Bat populations are in steep decline and presently, 16% of all species are classified as “threatened”. One main driver identified for this decline is the loss of natural roosting opportunities, caused by the removal of natural habitats. Installation of bat boxes is one solution to compensate for the lack of natural roosting opportunities. Current recommendations for box design emphasize low maintenance costs and are rarely based on empirical evidence.

We investigated occupancy of 13634 bat boxes in northern Bavaria, Germany. In our study boxes differed in type, age and mounting height, as well as in maximum community age, i.e. the length of time a group of boxes had been installed in a particular place, the size of box groups and the distance to the next box in the surrounding area, i.e. box isolation.

Our results showed that box occupancy depended on box type and bat species. As a case study, we analysed the two most common species found within the investigated boxes, Pipistrellus pipistrellus and Myotis nattereri, in more detail. Both species showed preference to a voluminous box that had a narrow entrance (“Gable box” 14 mm). For P. pipistrellus, only box type affected occupancy, whereas for M. nattereri, the relationship between box type and box age were important. Older boxes, and boxes in areas with higher maximum community age of boxes showed higher box occupancy by bats. Box occupancy decreased with the distance between adjacent box groups (“box isolation”). High mounting height showed a tendency for increased box occupancy, but the effect was only weakly significant.

During 2017, 13634 boxes were surveyed by volunteers that acted independently of one another. There was no systematic sampling design, volunteers were asked to check boxes at least once between June and October, on a cold day to avoid bat fly out and locally asynchronous to avoid double counting.

For every box, northing, easting, elevation, calendar week of the check, box type, mounting height and age was recorded. A box was considered occupied when a bat was present or when indirect evidence of bat box use was obtained. Also, other species groups were recorded. Landscape-level variables included the land cover around the box, and the spatial placement of boxes and were calculated with ArcGIS version 10.4.1.

The data 13634 included 26 box types of varying ages and mounting heights. Boxes older than 16 years were dominated by one box type, mostly installed at a height of 1–1.5m and up to 100m to its nearest neighbour. Boxes installed later were predominantly placed at a height of 3–4m, in groups of 4–8 boxes. In addition, boxes within the same area often were of the same box type and age. Due to this unbalanced data structure, we decided to focus on three woodcrete box types: Gable box, Colony box and Flat box, where each had more than 1000 replicates in areas where they occurred together resulting in our dataset "1_box occupancy.txt".

To test for consistency of model results, we ran the identical model setup for four other subsets:

only adult bats: We excluded boxes with only indirect evidence of bat occupancy and boxes with nursery groups.

nursery groups: We excluded boxes with only indirect evidence of bat occupancy and boxes with only adult bats.

Pipistrellus pipistrellus: We only considered areas where P. pipistrellus and M. nattereri occurred together. Boxes with indirect evidence of bat occupancy or boxes with unidentified individuals were excluded. Multiple species boxes that were occupied by species other than P. pipistrellus were removed from the dataset.

Myotis nattereri: We only considered areas where P. pipistrellus and M. nattereri occurred together. Boxes with indirect evidence of bat occupancy or boxes with unidentified individuals were excluded. Multiple species boxes that were occupied by species other than M. nattereri were removed from the dataset.