L’objectif général de la thèse est de modéliser les principaux processus biologiques et économiques régissant des pêcheries multi-espèces et multi-flottilles afin de proposer des stratégies viables pour la gestion durable de ces pêcheries mixtes, dans un contexte stochastique et multiobjectif. Plus spécifiquement, cette thèse utilise des analyses de co-viabilité stochastique pour étudier les arbitrages entre des objectifs contradictoires de gestion (conservation, et viabilité économique et sociale) des pêcheries mixtes. Deux pêcheries mixtes sont analysées dans cette thèse : la pêcherie française mixte démersale du golfe de Gascogne et la pêcherie crevettière australienne du Nord (NPF). Ces deux pêcheries sont multi-espèces, et utilisent des stratégies multiples de pêche, induisant des impacts directs et indirects sur les écosystèmes. Cette thèse propose une application de la co-viabilité stochastique à ces deux cas, en prenant en compte leur histoire, leur contexte socio-politique et les différences dans les stratégies et objectifs de gestion. Les résultats suggèrent que le status quo peut être considéré comme une stratégie biologiquement durable mais socio économiquement à risque dans les deux pêcheries (ainsi qu’à risque écologique dans le cas de la pêcherie australienne). Les simulations réalisées pour le golfe de Gascogne permettent de comparer les arbitrages associés à différentes réductions de capacités par flottille et de montrer qu’il existe des solutions de gestion permettant la co-viabilité du système (viabilité biologique des différentes espèces considérées et viabilité socio-économique des flottilles) contrairement à des stratégies de gestion mono-spécifiques ou basées sur la maximisation de la rente. Dans la pêcherie crevettière australienne, l’analyse montre que les stratégies de diversification permettent de limiter le risque économique contrairement aux stratégies plus spécialisées. / Empirical evidence and the theoretical literature both point to stock sustainability and the protection of marine biodiversity as important fisheries management issues. Decision-support tools are increasingly required to operationalize the ecosystem-based approach to fisheries management. These tools need to integrate (i) ecological and socio-economic drivers of changes in fisheries and ecosystems; (ii) complex dynamics; (iii) deal with various sources of uncertainty; and (iv) incorporate multiple, rather than single objectives. The stochastic co-viability approach addresses the trade-offs associated with balancing ecological, economic and social objectives throughout time, and takes into account the complexity and uncertainty of the dynamic interactions which characterize exploited ecosystems and biodiversity. This thesis proposes an application of this co-viability approach to the sustainable management of mixed fisheries, using two contrasting case studies: the French Bay of Biscay (BoB) demersal mixed fishery and the Australian Northern Prawn Fishery (NPF). Both fisheries entail direct and indirect impacts on mixed species communities while also generating large economic returns. Their sustainability is therefore a major societal concern. A dynamic bio-economic modelling approach is used to capture the key biological and economic processes governing these fisheries, combining age- (BoB) or size- (NPF) structured models of multiple species with recruitment uncertainty, and multiple fleets (BoB) or fishing strategies (NPF). Economic uncertainties relating to input and output prices are also considered. The bioeconomic models are used to investigate how the fisheries can operate within a set of constraints relating to the preservation of Spawning Stock Biomasses (BoB) or Spawning Stock Size Indices (NPF) of a set of key target species, maintenance of the economic profitability of various fleets (BoB) or the fishery as a whole (NPF), and limitation of fishing impacts on the broader biodiversity (NPF), under a range of alternative scenarios and management strategies. Results suggest that under a status quo strategy both fisheries can be considered as biologically sustainable, while socio-economically (and ecologically in the NPF case) at risk. Despite very different management contexts and objectives, viable management strategies suggest a reduction in the number of vessels in both cases. The BoB simulations allow comparison of the trade-offs associated with different allocations of this decrease across fleets. Notably, co-viability management strategies entail a more equitable allocation of effort reductions compared to strategies aiming at maximizing economic yield. In the NPF, species catch diversification strategies are shown to perform well in controlling the levels of economic risk, by contrast with more specialized fishing strategies. Furthermore analyses emphasize the importance to the fishing industry of balancing global economic performance with inter-annual economic variability. Promising future developments based on this research involve the incorporation of a broader set of objectives including social dimensions, as well as the integration of ecological interactions, to better address the needs of ecosystem-based approaches to the sustainable harvesting of marine biodiversity.