Simulation de la dynamique de pêcherie : modélisation, complexité et incertitude

Fisheries are defined by a set of fishermen and of marine populations impacting each other in a spatial-limited marine ecosystem. Within the over-exploitation context and according to ratified sustainable management objectives, my research has been focused on the development and a caution use of a mechanistic model describing fisheries dynamics. This model allows for integrating existing knowledge of fisheries. It aims at improving the understanding of fisheries functioning and at investigating the impact of fishing regulations on fisheries dynamics. This modelling exercise has supported many methodological developments for its parametrisation and its use in uncertainty context. Concerned with promoting a correct use of a complex model (that ambiguity link to numerous uncertainties could lead to an instrumental use of models moving beyond the scientific framework), I have been interested in a statistical approach of sensitivity analysis. The first part of my manuscript describes the fisheries dynamics model ISIS-Fish (Integration of Spatial Information for Simulation of Fisheries) and its use for several case-studies. The second part focus on statistical methods to estimate model's parameters. Finally, in the third part I present some developments on sensitivity analysis to take into account for inputs uncertainty on simulation outputs. I conclude by giving some perspectives of my work.