Abstract
We developed models to predict foraging habitat of adult female northern fur seals (Callorhinus ursinus) using stable carbon (δ13C) and nitrogen (δ15N) isotope values from plasma and red blood cells. Binomial generalized linear mixed models were developed using blood isotope samples collected from 35 adult female fur seals on three breeding colonies in Alaska during July-October 2006. Satellite location and dive data were used to define habitat use in terms of the proportion of time spent or dives made in different oceanographic/bathymetric domains. For both plasma and red blood cells, the models accurately predicted habitat use for animals that foraged exclusively off or on the continental shelf. The models did not perform as well in predicting habitat use for animals that foraged in both on- and off-shelf habitat; however, sample sizes for these animals were small. Concurrently collected scat, fatty acid, and dive data confirmed that the foraging differences predicted by isotopes were associated with diet differences. Stable isotope samples, dive data, and GPS location data collected from an additional 15 females during August-October 2008 validated the effective use of the models across years. Little within year variation in habitat use was indicated from the comparison between stable isotope values from plasma (representing 1-2 weeks) and red blood cells (representing the prior few months). Constructing predictive models using stable isotopes provides an effective means to assess habitat use at the population level, is inexpensive, and can be applied to other marine predators.