Identifying narwhal vocalizations to assess marine conservation areas in the changing Arctic

Climate-change related increases in sea surface temperatures and associated declines in sea ice in the Arctic are driving phenological shifts in habitat use of several species of marine mammals, which are also facing higher noise levels from increasing human activities. While some protected areas have been established, their effectiveness for species of interest, such as narwhal, have not been assessed. Sound is important for communication and navigation in marine mammals, which makes passive acoustic monitoring an appropriate tool for studying these species, being particularly effective in remote and harsh Arctic environments. Recordings collected in the Disko Fan and Davis Strait Conservation Areas (southern Baffin Bay), known habitat for narwhals, are analyzed using a new method for identifying narwhal whistles in long-term passive acoustic datasets. We identify narwhal whistles by assuming that any whistles produced around the same time as narwhal echolocation clicks (manually identified on spectrograms) were produced by narwhal. These whistles are then used to train a new deep learning detector to process long-term passive acoustic datasets for narwhal presence. This is an important step toward understanding the impacts of climate change on the distribution of this species and the effectiveness of conservation areas.