Black and white or shades of grey? Detectability of Adélie penguins during shipboard surveys in the Antarctic pack‐ice

Abstract

Summary Estimates of penguin abundance are important for developing marine ecosystem models and assessing potential competition between penguins and fisheries operations. Most shipboard surveys of penguins use strip transect (ST) or conventional distance sampling (CDS) methods to estimate abundance, but the assumptions of these methods are largely untested. To test their validity for surveys of Adélie penguins in the Antarctic pack‐ice, we recorded mark–recapture line‐transect data and estimated detectability using a point‐independence (PI) analysis. Contrary to ST assumptions, detectability declined markedly with distance from the transect line and varied with group size, substrate and observer position. Estimated detection probabilities across a 300‐m strip width, which has frequently been used in shipboard surveys, ranged from 0·09 for single penguins in water to 0·91 for groups of > 5 penguins on ice floes. Contrary to CDS assumptions, only two‐thirds of detections close to the transect line by one observer team were detected by the second team. Estimated detection probabilities on the transect line ranged from 0·30 for single penguins in water to 0·92 for large groups of penguins on ice. Synthesis and applications. Most shipboard surveys have not accounted for incomplete detection, potentially resulting in large negative biases that may vary between surveys. Recent theoretical improvements provide the potential for these biases to be addressed, but this requires application of more sophisticated and rigorous survey protocols. Application of PI analysis to mark–recapture line transect data demonstrated that substantial improvement to abundance estimates could be achieved for penguins compared with previously used methods. The protocols required for PI estimation can be applied to shipboard surveys of slow‐moving species such as penguins and seals, but may be difficult to apply to species moving faster than the survey platform, such as flying seabirds. The benefits of multiple observers are maximized only if they operate independently. For multi‐species surveys, it would be beneficial to have multiple teams of observers, each focussing on a species group. Improved estimation of marine predator abundance will facilitate the development of more realistic ecosystem models and allow improved management of the impact of fisheries operations on non‐target species.