IMIS

Fine scale density map of the xenophyophores Syringammina fragillissima (ind. m-2) at a 1200m deep station off Rockall Bank (North East Atlantic) by The University of Plymouth for the Deeplinks project. The drivers of the distribution of this species are poorly known despise its Vulnerable Marine Ecosystems (VME) status and more knowledge on its biology is needed to efficiently manage its conservation. A Convolutional Neural Network was used to measure its density in 58,147 images collected by the AUV autosub6000 during the JC136 cruise in May 2016. Several environmental data layers (topographic and oceanographic) at 2.5 x 2.5 m resolution were collected by the AUV during the same cruise and used in the modelling. Random Forest regression was used to predict the density of the target species in the extent of the environmental layer representing around 3 km2.