Summary
Knowledge of penguin abundance at regional and circumpolar scales across the Southern Ocean is important for the development of ecosystem models and to estimate prey consumption by penguins to assess potential competition with fisheries’ operations. One means of estimating penguin abundance is to undertake aerial surveys across the pack-ice surrounding Antarctica where penguins forage. However, it has long been recognised that aerial counts and resultant abundance estimates are likely to be negatively biased unless detectability is estimated and taken into account. Mark–recapture line-transect methods were used to estimate the detectability of penguin groups resting on ice floes during helicopter surveys over the pack-ice off Antarctica. Group size had the greatest effect of several measured covariates on detectability. Despite a concerted effort to meet the central assumption of conventional line-transect sampling (all objects on the transect line are detected), this was close to being achieved by single observers only in the case of the occasional very large group of >20 penguins. Emperor penguins were more detectable than Adélie penguins. Although observers undertook an extensive simulation training program before the survey, overall they improved in their ability to detect penguin groups throughout the survey. Mark–recapture line-transect methods can provide less biased estimation than conventional line-transect methods in aerial survey applications. This improvement comes with some costs, including the need for more demanding data-recording procedures and the need to use larger, more expensive aircraft. These additional costs will often be small compared with the basic cost, but the gain in terms of improved estimation may be substantial.