Résumé
Most survey methods developed to estimate abundance of killed animals on motorways may be biased due to the unequal detectability of carcasses, their persistence time on the lanes, and scavengers activities. Unbiased surveys are needed to evaluate the relationships between bird casualties (mortality), motorways characteristics, and the neighbouring avifauna. The present study conducted on four motorways in France, aimed to evaluate factors affecting persistence and encounter probabilities and variations in scavenging activity to obtain unbiased estimates of bird traffic casualties. Each motorway was surveyed once per season during multiple years and we used capture-recapture methods to estimate detection and carcass persistence rates. Results showed that surveys by car were as efficient as surveys by foot in detecting carcasses on the pavement, but less efficient for carcasses on verges. Passeriformes represented the most numerous casualties, and the Barn Owl (Tyto alba) was the most frequently killed species. Encounter probabilities were constant and high (0.957 ± 0.007). Average daily persistence probability was 0.976 ± 0.003. Persistence probabilities were higher for large and old carcasses, during summer, and differed between seasons, but were relatively similar between years. Scavenging activities, estimated using experimental carcasses disposed on the safe lanes of motorways, varied between years, seasons, and differed between diurnal and nocturnal periods. A peak in scavenging activity occurred during diurnal periods in spring. Results suggest that surveys must take into account carcass characteristics and seasonal variability to obtain unbiased estimates of road killed birds on motorways, as well as variation in scavenging rates.