Résumé
Tropical tuna species are among the world's most commercially important fish. Numerous industrial and artisanal fisheries in developing countries target them using artificial floating structures (Fish Aggregating Devices or FADs). Although tunas' associative behaviour with such structures is well known to fishers, knowledge of how they interact with FADs remains fragmentary. Using echo-sounder buoy data attached to FADs, we assessed tuna aggregation dynamics within an array of anchored FADs (aFADs) in the Maldives—one of the world's leading tuna fishing nations—over two years (2017–2018). Predictions of tuna presence and absence from a random forest model showed good accuracy when evaluated against semi-quantitative logbook data (78 %) but performed poorly with phone survey data (∼49 %). Occupancy at aFADs exceeded 60 % in both years, peaking above 80 % each April. This suggests FAD density is low enough for continuous use by tuna, the population is large enough to sustain occupancy, or both. Aggregations’ continuous residence time (aCRT) averaged 7.96 days, while continuous absence times (aCAT) averaged 3.89 days. Both metrics followed exponential distributions, suggesting the underlying biological process is time-independent and memoryless. Spatial autocorrelation showed partial correlation in tuna absence among aFADs, despite some local effects, supporting previous findings that aFADs in the Maldives do not function as a network. These findings highlight the value of echo-sounder buoy data, offering robust aggregation metrics and insights into tuna behaviour that could inform fisheries management.