Mass estimation of Weddell seals through photogrammetry
Ireland, Darren Scott.
MetadataShow full item record
Food resources of pelagic marine predators have traditionally been difficult to monitor and annual monitoring of food-resource availability is not currently feasible for the Weddell seal (Leptonychotes wedellii) population in Erebus Bay, Antarctica. Body mass measurements of parturient females, or their weaned pups, on an annual basis may be used as an indicator of food availability during the previous year and also provide a link between population vital rates and environmental fluctuation. Traditional methods of acquiring mass measurements, including physical restraint and/or chemical immobilization, limit the ability to sample adequately large numbers of individuals from the population of interest. Previous researchers have developed methods to estimate the mass of large seals using scaled photographs, but later application of these methods have not explicitly included uncertainty around estimates derived from predictive equations. I therefore developed the equipment and methods for estimating the mass of Weddell seals using digital photographs and image-analysis software. I then applied the method at a small scale to determine how prediction intervals may be incorporated into calculations based on mass estimates and what affect the explicit use of these intervals would have on the ability to detect differences between the mass of individuals or groups of seals. Scaled photographs of adult female and pup Weddell seals were taken from overhead, ground-level side, and ground-level head or tail perspectives. Morphometric measurements from scaled photographs (photogrammetric measurements) were then correlated using regression against the measured mass at the time of photography. Sampling occurred throughout the nursing period in order to build regression models over a wide range of masses. Resulting regression models predict the mass of adult female seals to within ±13.8% of estimated mass, and ±25.9% of estimated mass for pups. In an application of the method, differences in mass transfer between experienced and inexperienced maternal females and their pups were detected when prediction intervals were explicitly included. Detection of differences between individual seals may not be possible when prediction intervals are included with mass estimates, but estimated mass measurements should be useful for comparing mean differences across a population between years in relation to environmental variation.