Modelling availability bias using existing time series count data: Adélie penguins as a case study

It is widely acknowledged that counts of Adélie and other penguin species at breeding sites in Antarctica during the breeding season are influenced by the date at which they are undertaken. A count of some attribute of the population on any date may therefore need to be adjusted by some date-specific factor (termed the availability fraction) if the count is to form the basis of an estimate of the breeding population. Existing time series count data of adult Adelie penguins from a range of sites and years around Antarctica were modelled using quantile regression to estimate the range of the availability fraction at times other than the date of last egg lay. Differences in the availability curve were found between the East Antarctic and Ross sea regions. Spatial and temporal variability in the availability fraction is much higher prior to late November and after early January than through December. Implications for the planning of future regional scale surveys of land-based predators are discussed.