Methodology

The data for all species, time spans and locations are stored in a standardized database and consist of time series of count data from various hibernacula. Ideally, the number of individuals per bat species in a hibernaculum is determined once a year in winter, but it can happen that a hibernaculum has not been visited in one or more years. In this case, no count value is included in the time series for the missing year. However, if a roost was counted in one year but the respective species was not seen, this year is included in the time series for this species as a 0 count. What is actually counted during a count depends on many factors. For example, surface structures (smooth concrete walls vs. brick walls with many cracks), inaccessible areas, different human experiences, weather conditions on the counting day, counting protocols and the behavior of the bats influence the counting results in very different ways. The statements we can make with this data are therefore only of limited reliability at roost level. However, a sufficiently large amount of data and good geographical coverage of the hibernacula can reveal rough trends in population size and make changes over the years visible.

If the user selects a species, a time period and an area, the corresponding data is automatically filtered out of the database. Based on the selected data, a map of the hiberncula (right), the trend calculation (left), a description of the data basis and the trend (top) and a list of data providers (bottom) are generated. Population trends are calculated in the background from data for all years of a selected region in order to utilize the greatest possible information. Selecting a time period displays the desired section of the trend calculated from all data.

For the selected data, the number of visits where bats were counted, the number of sources and the number of roosts are indicated here. A single source can summarize different data providers, e.g. if the state offices have provided the data of all bat counters in their area (see "Sources").If, for example, 5 counts were made at one location between 2000 and 2010 and 7 at a second location and the data was provided by a single data provider, the following is shown here: based on 12 visits from 1 source in 2 hibernacula.

The distribution area of the species in Germany is shown in red on the map (German Federal Agency for Nature Conservation (BfN)).If a quadrant (20 x 20 km) is colored dark, this means that at least one roost was counted in this quadrant, the data of which are included in the trend calculation on the left.

This shows the long-term trend over all years (black line) with confidence interval (light grey area) and short-term fluctuations (dark black dots) with confidence interval (vertical lines). Significant increases or decreases in the trend are marked in green vs. red. The image can be downloaded using the download button in the top right-hand corner of the image. The population trends are calculated using Hierarchical Generalized Additive Models, for the exact methodology see Pedersen 2019, Knape 2016 and Fewster 2000. Here is a brief explanation of the procedure: The data are assumed to be negatively binomially distributed and linked to the explanatory variables time and location via a log-link function.The model consists of three components:

• a fixed component that estimates a common long-term trend across all data over the years,
• a year-specific component that estimates fluctuations between years (Knape 2016),
• and a local component that estimates a local trend for each counting location that is as close as possible to this common trend (GS model, Pedersen 2019).

The trends are estimated using a smoothing function and can therefore also depict non-linear relationships. Significant increases and decreases are marked in color in the trend curves if the slope of the curve is significantly greater or less than 0. Colored bars above the x-axis indicate a significant change in the slope.

• Fewster, R.M., Buckland, S.T., Siriwardena, G.M., Baillie, S.R. & Wilson, J.D. (2000) Analysis of population trends for farmland birds using generalized additive models.Ecology, 81, 1970-1984.
• Knape J. Decomposing trends in Swedish bird populations using generalized additive mixed models. Journal of Applied Ecology. 2016;53(6):1852–1861.
• Pedersen EJ, Miller DL, Simpson GL, Ross N. Hierarchical generalized additive models in ecology: an introduction with mgcv. PeerJ. 2019;7:e6876.

Here, years with increasing, decreasing and stable population numbers are summarized from the estimated trend in the middle left graph. They are intended to provide a quick overview of how the population of the selected species has changed over the period shown.

Individuals or institutions that provided the data from which the trend in the center left was calculated. Further details on the sources can be displayed using the drop-down menu. When using the information provided, we recommend citing the persons named here as data providers together with the website as the source, e.g. "www.batlas.info retrieved 17.4.2024, data source: Expert1, Expert2, BfN, LfU, Batman, Robin.

English and Latin name of the species for which the trend is calculated. Click on "Show information and images" to view more detailed information on the respective species and additional images.