Weighted likelihood recapture estimation of detection probabilities from an ice‐based survey of bowhead whales

Abstract

Estimating visual detection probabilities of migrating western Arctic bowhead whales from independent observers on two ice‐based perches presents challenges that cannot be resolved using standard capture‐recapture methods. Distant visual sighting and resighting of intermittent whale surfacings is very unlike recapturing banded birds or tagged fish. We develop several bias correction strategies that are essential to confront difficulties introduced by sighting ambiguities, match uncertainty, group size inconsistencies, and the survey protocol. These are implemented using a weighted likelihood adaptation of a standard capture‐recapture model. This model estimates the relationship between detection probabilities and covariates associated with the sighting. We present a complete analysis of the bowhead example and discuss how the data configuration, weighting, and estimation methods presented here highlight issues that are confronted in the analysis of many other complex capture‐recapture datasets. Copyright © 2014 John Wiley & Sons, Ltd.