Finding the hotspots within a biodiversity hotspot: fine-scale biological predictions within a submarine canyon using high-resolution acoustic mapping techniques

Submarine canyons are complex geomorphological features that have been suggested as potential hotspots for biodiversity. However, few canyons have been mapped and studied at high resolution (tens of m). In this study, the four main branches of Whittard Canyon, Northeast Atlantic, were mapped using multibeam and sidescan sonars to examine which environmental variables were most useful in predicting regions of higher biodiversity. The acoustic maps obtained were ground truthed by 13 remotely operated vehicle (ROV) video transects at depths ranging from 650 to 4000 m. Over 100 h of video were collected, and used to identify and georeference megabenthic invertebrate species present within specific areas of the canyon. Both general additive models (GAMs) and random forest (RF) were used to build predictive maps for megafaunal abundance, species richness and biodiversity. Vertical walls had the highest diversity of organisms, particularly when colonized by cold- water corals such as Lophelia pertusa and Solenosmilia variabilis. GAMs and RF gave different predictive maps and external assessment of predictions indicated that the most adequate technique varied based on the response variable considered. By using ensemble mapping approaches, results from more than one model were combined to identify vertical walls most likely to harbour a high biodiversity of organisms or cold- water corals. Such vertical structures were estimated to represent less than 0.1% of the canyon's surface. The approach developed provides a cost- effective strategy to facilitate the location of rare biological communities of conservation importance and guide further sampling efforts to help ensure that appropriate monitoring can be implemented.