Abstract
In order to cope with the human induced rapid environmental changes, many public actions for the protection of nature have been undertaken. Among these are those biodiversity offsetting measures that can enable the restoration of species habitats. These can however create ecological traps. This risk is particularly great for artificial devices which have been widely acclaimed but whose impact on the target populations is often untested. In this thesis, I use the examples of the European roller and the colonial Laridae breeding in the south of France, to show that the test of the ecological trap hypothesis is well adapted to the evaluation of artificial devices, and is generalizable to other contexts. The study of trap creation mechanisms also allows the formulation of good practice recommendations. I propose an adaptive approach for the design, implementation and evaluation of restoration and biodiversity offsetting projects.