Monitoring spatio-temporal variations in species abundance requires huge amount of data. Scientific labs do not have the resources to collect such data by standard means. Meanwhile, with the internet, large amounts of data can be collected with citizen science programs. For example, the Cornell lab of Ornithology has launched with the National Audubon Society the website ebird.org where any citizen can download his own observations. Such data are very massive (for example the Global Biodiversity Information Facility gathers more than 500 million of counts worldwilde), but they are very heterogeneous and (most of the time) sampled without any protocol. Estimating variations in species abundance from this data is then very challenging. In this talk, we will present a rationale for exploiting these unstandardized opportunistic data for relative abundances monitoring. In particular, we will investigate (in theory and in practice) the gain of using such data. We will also illustrate with one example how these data could enable to investigate new questions in Ecology.

Joint works with R. Julliard, C. Calenge, E. Porcher and C. Coron.