Geodiversity of La R´eunion

(1)

Geodiversity of La

R´

eunion

Bachelor Thesis

Sophie Baartman

Faculty of Science

Institute for Biodiversity and Ecosystem Dynamics

Supervisors:

Mr. Dr. Kenneth Rijsdijk

Mr. Dr. Thijs de Boer (co-assessor)

(2)

(3)

Abstract

Biodiversity and habitat areas on small islands are declining due to human influences (Wong et al., 2005). Because they have a lower bu↵er capacity, the vulnerability of small islands is higher and natural diversity is threatened (Nurse et al., 2001). In order to protect this natural diversity, the extent of habitat areas must be mapped and the abiotic diversity must be identified. A lot of research has already been done on biodiversity, but not on the influence of abiotic landscape factors (geodiversity) on biodiversity. The aim of this study was to quantify geodiversity of the island of La Runion and analyze the possible linkage between geodiversity and habitat area distribution, so that geodiversity could be used as a predictor for biodiversity. La R´eunion is part of the Mascarene Islands, located in the Indian Ocean, east of Madagascar. Geodiversity of R´eunion was quantified using the Geodiversity Index produced by Chambers (2014), which included geology, geomorphology, pedology and slope angle range and its standard deviation. The geodiversity hotspots were compared to the spatial distribution of natural habitats defined by Strasberg et al. (2005). This spatial analysis was done using ArcMap, ArcCatalog and Excel. The results showed that 55% of the geodiversity hotspots were located within the defined habitat areas. Windward Mountain Rainforest was proven to have the largest area of geodiversity hotspots on the island. Some improvements could be made on the used method. Furthermore, a lot of research is still needed in order to adjust the GDI for the correlation with detailed biodiversity data of the island.

(4)

(5)

1 Introduction

1.1 Relevance

Biodiversity and habitat areas are globally declining because of deforestation, urbanization and the abandonment of degraded land (Wong et al., 2005). This has a large impact on ecosystems all over the world, but for small island systems the impacts are more disastrous than for the mainland. The bu↵er capacity of an island is lower than that of the mainland because of the limited physical size of the insular ecosystem and the limited availability of natural resources Nurse et al. (2001). Bu↵er capacity is the ability to absorb disturbances from outside the ecosystem and to adapt and to future changes in for instance the climate (Nurse et al., 2001). It is because of these factors that the vulnerability of an island ecosystem is relatively high and the natural diversity is threatened. Widespread research on biodiversity and habitat area distribution has been done for several small islands, but geodiversity is a relatively new concept, which needs to be explored. In order to fully understand the e↵ects of the abiotic landscape features on ecosystems, geodiversity and habitat area distribution and their possible linkage should be analyzed. Before this linkage can be examined, first a clear quantification of the geodiversity of the study area should be made.

1.2 Theoretical framework

According to Serrano and Ruiz-Fla˜no (2007) natural diversity has two components: the abiotic and biotic characteristics of a system. Biotic elements include living features of an ecosystem such as species diversity. Abiotic elements include nonliving characteristics of an ecosystem such as geology, geomorphology, relief, water and soils. These last characteristics are often categorized under the term geodiversity. Geodiversity includes the dynamic abiotic features of a landscape. A clear definition of geodiversity is given by Serrano and Ruiz-Fla˜no (2007):

”Geodiversity can be defined from a theoretical point of view as the variability of abiotic nature, including lithological, tectonic, geomorphological, soil, hydrological, topographical elements and physical processes on the land surface and in the seas and oceans, together with systems generated by natural, endogenous and exogenous, and human processes, which cover the diversity of particles, elements and places.”

According to Serrano and Ruiz-Fla˜no (2007), geodiversity should be given a priority when look-ing at land management, nature conservation, sustainability programs and education. Hjort and Luoto (2010) stress that when identifying areas for protection, one must look at geodiversity. The reason for this is that these are areas which could harbour high biodiversity, even if the species composition would change over time. In this way it will be easier to identify areas for long-term preservation of biodiversity as compared to analyzing biodiversity data Hjort and Luoto (2010).

