Wearables for Sea Lions Deliver Precious Data for Undersea Habitat Mapping, Prediction

Researchers avoid the troubles of autonomous and remote-operated submersibles by giving the job of undersea mapping to experts.

Researchers from the University of Adelaide, the South Australian Research and Development Institute (SARDI), Honolulu's Department of Land and Natural Resources, and Port Lincoln's Department for Environment and Water have turned to an unexpected camera crew to help map underwater habitats: sea lions equipped with wearable cameras.

"Using animal-borne video and movement data from a benthic predator is a really effective way of mapping diverse benthic habitats across large areas of the seabed," explains first author Nathan Angelakis of the team's work. "These data are useful both for mapping critical habitats for an endangered species such as the Australian sea lion, and more broadly, for mapping unexplored areas of the seabed."

Exploring beneath the waves, in particular the benthic zone at the very lowest level, is a challenge for traditional robotics. Rather than using autonomous or remote-operated submersible vehicles, then, Angelakis and colleagues drafted in the sea lion โ€” a predator that stalks the very region the team was hoping to map.

The researchers equipped eight adult female Australian sea lions, from Olive Island and the Seal Bay colonies, with compact wearable cameras and tracking instruments. โ€œWe deployed the instruments on adult females so we could recover the equipment a few days later when they returned to land to nurse their pups," explains Angelakis. "We used satellite-linked GPS loggers on the sea lions, which meant we could track their position in real-time and knew when they had returned to the colony."

From the 89 hours of video footage recovered as a result, the team were able to identify and map six different habitats: macroalgae reef, macroalgae meadow, bare sand, sponge/sand, invertebrate reefs, and invertebrate boulder. Putting recovered data into a machine learning model resulted in a large-scale map predicting habitats across a swathe of Southern Australia's continental shelf.

Researchers fed 89 hours of footage, together with previously-gathered data, into a machine learning system for wide-area habitat prediction. (๐Ÿ“น: Angelakis et al)

"The sea lions from both locations covered quite broad areas around the colonies. In our calculations, we kept the area in which we predicted habitats small to maximize the precision of our predictions," Angelakis notes. "This allowed us to model benthic habitats across more than 5,000 square km [around 3,100 square miles] of the continental shelf."

The team's work has been published under open-access terms in the journal Frontiers in Marine Science.

Gareth Halfacree
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles