Modelling heterogeneity in detection probabilities in land and aerial abundance surveys in humpback whales (Megaptera novaeangliae)

Abstract

AbstractThe effective management and conservation of animal populations relies on statistically‐sound and replicable surveys to obtain estimates of abundance and assess trends. Surveys of cetaceans, such as humpback whales Megaptera novaeangliae, are difficult to conduct and are particularly affected by bias in detection probability. For example, the probability of detection of whales from land decreases substantially with increased distance from the platform. This distance effect is also true for aerial surveys, combined with the problem that animals are unavailable for detection (underwater) whilst in the field of view. We present a novel approach that combines corrected double‐platform land surveys with corrected aerial surveys to obtain a robust estimate of g(0), the probability of detection on the survey line, for aerial surveys of migrating humpback whales. Several sources of heterogeneity in detection probabilities were identified within the land and aerial surveys (including group composition, bearing of first sighting, number of groups being tracked simultaneously and cloud cover). After including these into our estimate of ĝ(0), we found that only 29% of available whales are being detected on the survey line (ĝ(0) = 0.288), which is a considerably smaller estimate than many available for humpback whales using other methods. Incorporating heterogeneity into the population surveys shows that we are likely to be underestimating the population size of whales on the east coast of Australia. The implications of this result for their conservation and management in light of increased whale‐human conflict is discussed.