A way to quantify geodiversity is to use a Geodiversity Index. Jonasson et al. (2005) created such an Index with which they compared geodiversity and habitat diversity for three di↵erent scales. Serrano and Ruiz-Flaño (2007) established another Index, which took into account a number of physical elements, the coefficient of roughness and the surface of the unit. Chambers (2014) modified this latter Geodiversity Index to make it applicable for the volcanic island of Tenerife. For this research, this Index produced by Chambers (2014) will be used and modified to be applicable for the island of La Réunion. Then, the Index will be applied in order to find the geodiversity hotspots of La Réunion. These hotspots will be compared to the current extent of di↵erent habitat

(8)

CHAPTER 1. INTRODUCTION

areas on the island, in order to find if there is a correlation between the geodiversity and habitat area distribution.

1.3 Research area

Figure 1.1: The location of La R´eunion and administrative boundaries. The inset shows the posi-tion of the island retrieved from Strasberg et al. (2005)

La Réunion is one of the volcanic islands that form the Mascarene Islands, together with Maur-itius and Rodrigues, located in the Indian Ocean east of Madagascar, see figure1.1. The French island is 2512 km2 and is formed above a fixed hot spot. The island consists of two shield volca-noes: Piton de Neiges and Piton de Fournaise, the latter is a currently active volcano. The island is strongly influenced by its volcanic nature; it has rugged topography and mainly volcanic soils (Schippers, 2015). Réunion has a tropical climate with monthly temperatures ranging between 25 and 30 Celsius, and around 18 Celsius at higher elevations (Schippers, 2015). Although the biodiversity of La Réunion is relatively high, when comparing it with the other Mascarene Islands, it is still threatened by transformation (Strasberg et al., 2005). The introduction of invasive spe-cies, agricultural expansion and urbanization pose the biggest threats for the biodiversity of the island. According to Laurance and Bierregaard (1997) 57000 ha of the primary forest still remains, which correspond with approximately 25% of the estimated original extent. The extent of remark-able habitat areas as defined by Strasberg et al. (2005) can be seen in figure 1.2. These areas were identified based on altitude, the rate of transformation of habitat, heterogeneity in geology and geomorhpology and species diversity and endemism (Strasberg et al., 2005). Furthermore, a large area has been reforested and now almost 40% of the island area is covered with forest (Schippers, 2015). However, at the island of Rodrigue virtually all of the original forest cover has been destroyed, and on Mauritius only 5% still remains. Deforestation of La Réunion still poses a big threat, as recolonization rates of deforested areas are very low (less than 1m2_{/year). The}

forest remnants are only located above 500m elevation, and mostly on the slopes of the active volcano. This is probably because these are the most difficult areas for deforestation (Laurance and Bierregaard, 1997). For protection and management of the habitat areas of the island, it is important to examine whether these remaining habitats are located on areas with a high or a low

(9)

CHAPTER 1. INTRODUCTION geodiversity, in order to assess the availability of undisturbed areas with high geodiversity.

(10)

CHAPTER 1. INTRODUCTION

1.4 Aim

The aim of this research is to gain more information about the correlation between habitat area distribution and geodiversity of the island. By doing so, it will be investigated whether abiotic landscape features could be a predictor for biodiversity and habitat area distribution. Furthermore, additional knowledge on the use of a Geodiversity Index for the quantification of geodiversity of small islands will be gained. Hopefully this will lead to a broader understanding of spatial variation of abiotic elements of the landscape at the island scale. In the end, the aim is to use geodiversity as a starting point when identifying areas for long-term habitat and biodiversity protection.

1.5 Research Questions

1. How can the Geodiversity Index of Chambers (2014) be used for the quantification of geo-diversity on the island of La R´eunion in order to correlate it with the current habitat area distribution?

a) What adjustments should be made on the Geodiversity Index of Chambers (2014) for the quantification of geodiversity on La R´eunion?

b) What is the most important factor of the Geodiversity Index when making the correl-ation with habitat area distribution on La R´eunion?

c) What grid size should be used for the Geodiversity Index to be relevant for forest biodiversity?

2. Is there a correlation between geodiversity hotspots and the distribution of the current habitat areas on the island of La R´eunion?

a) What percentage of the geodiversity hotspots lies within the current habitat areas? b) Which habitat type area corresponds with the highest geodiversity on La R´eunion?

(11)

2 Methods

The methodology includes a description of data collection, an explanation of the computational analysis and a description of the statistical comparison of habitat area distribution and geodi-versity.

