Unraveling planktonic ecosystems dynamics using imaging sensors on BGC-Argo floats

Simultaneous detection and sizing of plankton and marine particles is now possible at global scale with the Underwater Vision Profiler 6 (UVP6) mounted on BGC-Argo floats. Combined with other biogeochemical sensors, the UVP6 delivers Essential Ocean Variables (EOVs), from nutrients to plankton and detritus, critical for monitoring and modeling. To date, over a hundred of UVP6 have been deployed by different laboratories across all oceans. When deployed on BGC-Argo floats, particle size distribution or taxa counts -obtained through embedded recognition, are typically the only available data, as the floats are generally not recovered. Here we report multi-year patterns of plankton and particles obtained from four successful deployments and recoveries at different latitudes, ranging from the equator to 50° South and depths down to 2000 m. Objects larger than 0.6 mm were classified using machine learning recognition (for plankton and particle) and k-means clustering (only for particles) methods. To date, five morphological categories of marine snow (particles > 500µm) were defined, based on shape, darkness, and structural heterogeneity, while plankton images were validated by experts in 20 broad categories. We show how these results can be used to assess plankton diversity, detritus composition, carbon vertical flux, and attenuation down to the bathypelagic layers in a wide range of environmental conditions. In cases of low mesoscale activity, results show that different phytoplankton blooms produce different marine snow morphotypes having different fates. Dense marine snow is found to be the most exported and also the deepest (down to 2000 m depth). Other morphotypes, such as filaments or porous marine snow, were generally not exported below the surface layer. Size and morphology were important to determine marine snow sinking speed. In high mesoscale activity, the steady marine snowfall is disrupted by ocean horizontal and vertical circulations and intermittent export events are observed down to 600m depth. When fully integrated in a global network of BGC-Argo floats, underwater cameras will complement existing global observations of biogeochemical variables and small planktonic organisms, detected by optical sensors, by also capturing data on larger organisms and particles.