Using environmental variation to optimize aerial surveys of harbour seals

Abstract

Abstract Harbour seals are surveyed aerially when they haul-out to moult in August. The proportion of the population hauled out throughout the year is related to temporal, environmental, and meteorological variables. Thus, monitoring is conducted under predefined ranges of conditions. Effects of variation within these ranges are rarely reviewed. We used linear models to assess effects of time, date and weather on the difference between counts predicted by a population growth model and observed counts, based on a 30-year time-series. Our top-ranked model explained 34.4% of the variance. Survey date and its interaction with survey year were the most important variables, with higher counts earlier in August, particularly early in the time series, where surveys may not have been timed optimally to capture the peak in the moult. Cloud cover, wind speed, temperature, and interactions between these were of lesser importance; there were fewer seals on land during cloudy, windy days and on clear, warm days. These effects of weather are likely related to temperature regulation. Power analyses suggested that correction for survey conditions would allow detection of a one percentage point annual change in population growth rate with 80% power 4 years sooner than without taking survey conditions into account.