2.1 Computational analysis

The first step of this research was the collection of the appropriate GIS data. A Digital Elevation Model (DEM) of 30 m resolution was found via the Global Landcover Facility (2008). Strassberg provided a current habitat distribution dataset of the island, which contained the location and area of 19 habitat types. The 1:100,000 Geological, 1:100,000 Geomorphological and 1:100,000 Pedological maps were found in JPG and PDF format, thus needed to be digitalized in ArcGIS in order for them to be used in the analysis. Appendix B contains the original maps which were digitalized and appendix A shows the digitalized maps. The JPG files are georeferenced in ArcMap to match the exact location of the DEM, using the coordinate system WGS 1984 UTM Zone 40S. After that, new feature classes are created for each map and polygons were drawn over the di↵erent geological, geomorphological and pedological units. The produced shapefiles were converted into raster files using the Feature to Raster tool, at a 500m resolution. The original maps legends were also converted into the ArcGIS datasets. Furthermore, the legends of the Geological and Pedological maps needed to be translated from French into English.

The grid cell size had to be determined first, before creating an island grid. This was done using the procedure outlined by Hengl (2006) by using resolutions that comply with the inherent prop-erties of the input datasets. Furthermore, in the article, an equation is proposed for calculating the right pixel size, which takes into account the resolution of the maps used in the analysis and a scale number of 0.0025 (equation2.1) (Hengl, 2006). When 1:100 000 is used as scale number, the outcome pixel size is 250m. However, when the maps were digitalized and converted to a raster, the resolution was changed to 500m. Thus, a 500m by 500m grid was used for the calculation of the Geodiversity Index.

p = SN_{⇤ 0.0025} (2.1)

Where p = pixel size, SN = scale number.

Equation 2.2 shows the GDI that Chambers (2014) used for the quantification of geodiversity on the island of Tenerife. This equation was slightly adjusted for the quantification of the geodi-versity of La R´eunion (equation2.3). According to Chambers (2014), a geodiversity assessment of a volcanic island should include, geology, pedology, geomorhpology and slope angle range and standard deviation. The range indicates the di↵erence between minimum and minimum slope angles. The standard deviation is the variance of the slope angles. As can be seen, Chambers (2014) used geology mega-units and sub-units, but these were not included in the data of La R´eunion. Thus, only one geological unit is put into the equation. Furthermore, the variety in geomorphology is added to the equation. Geology, geomorhpology and pedology are all weighed equally, and the range of the slope angle category is weighed twice. The slope angle category is assumed to be more important because a high value would imply a high diversity of slope angle, which is important for geodiversity Chambers (2014). The variety is calculated per grid cell; that means that if for instance there are two geological units present in one cell, it will get a value of two.

(12)

CHAPTER 2. METHODS

GDI = Gl + Gm + P + 2(Sr) + Ss (2.2) Where: GDI = Geodiversity Index, Gl = Variety of geology, Gm = Variety of geomorphology, P = Variety of pedology, Sr = Range of slope angle category, Ss = Standard deviation of slope angle category.

GDI1 = 2(Gm) + Gs + P + 2(Sr) + Ss (2.3) Where GDI1 = Geodiversity Index 1, Gm = Variety of geology mega-units, Gs = Variety of geology sub-units, P = Variety pedology, Sr = Range of slope angle category, Ss = Standard deviation of slope angle category.

To obtain the right parameters for the Geodiversity Index (equation 2.1) the steps that were used and explained by Chambers (2014) were used. First, a slope map was created using the DEM to Slope tool, with the 30m DEM as input file. After that, the range and standard deviation of the slope were calculated using the Zonal Statistics Tool, selecting ALL as output. These values were reclassified using the Field Calculator and python scripts provided by Chambers (2014), so that the values were put into classes shown in table 2.1 and 2.2. For every unit, five di↵erent classes of variety are used. These scripts can be found in Appendix C. The variability of the geology, geomorphology and pedology units per grid cell was needed for calculating the GDI. This variability could be calculated using the Zonal Statistics tool of ArcMap, with the raster file of the designated map and the grid as input files, selecting VARIETY as output.

Table 2.1: Categories of the slope angle standard deviation

SS Category 1 2 3 4 5

Value 0-7 7.01-8.77 8.87-10.62 10.63-16.09 >16.1

Table 2.2: Categories of the slope angle range

Sr Category 1 2 3 4

Value >45.57 45.48-55.85 55.86-66.81 >66.82

In order to make the GDI calculation, the output tables of the Zonal Statistics were joined to the attribute table of the grid. A new field was added to the attribute table of the grid and named GDI value. The Field Calculator was used to calculate the Geodiversity Index value per grid cell using equation2.2. The GDI value map was converted into a raster file using the Feature to Raster tool. Then, the values were classified into five classes using the Natural Jenks into the classes Very Low, Low, Medium, High and Very High, shown in table2.3.

