2010

2012

1.Shelter inventory (Atlas)

The observation of individuals inside their shelters allows the identification of some of the species through their morphological characteristics. To this end, and in conjunction with various training actions aimed at Atlas project volunteers, a bat morfological identification key was prepared (Appendix I in Atlas Bats from Continental Portugal 2013). However, the observation of the discriminating characteristics of many of the species often requires handling the individuals. As the handling of bats is subject to a license, issued only to people who prove their ability to handle the species without risk to themselves or to the bat, it was recommended that volunteers without a license should obtain images of the bats observed. Not all the images collected allowed the identification of the species, but still this image collection provided a significant contribution of information to the Atlas.

As expected, the ability to locate shelters is not uniform for all types of structures used by bats. Therefore, and as a consequence, this methodology is not generally applicable to all bat species. As an example, we refer to arboreal species that often use inaccessible cavities inside tree trunks, which makes it impossible to detect them except in cases of tree felling or with the use of radiotelemetry methods. These species were surveyed mainly by acoustic methods, net captures or occasional observations.

Whenever security conditions did not allow visiting the shelter interior, listening points were made with the aid of an ultrasound detector, near the shelter entrance.

The information collected in shelters of recognised importance and included in the Monitoring Programme of Underground Shelters of National Importance for Bats were also integrated in the Atlas project.

2.Acoustic identification (Atlas)

The acoustic data collection during the Atlas was done in five or more sampling points. Each point had a duration of 10 minutes and the five points were distributed in each grid 10x10 km in order to cover the main land uses of the region, but also focusing on habitats with recognized value for bats, such as riparian areas. This sampling was carried out during the first three hours of the night, between April and September, in meteorological conditions favourable to bat activity (e.g. no rain, strong wind or fog).

Digital recordings of bat vocalizations made during the field sampling were subsequently analyzed using a sound analysis software (e.g. Audacity v. 2.0.4 http://audacity.sourceforge.net/ ). In order to assist the Atlas project volunteers and to standardize acoustic identifications, several training sessions were held and a key for the acoustic identification of bats was also prepared (Appendix II in Atlas Bats from Continental Portugal 2013).

The identification of bats from their vocalizations is not always possible. The low quality of many recordings, sometimes resulting from the surrounding noise, may prevent the identification of the vocalization. But more importantly, the vocalizations emitted by many species are too similar, making it impossible to identify them without recourse to more complex numerical analyses that were beyond the scope and objectives of this project. Thus, it was decided not to use acoustic identification data on the following species or groups of species: Myotis spp. (except M. escalerai), Plecotus spp., Eptesicus spp., Rhinolophus euryale and R. mehelyi, Miniopterus schreibersii and Nyctalus noctula.

In the particular case of these last two species, some vocalizations have been, in this project, classified as belonging to the two species with identical vocalizations, respectively Pipis- trellus pygmaeus and Nyctalus lasiopterus. This procedure was chosen because it was considered that, given the vast area of occurrence of P. pygmaeus and N. lasiopterus, these errors would not cause significant deviations in the mapping of their distribution, and would certainly be more informative than the total exclusion of acoustic information for the four species.

3. Mapp Elaboration (Atlas)

Each observation made in the field and inserted in the national database is associated with a geographical location, usually obtained through a GPS (Global Positioning System), or through map viewing programs and geographical information available on the internet (e.g. Google Earth). This database can then be easily exported to a GIS (Geographic Information System), ensuring that the basic cartography of the species distribution is drawn up quickly and correctly.

The maps of the known distribution of each species were prepared automatically, using the Range-tool tool (Urda, D.& Maxim, I., 2012.) for ArcGIS 10 (ESRI). This tool generates maps of distribution and areas of occurrence, derived from a cartographic base of confirmed observations and a reference grid, in the same projection as the cartographic base.

In its simplest form, the product of this tool makes it possible to map the 10x10 km UTM grid squares where the presence of the species has been confirmed. However, for this publication the information was more detailed, and the presence of each species was classified according to the identification methodology, namely morphological, genetic or acoustic.

4.Preparation of the dataset for GBIF publication (PORBIOTA)

Preparation of the dataset for GBIF publicatio used as source the geographic layer in shape file format obtain in https://sig.icnf.pt/portal/home/item.html?id=0d26526ca5b049a3a6e8da783f32fb9c

The following steps were performed:

- The original shapefile of the bat atlas (3763) was imported into QGIS;

- An overlay was made with the "Grid_ED50UTM_transformed_ETRS89TM06_10K" to which all information by location was extracted (as the original shapefile did not contain the 10x10km grid squares);

- The 10x10 grid was exported, in the original projection system (3763);

- In R Studio1 the bat grid shapefile was imported into R and footprintWKT was calculated in the original projection system (3763);

- The centroids of the polygons and the coordinateUncertaintyInMetters field was calculated;

- To calculate the coordinatePrecision field, the grid shapefile was converted to 4326;

- The fields with the original coordinates in 3763 were created (verbatimLongitude, verbatimLatitude, and verbatimSRS);

- The CAOP shapefile of Continental Portugal was imported into R, also in 3763, and an intersection was made to extract the following information: stateProvince, municipality, and locality;

- The shapefile was converted into 4326 to calculate the decimalLongitude and decimalLatitude fields, and the geodeticDatum field was added.

- The footprintWKT, coordinateUncertaintyInMetterand coordinatePrecision information was added to the dataset.

Legend

georeferenceRemarks: refers to the number of the map grid in the atlas

The data collection work for the Atlas project took place during 2011 and 2012 and focused exclusively on bat species that occur in the territory of continental Portugal. Data collected during 2010 were also considered, when available.

The methodology developed was based on the methodology of the work of Rainho et al. 1998, adjusted to a sampling based on a grid of squares of 10 km side. We also integrated some of the adaptations implemented in the ICNF recommendations for monitoring programs (2009) which facilitated the inclusion of data collected by the Environmental Impact Assessment and Monitoring teams and thus allowed extending the coverage of the national sampling.

The location of potential shelters often resulted from information obtained from habitants of the region. The Atlas methodology did not define a minimum number of shelters to be visited per grid. Each site identified as a potential shelter was visited to check for the presence of bats or traces of bats (e.g. accumulation of droppings). When traces were detected, whenever possible, the shelter was visited again at other times of the year.

The maps produced were assessed by a panel of specialists, who listed the observations that needed confirmation. These consisted essentially of observations isolated from the remaining range, or observations in regions where the species was previously unknown. The validation of the listed observations was performed using the sound files and photographs inserted in the database. Observations listed for validation that had no sound files or photographs associated were excluded from the maps. No validation was performed on observations within the known area of occurrence.

All initial data was inserted in a specific relational database built specifically for the project, which implemented several quality control methods. In the preparation of the dataset for GBIF publication, data consistency was verified using cluster algorithms using OpenRefine.