Une méthode de comptage des oiseaux à partir de photos

Communal roosting occurs in birds of many different taxa and takes place in highly diverse habitats. Some of these concentrations may contain millions of individuals. Accurate estimates of bird numbers at large roost sites are important for different purposes (understanding their biology, establishing population trends and implementing management and conservation actions).
Juan M. Pérez-García, biologist at the Miguel Hernández University of Elch (Spain) has described in the Revista Catalana d’Ornitologia journal the use of a low-cost methodology for conducting censuses of large
concentrations of birds based on digital photos that can be processed with a standardized automatic image analysis program. He applied this technique to estimate the number of birds in a mixed starling roost in south-east Spain, which estimated a total of 125,197 birds (CI 95% 122,829 -130,036). The software showed a mean error of 2.85% ± 3.75%, which was directly related to the number of individuals per photograph. The results indicate that this methodology is not only simpler and less expensive than previously used techniques, but can also provide accurate, comparable quantitative data on the number of individuals in large roosts. It can be applied to different species ( herons, raptors, waders, seabirds..) and situations (breeding colonies, post-breeding, feeding and migratory groups).