Table 2.3: Categories of the GDI based on equation 2.1 Geodiversity Very Low Low Medium High Very High

Value 3-7 8-9 10-11 12-13 14-19

2.2 Statistical analysis

First, the GDI was analyzed in order to find out the importance of the slope factor in determining the output values. This was done using regression analysis with the R-square method between the

(13)

CHAPTER 2. METHODS slope map and the GDI map. Secondly, statistical analysis of the data was needed, in order to correlate between habitat area distribution and the spatial distribution of geodiversity hotspots on the island of La R´eunion. The habitat area distribution map provided by Strasberg et al. (2005) is shown in figure1.2. This map was made based on existing data and expert knowledge. First, the percentage of geodiversity hotspots that lay within habitat areas was calculated. Geodiversity hotspots are defined as being grid cells with a geodiversity of High or Very High. The Tabulate Intersection tool of ArcMap was used to calculate the percentage of these geodiverse areas that are ’COMPLETELY WITHIN’ the habitat areas. This overlapping area was then divided by the total area of High to Very High geodiversity of the island. Secondly, the mean geodiversity value per habitat area was calculated using the Zonal Statistics tool, with the GDI map and the Habitat map as input values, selecting ’MEAN’ as output value. Lastly, the largest area of geodiversity hotspots per habitat type was calculated. This was done using the Tabulate Intersection tool, again calculating the overlapping areas of High to Very High geodiversity and habitat areas. Afterwards, the sum of the area of geodiversity hotspots per habitat area was calculated.

(14)

(15)

3 Results

In this section, the results of this research will be presented. First, the outputs of the quantification of the geodiversity of La R´eunion will be shown. Then, the results of the statistical comparison of habitat area distribution and the geodiversity of the island will be presented.

3.1 Geodiversity Index

(16)

CHAPTER 3. RESULTS

Figure 3.2: GDI frequency distribution

Figure3.1shows the geodiversity map of R´eunion. It shows a range in geodiversity from Very Low to Very High. The coastal zones and the lowland agricultural areas mostly have values of Very Low or Low geodiversity. The inland and the slopes of the volcanoes show High to Very High geodiversity. The craters show a lower geodiversity than the surrounding areas. Figure3.2

displays the frequency distribution of the GDI values. One can see that the distribution is slightly right or positive skewed, with a minimum of 3 and a maximum of 19. The overal mean GDI is 10.34 and the standard deviation is 2.79 with a total count of 10260 grid cell values. When looking at the correlation of the slope and the GDI, the R-square method gave a value of 0.772, which is significant.

3.2 Comparison with habitats

Figure 3.3shows the Habitat map by Strasberg et al. (2005) plotted over the created GDI map. Several overlapping areas can easily be observed. Spatial analysis showed that 55% of the High to Very High geodiversity areas fall within these habitat areas, thus 45% does not. The habitat area with the highest mean GDI value is the Lowland Savanna, which is a small area located in the northern coastal zone of the island. The habitat area with the largest area of High to Very High geodiversity is the Windward Mountain Rainforest.

(17)

(18)

(19)

4 Discussion

This section includes the interpretation of the results, a discussion of the methods and some recommendations for future research.

4.1 Interpretation of the results

The GDI map, the DEM and the Slope map, show that the geodiversity hotspots are mainly located in the inland and on the steeper slopes of the volcanoes. The lower lying areas, the coastal areas and lava flows generally have a lower GDI value. This is as expected as the slope is an important factor in the GDI, with a R-square value of 0.772. Additionally, the geology, geomorphology and pedology maps likely relate to the topography of the island, which could be the reason for the strong correlation. Topographic features, such as slopes are commonly used to define boundaries between map units. Strasberg et al. (2005) used topographic features as an indicator for habitat boundaries.

There is one unexpected result regarding the overlap between geodiversity hotspots and habitat areas. The volcano and lava flow of the Piton de la Fournaise has a relatively low geodiversity, but it is defined as a remarkable habitat area by Strasberg et al. (2005). They defined it is as being a ’Lava Flow’ habitat area. The low geodiversity can be explained by looking at the separate maps of the GDI parameters. The geology map shows only one unit of ’alluvium’ in this area, the geomorphology and pedology have two di↵erent units for this area and the slope is relatively homogeneous and low. This homogeneity is understandable as it is one defined lava flow. However the definition of a remarkable habitat area is also understandable as lava flows are very often the basis for unique habitat types and harbor the most diagnostic species (Vel´azquez and Bocco, 2001).

