Seamounts as hot-spots of large pelagic aggregations

Abstract

The distribution of four top predators in the Tyrrhenian Sea, a sub-basin of the Mediterranean Sea, was investigated by means of random forest regression considering depth, distance from the coast, seafloor slope, and distance from seamounts as habitat descriptors on a 2x2-nautical mile regular grid. RF results are processed to estimate variable importance and model performance. Random forest architecture reached optimal sensitivity and specificity, thus providing a consistent support tool for identifying suitable habitats. The considered species are characterized as having patched suitable habitats with a number of hot-spot areas where the different species' habitats overlap. These hot-spot areas' locations correspond to those of specific seamounts identifying the attraction effect of these topographic structures. The mean features typifying the most attractive seamounts are investigated and found to be shallow peak and base depths but wide base area and high relative elevation.