In the following, we will describe the ecological datasets used for mapping purposes and/or for generating the indicators and statistics used to characterize each protected area.

1. Land Cover

Each protected area can be characterized by its land cover. There are a few land cover maps available and we are proposing to use two global products which have a reasonable accuracy and global coherence, namely the Global Land Cover for the year 2000 (GLC2000, see Bartholomé & Belward, 2005) and the GlobCover product for the year 2005[1] produced by the European Space Agency (ESA) in partnership with JRC, EEA, FAO, UNEP, GOFC-GOLD and IGBP.
Note that these two products have not only been generated for different years, but they have different classes and have been prepared on different sampling support. The resolution of GLC 2000 is of around 1 km while GlobCover 2005 has a resolution of 300 m.
The classes of land cover types found in each protected area, the relative surfaces (in %) of these classes and their surface in km2 in the protected area can be displayed and downloaded in Excel format.

2. Soil map

Another important environmental information that has been made available in the interface is a global soil map made available by the FAO/UNESCO. The information can be found at the following address:
http://www.fao.org/soils-portal/soil-survey/soil-maps-and-databases/faounesco-soil-map-of-the-world/en/

3. Crops

The proposed crop map shows the cropland percentage map for the baseline year 2005 (Fritz et al., 2015). This new global cropland map from the IIASA-IFPRI combines multiple satellite data sources, reconciled using crowdsourced accuracy checks, to provide an improved record of total cropland extent as well as field size around the world. The data can be accessed from http://www.iiasa.ac.at/web/home/about/news/150116-Cropland-Maps.html
This information is used to assess agriculture pressure on protected areas as discussed in chapter 9.

4. Fires

Fires are central in the ecology of tropical ecosystems and can act as a threat or a regenerating factor depending on the ecosystem adaptations to it. Many ecosystems in the world are fire-dependent and for them fire is essential to maintain their functionalities and their biodiversity. Hardesty, Myers and Fulks (2005) have estimated that “around 84% of the ecoregions identified by scientists as critical for global conservation have altered fire regimes. This alteration can cause biodiversity loss and habitat degradation”. Besides its relevance for conservation, fire is also a common practice for land management. The information made available here is the active fire density measured over the last month made available by the MODIS team and hosted at NASA EarthData (https://earthdata.nasa.gov/data/near-real-time-data/firms/active-fire-data).
For a more detailed analysis of fires and burnt areas in protected areas in near real time, we refer to the Fire Monitoring Tool (Palumbo et al., 2013) which is available from http://lrm-maps.jrc.ec.europa.eu/firereport/

5. Habitats

Defining funding priorities considering ecological features of a protected area is not an easy exercise as one needs to look at the various supporting, regulating, provision and cultural services provided by the ecosystems found in the protected areas. One also needs to look at the protected area as a part of a larger landscape. While large protected areas are more effective in conserving biodiversity, a small area can be critical as an element of a network of protected areas. Much remains to be done in DOPA as there are only a few indicators that have been adopted globally considering all the above issues. What we have proposed, so far, is to highlight those protected areas that have unique ecological features at the country and ecoregion level as these are more likely to host endemic species and most vulnerable considering the small likelihood of finding such ecological features elsewhere.
DOPA Explorer Beta used eHabitat (Dubois et al., 2013a; Skøien et al., 2013) to compute for each protected area the likelihood of finding anywhere in the ecoregion a set of ecological characteristics similar to the one found in that protected area. This approach is very similar to the one used for ecological niche modelling, where a set of selected thematic ecological maps (e.g. climatic and land cover data, elevation and slopes…) is used to identify areas where a given species has highest probability of being found. To compute for each pixel the similarity to a reference location, one popular approach is based on the Mahalanobis distance (Mahalanobis, 1936). Mathematically simple and fairly easy to understand, the model performs relatively well compared with most other models (Tsoar et al., 2007) and is computationally fast. For each protected area, a similarity map was thus presented highlighting areas that are most similar to those of the protected area. The final result allows end users to assess how ecologically isolated a protected area is as well as the internal ecological variability. One of the main limitations of this approach was that protected areas with heterogeneous landscapes led to an overestimation of the probabilities of finding similar areas elsewhere, because the statistical approach considers an “average habitat” over the whole surface of the analysed protected area. The variables characterizing the “average habitat” may be represented by a range of values that is too broad, leading consequently to a high variance in the final results (Dubois et al., 2013a).
The approach proposed in DOPA Explorer 1.0 is based on a segmentation process that automatically decomposes each protected area into a set of independent areas, representing habitat types, which are then assessed individually in terms of the probability of finding elsewhere similar ecological conditions.By reducing the variability within landscape patches, similarity values can be considered to be more accurate. Implemented in eHabitat+ (Martínez-López et al., in prep.), this segmentation step was applied to all protected areas as large as 100 km2 and the number of segments are reported in the DOPA Explorer 1.0. We have also computed for each segment a map of ecological similarities but this information will be made available in a different interface, the DOPA Analyst, that is currently in development.
The segmentation algorithm used requires two main parameters, a minimum patch size and a similarity threshold, that were here optimized. The minimum patch size was set to be the square root of the total bounding box area of the park, to avoid obtaining segment sizes that were too small to represent any manageable functional habitat types, and the similarity threshold (ranging from 0 to 1) was set to 0.5 to obtain segments representing medium scale landscape patches.