This research shows that most of the geodiversity hotspots (55%) on R´eunion are located within the defined habitat areas, which emphasises the predictive power of the GDI regarding habitat area distrubtion. This agrees with the results of previous studies. Hjort and Luoto (2010) did research on the inclusion of explicit measures of geodiversity in order to predict biodiversity in a boreal landscape. They included geological, geomorphological and hydrological diversity in a plant species richness prediction model Including climate and topography variables. These geodiversity factors improved the explanatory power, predictive ability and robustness of the model. Therefore, geology, gemorphology and hydrology seem to be promising to predict biodiversity in this area (Hjort and Luoto, 2010). Although the above mentioned study looked at the predictive ability of geodiversity of plant species richness, and not habitat distribution, the results can be compared to this study. Assuming that there is a positive correlation between plant species richness and habitat area distribution, geodiversity could be a predictor for habitat area distribution. Another study done by Burnett et al. (1998) also shows a positive correlation between abiotic and biotic features of the landscape. They proved that a larger heterogeneity of abiotic conditions provide a greater diversity of potential niches for plant and animals than homogeneous landscapes. For this research they used an index, which included terrain and soil variation and tested it in a deciduous forest in the northeastern United States.

As already mentioned in the introduction of this paper, if there is a strong correlation between geodiversity and biodiversity, the identification of protected areas could be focusing on geodiversity hotspots in the future. These protected areas would have a high probability of harbouring high biodiversity, even is the species composition would change over time (Hjort and Luoto, 2010). The results of this study are a good step towards understanding the influence of abiotic factors on biotic features, such as biodiversity, and habitat area distribution.

(20)

CHAPTER 4. DISCUSSION

4.2 Methodological discussion

Some possible errors of the used data should be mentioned. First, the original maps of the geology, geomorphology and pedology were manually digitized because of the shortage of appropriate GIS data. This could have lead to possible errors and bias in the maps. Secondly, the 30m DEM of R´eunion that was used during this research originally contained some sinkholes, which have been filled with the Fill tool before it was uploaded to the Global Landcover Facility. This lead to some triangle structures in the areas which had been filled. This could have lead to possible flaws in the slope map in these areas and therefore have an influence on the GDI.

The use of equation 2.1, based on Chambers (2014) could be discussed as well. One could say that the equation is too simple to quantify geodiversity. However, as there was no biodiversity data available for La R´eunion, the equation could not be adjusted to its purpose. The equation could be made more complex if detailed biodiversity is available in the future. Then it could be adjusted to be fit for the prediction of the biodiversity distribution of the island.

More importantly, because there was no detailed biodiversity data available of R´eunion, the cor-relation between geodiversity and biodiversity cannot be made. However, according to Butchart et al. (2010) the extent of habitat is one of the most important indicators for assessing biodiversity. They evaluated the recent biodiversity decline according to several parameters, and habitat frag-mentation and decline in habitat area were two of these parameters. Furthermore, the habitat map used for this research, made by Strasberg et al. (2005) was aimed to derive biodiversity pat-terns and processes. The map’s purpose was to serve as a sound basis for defining priorities for the conservation of biodiversity hotspots, which were stressed to be in the defined habitat areas. Assuming that there is a positive correlation between biodiversity hotspots and the defined habitat areas by Strasberg et al. (2005), it could be hypothesised that there is also a positive correlation between biodiversity and geodiversity hotspots. However, more research is needed in order to test this hypothesis.

4.3 Future research

As mentioned before, the hydrology is an important factor for predicting biodiversity (Hjort and Luoto, 2010), but it was not included in this research. Therefore, a recommendation for further research is to include hydrological variety in the GDI of R´eunion, as this would probably improve its predictive ability for biodiversity. Secondly, if in the future, detailed biodiversity data is available, such as plant species richness data, the predictive ability of the GDI can be tested. This could be done by an extensive correlation analysis, such as Chambers (2014) did in her paper on Tenerife. Another feature that could be improved in the GDI is the geomorphology factor, as new research on the optimization of geomorphological mapping is being done. Anders et al. (2011) state that semi-automated geomorphological mapping techniques are replacing classical techniques because of the growing availability of high-resolution data. They propose the use of DEM derived data such as slope angle and topographic openness, and if available Light Detection and Ranging (LiDAR) data for identifying geomorphological features. The use of this approach in combination with the GDI would probably improve the explanatory power of the GDI.

