Science, 4, 51.

(27)

Wei, J., Lee, Z., Garcia, R., Zoffoli, L., Armstrong, R. A., Shang, Z., ... & Chen, R. F. (2018). An assessment of Landsat-8 atmospheric correction schemes and remote sensing reflectance products in coral reefs and coastal turbid waters. Remote Sensing of Environment, 215, 18-32.

Woodhead, A. J., Hicks, C. C., Norström, A. V., Williams, G. J., & Graham, N. A. (2019). Coral reef ecosystem services in the Anthropocene. Functional Ecology, 33(6), 1023-1034.

Wozniak, B., & Dera, J. (2007). Light absorption in sea water (Vol. 33). New York: Springer. Zawada, D. G., Piniak, G. A., & Hearn, C. J. (2010). Topographic complexity and roughness of a tropical benthic seascape. Geophysical research letters, 37(14).

(28)

8 Appendices

Appendix A: Deliverables Sub-indices Maps

(29)

(30)

(31)

(32)

(33)

(34)

(35)

Appendix B : Logistic regression coefficients

Table 6. Output logistic regression model Saba Bank.

COEFFICIENT STD.ERROR Z-VALUE P-VALUE (WALD) INTERCEPT -2.9340 0.9091 -3.227 0.001249 GDI 1.1954 0.3617 3.305 0.00951

Table 7. Output logistic regression model St. Eustatius

COEFFICIENT STD.ERROR Z-VALUE P-VALUE

(WALD)

INTERCEPT 0.7473 0.9055 0.825 0.409

(36)

Appendix C: ArcGIS Model Builder

Figure16. Environment pre-processing, selecting submarine areas in ArcGIS model builder.

Figure18. External grid calculation in ArcGIS model builder.

(37)

Figure20. Roughness diversity calculation in ArcGIS model builder.

Linking coral reef locations to seafloor geodiversity in the Dutch Caribbean