Résumé
The aggregative behavior of tuna around floating object is widely exploited by the industrial purse-seine fishery, which deploy thousands of floating objects each year in all oceans in order to improve their catches. These fish aggregating devices (FADs) are generally equipped with echosounder buoys that can collect acoustic data, conferring to these devices the status of privileged observation platforms for the fish communities that aggregate. Using a classification model based on supervised learning algorithms trained on M3I buoy data, we were able to translate the acoustic data collected along the trajectories of 5748 drifting FADs newly deployed between 2016 and 2018 in the Indian Ocean into presence or absence of tuna aggregation. Analysis of the resulting time series indicated that drifting FADs are colonized by tuna aggregation over an average of 39 days. The results also revealed, for the first time, that the residence time of a tuna aggregation around a single DFAD is about 6 days and that DFADs spend on average 9 days without tuna. After colonization, DFADs appear to be occupied by tuna aggregation about 43 % of their soaking time. Finally, we showed that these metrics can manifest spatial and temporal variations.