Trackline detection probability for long-diving whales
Jay Barlow
Abstract
Long-diving whales such as dwarf and pygmy sperm whales (Kogia spp.) and beaked whales (Mesoplodon spp., Ziphius cavirostris, and Berardius bairdii) are often missed on visual line-transect surveys because they do not always surface within an observer’s field-of-view. I develop a simulation model to estimate the probability of detecting these species. This model differs from similar models in that it more completely considers parameter uncertainty. Search parameters are estimated from data that were collected on the behavior of observers using 25x binoculars during cetacean surveys. Whale dive times were estimated from visual observations. Conditional detection probabilities (given that a previously undetected animal is at the surface within an observer’s field-of-view) were fitted to observed distributions of radial sighting distances. The probability of detecting a whale on the trackline, g0 , was estimated from the simulation model. For the given methods (>50 m ship, 25x binoculars, etc.), the estimates of go are 0.35 (CV=0.29) for Kogia spp., 0.45 (CV=0.23) for Mesoplodon spp., 0.23 (CV=0.35) for Ziphius cavirostris, and 0.96 (CV=0.23) for Berardius bairdii. These estimates are most sensitive to estimates of vessel speed and the duration of long dives.