Combining datasets into a basin wide approach to model habitats of deep-diving whales

Aim Deep-diving cetaceans are oceanic species exposed to multiple anthropogenic pressures including high intensity underwater noise, and knowledge of their distribution is crucial to manage their conservation. Due to intrinsic low densities, wide distribution ranges and limited presence at the sea surface, these species are rarely sighted. Pooling data from multiple visual surveys sharing a common line-transect methodology can increase sightings but requires accounting for heterogeneity in protocols and platforms. Location North Atlantic Ocean and Mediterranean Sea. Time period 1998 to 2015. Major taxa Ziphiidae; Physeteriidae; Kogiidae. Methods About 1,240,000 km of pooled effort provided 630 sightings of ziphiids, 836 of physeteriids and 106 of kogiids. For each taxon, we built a hierarchical model to estimate the effective strip width depending on observation conditions and survey types. We then modelled relative densities in a generalized additive modelling framework. Geographical predictions were limited to interpolations identified with a gap analysis of environmental space coverage. Results Deeper areas of the North Atlantic gyre were mostly environmental extrapolation in the predictions, thereby highlighting gaps in sampling across the different surveys. For the three species groups, the highest relative densities were predicted along continental slopes, particularly in the western North Atlantic Ocean where the Gulf Stream creates dynamic frontal zones and eddies. Main conclusions Pooling a large number of surveys provided the first basin-wide models of distribution for deep-diving cetaceans, including several data-deficient taxa, across the North Atlantic Ocean and the Mediterranean Sea. These models can help the conservation of elusive and poorly known marine megafauna.