Published online by Cambridge University Press: 12 February 2010
Aerial surveys of harbour seals (Phoca vitulina) are usually carried out to provide an index of population size. This can be normalized, either by design or by post-hoc analysis to reduce the effects that date, time of day, tide and weather might have on the number of seals counted. In order for long-term trends to be determined from these counts it is assumed that the mean number of seals at a particular site does not vary during the survey period, and that the start and duration of the survey window does not vary with location or between years. This study used a combination of repeat land-based and aerial surveys to test the assumption for constancy of counts during the survey period. The study focused on harbour seal abundance at haul-out sites around the Isle of Skye in north-west Scotland. The coefficient of variation in these counts was estimated to be 15%, based on repeat aerial surveys using thermal imaging. Land-based counts were used to examine the effect of covariates on seal numbers using generalized additive modelling. This site-specific model predicted that the current aerial survey window for harbour seals in the UK, which is a three-week period during the moult, is about a week too early and that count variation could be reduced by surveying 1 1/2 hours earlier in the tidal cycle. Furthermore, the pupping period showed even higher (though more variable) abundance of hauled out seals than during the moult.