Review
Roost site selection by tree-dwelling bats across biogeographical regions: an updated meta-analysis with meta-regression
Editor: KH
Abstract
- Bats roosting in tree cavities, under loose bark or just on branches in foliage, so-called tree-dwelling bats, are a very diverse and abundant group of species. Although they can be very different species with locally distinct needs, radical exploitation of forest habitats and global changes have meant that many of them are regionally endangered and survive today only in small populations. To help develop appropriate conservation policies and management plans, much research in recent decades has been focussed on inferring habitat selection by tree-dwelling bats. However, large variability in the reported selection patterns makes it difficult to suggest some more-or-less universal and easily applicable management recommendations, also for regions in which nature conservation awareness is low.
- We created a meta-analytic model to explore selection patterns at a global scale. Phylogenetic relationships among species and additional variables were included to explain discrepancies between studies.
- A meta-analytic model showed that bats are selecting trees with trunk diameters that are larger than those of randomly selected trees, and this pattern is consistent within each biogeographical region. However, meta-regression revealed that the detectability of the selection depends strongly on the structural variability within the study environment (size of randomly selected or surrounding trees in forest stands, forest fragmentation, habitat disturbance) and on the methodological approach that has been applied (the length of the study). We found stronger selection for larger trees in non-fragmented and less disturbed forests than in fragmented forests with high habitat disturbance, and, strikingly, short-term studies yielded results with stronger selection than long-term studies.
- Our results suggest that patterns of roosting habitat selection by bats may be overestimated in some studies. In conclusion, we propose that further research should be conducted in all types of forest ecosystem (data from the tropics are currently missing). Future studies should include at least three years of data collection, in order to avoid estimation bias in habitat selection patterns.