5.1. The Habitat Diversity Index (HDI)

To summarize the information provided by the segmentation step in DOPA Explorer 1.0 we simply indicate for each Protected Area the number of segments, a proxy for the habitat types encountered in the protected area, and this number can further be used to assess the ecological complexity of the park (see e.g. McCoy and Bell, 1991).
A large number of segments alone is not sufficient to highlight the ecological variability of the protected area as it indirectly favours large protected areas. Therefore, we also propose a Habitat Diversity Index (HDI) which is defined here as the number of distinct segments or habitat types divided by the square root of the surface of the protected area (in km2) and then multiplied by 1000.
For marine protected areas, the HDI is defined as simply as the standard deviation of bathymetry. This variable provides information on the vertical relief variability and has been used to identify habitats most likely to support a larger variety of species as topographic complexity is often considered positively associated to biodiversity (see e.g. Thrush et al., 1997, 2001). The value presented as the HDI has been log-transformed in order to generate meaningful distinctions across a wide range of values.
Note that the HDI for terrestrial and marine ecosystems are both systematically shown in the radar plot characterizing each protected area but values for both HDIs will only be found for mixed protected areas, namely those with a marine and terrestrial component.

5.2. Biophysical data used by eHabitat+

Altogether, we are using percentage of tree cover, percentage of grassland cover, bathymetry, slope, aridity, biotemperature, precipitation, aridity, Normalized Difference Vegetation index (NDVI Max. and Min.) and Normalized Difference Water index (NDWI), some of them representing long term annual averages for the terrestrial ecosystems, and the bathymetry for the marine ecosystems.

% woody vegetation and grassland cover

This is derived from the Vegetation Continuous Fields collection which contains proportional estimates for vegetative cover types: woody vegetation, herbaceous vegetation, and bare ground. Only the first two variables are used as the third one is a function of the first two. The product is derived from all seven bands of the MODerate-resolution Imaging Spectroradiometer (MODIS) sensor on board NASA's Terra satellite (DiMiceli et al., 2011). These data have been accessed from http://glcf.umd.edu/data/vcf/

Normalized Difference Vegetation Index (NDVI)

The MODIS Normalized Difference Vegetation Index (NDVI) data set used is available on a 16 day basis for a ten year period between 2001 and 2010. The product is derived from bands 1 and 2 of the MODerate-resolution Imaging Spectroradiometer on board NASA's Terra satellite (Carroll et al., 2004). These data have been accessed from http://glcf.umd.edu/data/ndvi/

Slopes

The relief (slope) on land is derived from the Shuttle Radar Topography Mission (SRTM 30) (USGS, 2004) data accessed from http://www2.jpl.nasa.gov/srtm/

Bathymetry

Marine data have been extracted from the General Bathymetric Chart of the Oceans (GEBCO) maintained by the British Oceanographic Data Centre on behalf of the International Hydrographic Organization (IHO) and the Intergovernmental Oceanographic Commission (IOC) of UNESCO. The dataset is a global 1 minute grid generated by combining quality-controlled ship depth soundings with interpolation between sounding points guided by satellite-derived gravity data. When available, data sets generated by other methods have been included to improve local accuracy.
Digital data accessed from http://www.gebco.net/data_and_products/gridded_bathymetry_data/

Annual bio-temperature, Surface Temperature, annual precipitation, annual evapotranspiration/precipitation ratio

Climate variables (monthly and annual temperature, bio-temperature (the T>0°C) and evapotranspiration/precipitation ratio) have been derived from WordlClim (Hijmans et al., 2005). The variables have been computed according to the definitions of Holdridge (1947). The original data can be accessed from http://www.worldclim.org/
WordClim[ provides gridded maps of current (1950-2000) and future climate variables at different latitude-longitude resolutions, i.e., 10 minutes, 5 minutes, 2.5 minutes and 30 seconds. The dataset for current climate (Hijmans et al., 2005) is the result of a spatial interpolation process using splines applied to measurements from climate stations. The 30 seconds resolution adopted here corresponds to grid cells of 0.86 km2 at equator, usually referred to as a 1 km grid.
The WorldClim data base gives long term monthly averages of precipitation and minimum, mean and maximum temperatures for each pixel. These variables are then averaged for each protected area and for each month.