Ecology, 90(11), 2009, pp. 3209–3221 Ó 2009 by the Ecological Society of America Sex-specific, seasonal foraging tactics of adult grey seals ( Halichoerus grypus) revealed by state–space analysis GREG A. BREED, 1,3 IAN D. JONSEN, 2 RANSOM A. MYERS, 1,4 W. DON BOWEN, 2 AND MARTY L. LEONARD 1 1 Department of Biology, Dalhousie University, 1355 Oxford Street, Halifax, Nova Scotia B3H 4J1 Canada 2 Bedford Institute of Oceanography, 1 Challenger Drive, Dartmouth, Nova Scotia B2Y 4A2 Canada Abstract. In many large pelagic animals, observing behavior is limited to observation by radio or satellite telemetry. In many cases, discriminating different behaviors from telemetry data has been a key, but often elusive, goal. Here we use state–space models (SSMs) to fit a correlated random walk (CRW) model that switches between two unobserved behavioral states (nominally foraging and traveling) to 41 male and 43 female adult grey seal (Halichoerus grypus) satellite telemetry tracks. The SSM results reveal markedly different spatial behavior between the sexes, fitting well with sexual size dimorphism and known dietary differences, suggesting that the sexes deal with seasonal prey availability and reproductive costs differently. From these results we were also able to produce behaviorally informed habitat use maps, showing a complex and dynamic network of small, intensely used foraging areas. Our flexible SSM approach clearly demonstrates sex-related behavioral differences, fine scale spatial and temporal foraging patterns, and a clearer picture of grey seal ecology and role in the Scotian Shelf ecosystem. Key words: animal movement; correlated random walk; foraging; grey seal; Halichoerus grypus; Nova Scotia, Canada; sex-specific foraging behavior; switching model. INTRODUCTION Foraging is central to an animal’s life history and ecology. Appropriately synchronizing foraging effort with reproductive costs, seasonal cycles, and environ- mental variability can mean the difference between success or failure of individuals or whole populations. For most pelagic animals, behaviors at sea are nearly impossible to observe directly. Instead, biologists have been attaching increasingly sophisticated electronic tags that record or transmit location, physiological, and environmental parameters. Satellite telemetry and other forms of tracking have filled vast gaps in our knowledge of ecology and natural history of many marine species (e.g., Stewart et al. 1989, Jouventin and Weimerskirch 1990, McConnell et al. 1992, Prince et al. 1992, LeBoeuf et al. 2000, Shaffer et al. 2006). The extent of ranging by species such as northern elephant seals ( Mirounga angustirostris), Wandering Albatross (Diomedea exu- lans), or Sooty Shearwaters (Puffinus griseus) was far beyond expectation. Tagging studies today are growing in number as tags become smaller and more reliable. Thousands of animal tracks have been logged around the world. Methods for analyzing tracking data, however, have not kept pace with the rapid improvement of tag technology. There have been some advances, perhaps the most significant is the idea of treating animal tracks as correlated random walks (CRW; Kareiva and Shigesada 1983, Marsh and Jones 1988, Turchin 1998, Okubo and Gross 2002). The idea of using CRWs to understand animal movement is quite old, but fitting CRW models to data proved difficult (see Turchin 1998, Okubo and Gross 2002). Within the past 10 years, state–space models (SSMs) have been increasingly employed to fit CRW models to animal movement data (Anderson-Spreher and Ledolter 1991, Newman 1998, Sibert et al. 2003). The approach differs from other methods because it simultaneously fits two kinds of error: measurement error (how well the location is known) and process noise (how much an animal’s movement deviates from the model being fit). Early SSM implementations used analytical or numer- ical methods to fit models with known dynamical parameters (e.g., equations of motion; Kalman 1960), but non-Gaussian and/or nonlinear problems generally need to be solved numerically using Markov Chain Monte Carlo (MCMC) simulations or particle filters (Gelfand and Smith 1990, Doucet et al. 2001). In this analysis, we used a state–space approach to analyze the movements of grey seals (Halichoerus grypus) in the northwest Atlantic. Grey seals are abundant upper-trophic-level predators inhabiting both sides of the North Atlantic. There is increasing evidence that marine mammals can have significant top-down effects on ecosystem functioning (Bowen 1997). In addition, there have been several attempts to model Manuscript received 10 September 2007; revised 23 December 2008; accepted 5 February 2009; final version received 3 March 2009. Corresponding Editor: J. R. Sauer. 3 E-mail: gbreed@dal.ca 4 Deceased. 3209