A Bayesian estimate of harbour seal survival using sparse photo-identification data

Survival rates have rarely been estimated for pinniped populations due to the constraints of obtaining unbiased sample data. In this paper, we present an approach for estimating survival probabilities from individual recognition data in the form of photographic documentation of pelage patterns. This method was applied to estimate adult (age 2+) survival for harbour seals in the Moray Firth, NE Scotland. An astronomical telescope was used to obtain digital images of individual seals, and high-quality images were used to document the annual presence or absence of individuals at a single haul-out site over a 4-year period. A total of 95 females, 10 males and 57 individuals of unknown sex were photographically documented during the study period. Survival and recapture probabilities were estimated using Jolly–Seber mark–recapture models in a Bayesian statistical framework. Computer-intensive Markov Chain Monte Carlo methods were used to estimate the probability distributions for the survival and recapture probabilities, conveying the full extent of the uncertainty resulting from unavoidably sparse observational data. The deviance information criterion was used to identify a best-fitting model that accounted for variation in the probability of capture between sexes, with constant survival. The model estimated adult survival as 0.98 (95% probability interval of 0.94–1.00) using our photo-identification data alone, and 0.97 (0.92–0.99) with the use of an informative prior distribution based on previously published estimates of harbour seal survival. This paper represents the first survival estimate for harbour seals in the UK, and the first survival estimate using photo-identification data in any species of pinniped.