Data were collected with a downward facing video camera with lighting and lasers for scale. The camera recorded to tape at the surface. Geographic position was annotated onto the screen from a GPS and depth was recorded by a depth gauge in the camera's field of view (depths were later adjusted for tide). Three transects were conducted at 26 sites by slowly drifting from either deep to shallow or shallow to deep. The camera would generally fly at about 1 m off the seabed, periodically touching down. An observer would later watch the recorded videos and assign habitat types to the images.