L'agrégation de thons de sub-surface au sein du système [DCP ancré - macronecton - environnement - pêche] en Martinique : étude hiérarchique par méthodes acoustiques, optiques et halieutiques

Artificial Fish Aggregating Devices (FADs) are known to concentrate tropical tunas. This aggregative behaviour drastically increases the vulnerability of these species to fishing. Alternatively, FADs can be used as oceanic observatories to study the aggregative behaviour of large pelagic fish. We conducted sea surveys in Martinique (Lesser Antilles) to i) establish a typology of the pelagic fish aggregations found around moored FADs ii) study their relations with their biotic and abiotic environment and the local fishery. The mean biomass of large pelagic fish aggregations was also estimated to provide scientific inputs to achieve a sustainable management of the moored FAD fishery. The first part presents the background of the study. We introduce the regional pelagic ecosystem, the aggregative behaviour of pelagic fish around FADs and the Martinican moored FAD fishery. The second part begins with a presentation of the conceptual framework of the study: the hierarchy theory. Observation scales and survey methodology are therefore described. Repeated echosounder surveys run in a star pattern were conducted around 2 moored FADs, in combination with underwater video observations, classical fishing experiments and CTD profiles. In the third part, the elements of the [moored FAD ¿ pelagic fish - environment - fishing] system and their interactions are numerically characterised in the vertical plane. The main type of pelagic fish aggregation was a large aggregation distributed in sub-surface (40-100 m) within a radius of 400 m of the FAD. This aggregation was mostly made of 60 cm FL blackfin tunas (Thunnus atlanticus). It appeared at sunrise and mostly spread away in the late afternoon. It was observed during all daytime surveys. The average density and the vertical structure of the micronektonic sound scattering layers (SSLs) were assessed locally around a FAD for 48h cycles. These SSLs descriptors showed positive correlation with the size and the abundance of the sub-surface tuna aggregation. Whereas this aggregation accounted for about 95% of the total biomass aggregated around a moored FAD, it was nearly unexploited by the local commercial fishermen. The fourth part focuses on the sub-surface tuna aggregation. A high variability of the spatial distribution of the aggregation was evidenced in the horizontal plane at daily, day to day and monthly temporal scales. These variations were related to fluctuations of the global abundance of the aggregation. The daytime average density of sub-surface tuna was modelled as a function of the distance to FAD, within an advection-diffusion framework. Based on these results, a universal kriging model was designed to assess the mean maximum daily biomass of the sub-surface tuna aggregation. This mean biomass estimate yielded 11 tons with an estimation error of 26%. To conclude, we present a hierarchical organization in space and time of a sub-surface tuna population distributed within a network of moored FADs, from micro to meso-scale. We suggest that the spatial distribution of tunas can be explained from fine to coarse-scales by behavioural processes (self-organization, feeding, aggregative behaviour) whereas their spatial distribution is mainly forced by environmental processes at broader scales.