An analysis, which would also be very interesting is the comparison of past and current habitat area extent of La R´eunion, which is shown in figure 1.2 and the correlation with geodiversity. In this way, the influence of the GDI could be investigated on the habitat area decrease. It is interesting to know whether how many geodiversity hotspots, which were part of remarkable habitat areas have been lost in the past due to human influences.

(21)

5 Conclusion

In this study, the correlation between habitat area distribution and geodiversity of the island of La Réunion was tested. The aim was to investigate whether geodiversity could be used as a predictor for habitat area distribution.This was done to provide information for the identification of protected areas. The geodiversity hotspots were hypothesised to be sites for potential high biodiversity. The geodiversity assessment included: geology, geomorphology, pedology and slope angle range and standard deviation. Slope was proved to be an important factor in this equation after a correlation analysis with an R-square value of 0.772. After the quantificaton, a geodiversity map was made in which the geodiversity hotspots could be identified and compared to the habitat area map created by Strasberg et al. (2005). The results show that 55% of the geodiversity hotspots are located within habitat areas. Based on these results of this study and previous studies, a positive correlation between geodiversity hotspots and biodiversity hotspots on the island of La Réunion could be hypothesised. However more research is needed on this topic to prove this correlation, as 55% is not a very strong correlation. As geodiversity is a relatively new topic, the opportunities for further research abound. The GDI could be improved to better correlate with biodiversity data, such as plant species richness data. Furthermore, the habitat area loss on Réunion and the amount of lost geodiversity hotspots could be investigated.

(22)

(23)

Bibliography

Anders, N. S., Seijmonsbergen, A. C., and Bouten, W. (2011). Segmentation optimization and stratified object-based analysis for semi-automated geomorphological mapping. Remote Sensing of Environment, 115:2976–2985.

Burnett, M. R., August, P. V., Brown, J. H., and Killingbeck, K. T. (1998). The influence of geomorphological heterogeneity on biodiversity i. a patch-scale perspective. Conservation Biology, 12:363–370.

Butchart, S. H., Walpole, M., Collen, B., Van Strien, A., Scharlemann, J. P., Almond, R. E., Baillie, J. E., Bomhard, B., Brown, C., Bruno, J., et al. (2010). Global biodiversity: indicators of recent declines. Science, 328:1164–1168.

Chambers, R. (2014). Development of a geodiversity index on a volcanic island: Tenerife, Canary Islands. University of Amsterdam. Unpublished MSc. Report.

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

Hjort, J. and Luoto, M. (2010). Geodiversity of high-latitude landscapes in northern finland. Geomorphology, 115:109–116.

Jonasson, C., Gordon, J. E., Koci´anov´a, M., Josefsson, M., Dvorak, I. J., and Thompson, D. B. (2005). Links between geodiversity and biodiversity in european mountains: case studies from sweden, scotland and the czech republic. The Mountains of Europe: Conservation, Management and Initiatives, pages 57–70.

Laurance, W. F. and Bierregaard, R. O. (1997). Tropical forest remnants: ecology, management, and conservation of fragmented communities. University of Chicago Press.

Nurse, L. A., Sem, G., Hay, J., Suarez, A., Wong, P. P., Briguglio, L., and Ragoonaden, S. (2001). Small island states. Climate change, pages 843–875.

Schippers, J. (2015). Islands of R´eunion and Mauritius, east of Madagascar.

Serrano, E. and Ruiz-Fla˜no, P. (2007). Geodiversity. a theoretical and applied concept. Geographica Helvetica, 62:140.

Strasberg, D., Rouget, M., Richardson, D. M., Baret, S., Dupont, J., and Cowling, R. M. (2005). An assessment of habitat diversity and transformation on la r´eunion island (mascarene islands, indian ocean) as a basis for identifying broad-scale conservation priorities. Biodiversity & Con-servation, 14:3015–3032.

Vel´azquez, A. and Bocco, G. (2001). Land unit approach for biodiversity mapping. Landscape Ecology Applied in Land Evaluation, Development and Conservation, ITC pub, 81:273–285. Wong, P. P., Marone, E., Lana, P., Fortes, M., Moro, D., Agard, J., Vicente, L., Thonell, J., Deda,

(24)

Geodiversity of La R´eunion