Analysing data from bat observations along transects


Thierry Onkelinx


July 2, 2017

This post will handle observations of bats along a set of transects. It is a work in progress simply because I’m still collecting more data. So come back once and awhile to see potential updates.

The survey

The principle of the survey is quite simple: walk around with a bat detector and note the route you took and were you encounter bats. Repeat this several times. I choose to take a different route each time so I can cover the same area at different times of night.

I use ObsMapp to trace the route and the observations. The bat detector is a Peersonic RPA2 with Philips SHB9850NC headphones. This set-up is quite handy. At the start of the route you start listening to the bat detector and tell ObsMapp to start tracking the route. Each time you encounter a bat you a) make a sound recording of the bat and b) mark the observation in ObsMapp1. At home you upload the observations to, or Then you check the observations based on the recorded sounds and update the observations on the website. Each survey is downloadable from the website under several formats. Here we will use both the KML format and the csv format. The KML is required because it contains both the track of each route and the observations. The csv is needed for the date, the start time and end time of the track.

Survey effort has unfortunately no API to download our data. We had to download every track manually as a KML file. Then we parse those files to get the observations and the tracks. The code to parse the files is available in the source code of the blog. The result is the set of all tracks and observations (Figure 1).

Figure 1: Map with all tracks and all observations.

We determine the survey effort as the cumulative proportion on the area of grid cells that a covered by a survey. So we start by defining a grid with 100x100m resolution. Then we create of buffer on 20m around the tracks because assume that we can here most bats up to this distance on the bat detector. The result is a ribbon marking the area where we could have detected bats. The total area of ribbon per grid cell is an indicator of the total survey effort per grid cell (Figure 2).

Figure 2: Map with total survey effort.

Presence of bats

The number of observations per species is quite different. Only the common pipistrelle (Pipistrellus pipistrellus) has currently enough observations for the analysis (Table 1). Note that the table also contains other mammal species because I note all mammal species that I encounter.

Table 1: Number of observations per species and per track
species 2017-06-04 2017-06-10 2017-06-19 2017-06-29
Pipistrellus pipistrellus 51 39 14 28
Erinaceus europaeus 0 2 0 0
Martes foina 0 0 1 1
Myotis mystacinus 2 0 0 0
Oryctolagus cuniculus 1 0 0 0
Pipistrellus spec 0 0 1 0
Vulpes vulpes 1 0 0 0

Next we check for each combination of track and grid cell whether we detected pipistrelles or not. Figure 3 displays the average presence over all tracks.