Species

= = toc = 1. Baseline species data used by the DOPA = DOPA uses species from the IUCN Red List of Threatened Species TM, the world’s most comprehensive information source on the global conservation status of animal, fungi and plant species. It contains a rich compendium of supporting information of the distribution range, ecological requirements, habitats and threats to species and on conservation actions that can be taken to reduce or avoid extinctions. For further details, see www.iucnredlist.org.

Global species richness maps for birds, amphibians and mammals from the 2017.2 version of the Red List have been made available in DOPA Explorer’s mapping interface and the species range polygons for all species from the Red List have further been intersected with the boundaries of all protected areas to produce indicative species lists for the protected areas. In summary, the species distribution maps have been rasterized in the DOPA on a 1 km grid and used here in two ways:

1) the **globally assessed taxonomic groups of mammals, birds and amphibians** ( 20,463 species) have been used across all protected areas and are the data behind all species indicators used in DOPA Explorer 2.0, namely the Species Irreplaceability Indicator (SRI) and the Species Coverage Indicator (SCI);

2) **other taxonomic groups** (e.g. Insecta, Reptilia, Bivalvia, …) from the Red List which have been **assessed mainly locally**, an additional set of 15,427 species, have been used for descriptive purposes only.

Overall, the species distribution maps used cover 35,890 species. These maps invariably represent current, known limits of distribution for individual species within their native historical range. Although these maps have many uses, they generally have a coarse resolution and consequent limitations. The species analyses are computed using the distribution range data for species that are categorized with the following attributes: the presence is either extant or probably extant; the origin is either native or introduced and the seasonality is breeding, non-breeding or resident.

Species included in the Red List are classified into the following nine Red List categories based on Red List criteria such as rate of decline, population size, area of geographic distribution, and degree of population and distribution fragmentation:
 * Extinct (EX) – No known individuals remaining.
 * Extinct in the Wild (EW) – Known only to survive in captivity, or as a naturalized population outside its historic range.

Threatened species fall into one of the following three categories:
 * Critically Endangered (CR) – Extremely high risk of extinction in the wild.
 * Endangered (EN) – High risk of extinction in the wild.
 * Vulnerable (VU) – High risk of endangerment in the wild.

All other species fall in these last categories:
 * Near Threatened (NT) – Likely to become endangered in the near future.
 * Least Concern (LC) – Lowest risk. Does not qualify for a more at risk category. Widespread and abundant taxa are included in this category.
 * Data Deficient (DD) – Not enough data to make an assessment of its risk of extinction.
 * Not Evaluated (NE) – Has not yet been evaluated against the criteria.

**Data limitations and assumptions of the SCI** The species distribution data included in this analysis reflect the current state of knowledge of the geo-graphical distributions of the taxon assessed. They do not, of course, represent all amphibian, mammal and bird species in existence, but instead they are used as indicators of the diversity within that taxon.

There are a number of factors which can cause inaccuracy, or inconsistency in the results of our analysis. These can be divided into factors related to species data collection, and factors relating to our analysis techniques.

Data collection limitations = 2. Species Indicators =
 * 1) // Uneven sampling. Since the species EOO data, derived from current literature and expert knowledge, are based on a priori studies, this means that the sampling density is not uniform across the whole conti-nent. Therefore sampling is biased towards easily accessible areas. Relatively inaccessible areas, such as dense lowland rainforest, or conflict areas, will have a lower sampling rate. This results in:(a) Amphibian species EOOs do not include all areas in which a species is found, and(b) Species endemic to inaccessible areas will remain undiscovered until sampling improves. //
 * 2) // Unsuitable areas within EOO. Currently, the GAA have only made available for download vector files for the EOO of each species. The academic literature suggests that modelling of suitable areas for each amphibian species has been done, but the resulting maps are currently not available for download. By using modelling techniques to identify suitable areas for each species, the AMD project showed that within mammal species’ EOOs there are in fact many unsuitable areas for each species. Such an effect may be further exacerbated for amphibian species given that many exist in close proximity to wetlands, which may occur disparately within a large EOO. This highlights a shortfall in EOO data for species diversity mapping. Irreplaceability Index limitations and assumptions //
 * 3) // Species assigned to PAs in which they do not occur. Given that there maybe many unsuitable areas within a species EOO, and the accuracy of an EOO boundary may reduce as the range of the species increases, it is likely that many PAs include commission errors. //
 * 4) // The conservation value of all species is equal. The concept of a flagship species, such as lions or elephants, is not accounted for in this analysis. This means that while the economic value of a protected area may be higher as a result of greater tourism, in terms of biodiversity it has no increased value just be-cause it contains a flagship species. Studies have shown flagship (mammal) species are poor predictors of overall mammal and breeding bird diversity, with 6 flagship mammals representing the same biodiversity as 6 randomly chosen mammal species (Williams et al. 2000) //
 * 5) // Mapping scales vary between taxa and species. For example, an amphibian with an extent of occurrence of 10km is mapped more precisely than a mammal which has a range of thousands of kilometers. The effect this would have on a PA’s RI would be to give more influence to species with smaller ranges. //
 * 6) // Does not account for networks of neighbouring PAs. Networks of connected PAs, such as the WAP park complex of Benin, Niger, and Burkina Faso, are vital to the maintenance of habitat, and essential for the maintenance of corridors for species with larger ranges. //

