Bats and road collisions: building a risk prediction model incorporating landscape variables

Bats use echolocation to operate during darkness, avoid obstacles and forage. Different species have different types of sonars, depending on the habitats they use. Aerial-hawkers use low quasi-constant frequencies which allow them to fly at heights, whereas gleaners use high-pitched modulated frequencies, better suited to fly close to the ground. Knowledge in bat movement has mainly been developed at large scales for the study of migration routes or at middle scales for the study of home ranges. Little is known about bat behaviour at a very small scale, where obstacle avoidance can be observed.

This lack of information is notably problematic when it comes to assessing collision risks with vehicles before road construction. It is commonly assumed that aerial hawkers are less prone to road collisions because they prefer to fly at heights, but some studies have reported otherwise. Many assumptions are also formulated concerning the reaction of each species to landscape features and hence the role of these features in road mortality. Gleaners are described as flying close to the ground and close to the vegetation whereas aerial-hawkers are supposedly more plastic. However, these elements rely mainly on bat experts’ experience and this flaw leads to hazardous impact assessment studies. Indeed, the current method for the determination of potentially high risk crossings is based on the highest activity rate along a road, although it has never been demonstrated that bat activity is proportionally linked to collision risk in all landscape situations.

In this context, a PhD project started in December 2015. The aim is to determine how several types of landscape features (hedges, forest edges, forest driveways, no three-dimensional element) alter collision risks with vehicles. Behavioural observations will be performed with acoustic flight path tracking over a large amount of road locations. Eventually, the goal is to achieve a prediction model for collision risks at pre-existing roads or at road projects before construction, based on local landscape characteristics. Preliminary results such as an interspecific comparison of flight height preferences will be presented and discussed.

bat; behaviour; ecology; road collisions; landscape; prediction model; impact assessment studies