Résumé
Collapse of important fish stocks have revealed that fish stock assessment based on fishery landings at age suffer from a number of limitations. Probably the most serious limitation is that the indication of population collapse is only perceived very late with such data. The present work aims at looking for indices, based on research survey data only, that could capture the spatial pattern of populations and changes across years. A series of candidate statistical indices has been selected or developed for that purpose. In order to handle diffuse population limits, these indices have been designed so that they do not depend on an arbitrary delineation of the domain. The spatial pattern of fish population is described by a variety of indices characterizing location (center of gravity, inertia and anisotropy, spatial patches), spatial extension (positive area, spreading and equivalent area) and microstructure. Collocation between different ages and years is summarized by a global index of collocation. The indices are estimated on the hake data series of bottom trawl surveys performed by IFREMER from 1987 to 2003 in the Bay of Biscay. Behaviour in the spatial indices is analysed, including interaction between the different indices, their relationship with abundance and age, as well as autocorrelation in the time series. The ability of the indices to capture patterns and provide early warning on incoming changes is discussed.