Abstract
of individuals observed by divers manually annotating depth and bearing every 30 s, while carrying a low-speed flowmeter to estimate average speed. Here, we improved this protocol through the use of electronic measurement devices to achieve enhanced feasibility, replicability, efficiency, and safety. Bearing and depth could be precisely measured at high frequencies using a logger fixed on an optimized diving tray, which allowed us to reduce tracking duration from 10 to 5 min, to track more individuals. It also permitted studying in situ the temporal dynamics of vertical speed and direction changes. All further steps, including data entry, sensor calibration, circular statistics and 3D track reconstruction (Madwick filter), were automated within interactive pipelines, enabling us to obtain results within 3 h after dives during fieldwork. We conducted in situ trackings for a diversified set of species (32 per ocean) during developments in the Caribbean (Guadeloupe), before being routinely applied in the Indian Ocean (Maldives) with a majority of trackings successfully carried out (74%) despite offshore conditions. High individual orientation accuracy, combined with great swimming/sinking abilities possibly dependent on depth/current, suggests that pelagic larvae/juveniles can swim in a correlated random-walk (CRW). This occurs even when orientation cues are too scarce for a consistent orientation among species/zones to emerge (biased CRW), marking a difference with their behavior in the coastal environment. Although biophysical models of dispersal ease the development of informed conservation strategies at large spatial scales, comparisons with genetic connectivity demonstrate that only models incorporating realistic active behaviors yieldcomparable outputs. Our methodological advances overcome various obstacles preventing the measurement of the parameters necessary for active models, not only for demersal fishes, but also for any small pelagic organism in any aquatic habitat.