2.1. Species Richness, Protection and Endemism
The IUCN Red List of Threatened Species providing conservation status, and distribution information on taxa that are facing a high risk of global extinction can be directly used as such to assess the number of threatened species encountered in a protected area or a country. We define hereafter the percentage of **protected species** as the percentage of the species with distribution maps falling at least partly in a protected area and **endemic species** as the species with distribution maps falling exclusively in the country. While the species lists made available for each protected area are derived from the distribution maps of all of the 35,890 species from the Red List we rasterized on a 1km grid, the country statistics are computed only for the three taxonomic groups that have been globally assessed by IUCN. Species lists for each protected area can be downloaded directly in an Excel format. Currently, the IUCN Red List data is largely based on expert opinion and the range maps are broad approximations for many species (especially species of least concern). Mapping scales also vary between taxa and species. Amphibians with an extent of occurrence of a few kilometres are likely to be mapped more precisely than a mammal which has a range of thousands of kilometers.

2.2. The Species Coverage Indicator (SCI)
The Species Coverage Indicator (SCI) proposed by Hartley et al. (2007) was initially called Species Irreplaceability Indicator by the authors and implemented so in the DOPA Explorer Beta (Dubois et al., 2013b). DOPA Explorer 1.0 proposes now two distinct indicators: the Species Coverage Indicator (SCI) and the Species Irreplaceability Indicator (SII) as defined by Le Saout et al. (2013). The SCI is calculated for each protected area by counting how many protected areas a species occurs in (//n//), and adding 1///n// to the SCI of each of those protected areas. The same procedure is carried out for all species in a given taxon. The higher the value of the SCI for a protected area, the higher the number of species found in very few other protected areas and/or the higher the number of endemic species in the protected area. In other words, the higher the SCI, the more important is the role of this PA for conserving biodiversity within the current PA network. Any change to the PA network or the size of the protected areas will impact the SCI. Further normalizing the SCI indicators on a scale of 1-100, one can have an idea of the relative conservation value of the protected area for each taxon by means of the radar plot or by a bar chart showing the ranking of each indicator of the protected area. The SCI suffers from the limitations indicated in Hartley et al. (2007) and Le Saout et al. (2013). Species with smaller ranges are more likely to trigger a higher SCI and species with large ranges will suffer from the fact that connectivity of protected areas is not taken into account and the critical role of corridors in maintaining viable habitats therefore not considered. There is also a concern that the species maps are sometime not accurate enough to be used in conjunction with small protected areas. Hartley et al. (2007) have tried various combinations of species maps and found that the ranking of protected areas based on the SCI is robust to changes in the species maps although this observation still needs to be further assessed with a multi-scale analysis of the SCI values, from country down to protected area level. One should note that the SCI attributes the same weight to all species independently of their taxon or their threat category on the IUCN Red List of Threatened Species. Because threatened species tend to have smaller distributions, and are therefore found in fewer protected areas, they have a greater effect on the indicator score of the protected area. However, this will still give more emphasize to small endemic species in comparison to larger species which might need to be protected by larger areas and more protected areas, such as rhinoceros and lions.

2.3. The Species Irreplaceability Indicator (SII)
The Species Irreplaceability Indicator (SII) in DOPA Explorer 2.0 now corresponds to the one developed by Le Saout et al. (2013). These authors calculated an irreplaceability score for protected areas as an aggregated measure of the degree of dependence of species’ on the protected area. Unlike in the SCI described above, this irreplaceability score for each protected area is independent of the degree of species coverage within other protected areas. Thus, within any given taxonomic group, irreplaceability values can be directly compared across sites worldwide. In contrast to the SCI, the SII is dominated by species for which each protected area has the most responsibility, with little contribution by species that overlap the site by very small percentages. The SII highlights protected areas of particular importance for avoiding the extinction of species (those with relatively high fractions of species ranges within them). The results of the irreplaceability analysis by Le Saout et al. (2013) were used by IUCN to identify potential candidate sites for inclusion in the natural World Heritage network (Bertzky et al., 2013). = 3. Data Processing = The computation of the SII requires a demanding multi-way intersection between over 200,000 protected areas and over 27,000 species ranges. This operation is achieved by a segmentation of the data on a one-decimal degree grid to form a list of either simple squares (where the grid cell is completely contained) or smaller polygons. When applied to all input datasets, this approach allows a straightforward SQL query using the Hadoop ESRI spatial framework, where an ST_INTERSECTION operation is only necessary for those cells which are not completely contained by either polygon. A bespoke Java class written to support this procedure can be found on GitHub, see Lars Francke, 2014 @https://github.com/lfrancke/jrc/blob/master/src/main/java/jrc/CellCalculator.java

Even when the task is thus simplified, there are still occasional errors in the first (union) step of the procedure, and these are much harder to debug once the data is distributed across the Hadoop cluster. Therefore the gridding and union steps were moved off the cluster. This has the added benefit of making helpfully-gridded and indexed data available for use on other platforms.