MARINE MAMMAL SCIENCE, 18(4):940-951 (October 2002) 0 2002 by the Society for Marine Mammalogy
AN ALGORITHM TO IMPROVE GEOLOCATION POSITIONS USING SEA SURFACE TEMPERATURE AND DIVING DEPTH C. A. BECK Department of Biology, Dalhousie University, Halifax, Nova Scotia B3H 4J1,Canada E-mail:
[email protected]
J. I. MCMILLAN W. D. BOWEN Marine Fish Division, Bedford Institute of Oceanography, Department of Fisheries and Oceans, P. 0. Box 1006, Dartmouth, Nova Scotia B2Y 4A2, Canada
ABSTRACT The at-sea movement of marine mammals is an important component of their foraging ecology, but has been difficult to study. Geolocation timed-data recorders (GLTDRs) estimate positions using measured light level to calculate day length and local noon. It is well known that these location estimates are imprecise (mean error of >lo). Satellite telemetry generally provides a more accurate, but also more expensive means of monitoring movement. We evaluated the accuracy and precision of geolocation positions by comparing these locations with satellite data from Service Argos for eight free-ranging gray seals (HukboeruJ grypu) equipped with both a satellite-linked data recorder (SDR) and a GLTDR. Geolocation positions differed by 1,026.0 ? 292.28 km from the corresponding Argos locations. We developed an algorithm to correct geolocation positions by comparing surface water temperature (ST) and dive depth collected by GLTDRs with existing sea-surface temperature and bathymetry databases. The corrected positions were significantly closer (P < 0.025)to the Argos locations of these seals (94.2 ? 8.22 km). The original geolocation positions would have led to incorrect conclusions about the use of space by gray seals; however, the corrected positions can be reliably used to study the large-scale spatial distribution of individuals. Key words: geolocation, gray seals, sea-surface temperature, bathymetry, Argos. O u r understanding of the foraging ecology of marine mammals has lagged behind other aspects of their biology owing to the difficulty in studying their
BECK ET AL.: GEOLOCATION ALGORITHM
94 1
behavior at sea. An important feature of such studies is the ability to determine the movement of individuals over time. Over the past several decades a variety of technologies have been developed which make this possible. Data collected using geolocation timed-data recorders (GLTDRs) can be used to estimate the geographic position of an animal by means of solar navigation equations, given estimates of day length and local noon (Hill 1994). These estimates are derived from light levels recorded by the GLTDR at regular intervals throughout the day. Despite the promise of this technology, positions determined from GLTDRs are subject to errors caused by weather and by the behavior of diving marine mammals. Both degrade the relatively imprecise predicted accuracy of such locations (Hill 1994, Welch and Eveson 1999). Geolocation positions have been used to study the large-scale movements of pinnipeds and fish (DeLong et al. 1992, Le Boeuf et al. 1993, Stewart and DeLong 1995, Block et al. 1998, Welch and Eveson 1999). Therefore, it is important to understand how the performance of GLTDRs under field conditions compares to their predicted accuracy. Several studies have attempted to validate the positions derived from GLTDRs (e.g., DeLong et al. 1992, Stewart and DeLong 1995, Le Boeuf et al. 2000) and others have attempted to improve the accuracy and precision of such estimates by matching sea-surface temperature data collected by the recorder and that derived from remote sensing of sea-surface temperature (e.g., DeLong et al. 1992, Campagna et al. 1995, Stewart and DeLong 1995, Le Boeuf et al. 2000). Nevertheless, explicit algorithms for doing so have not been published. Here we report an algorithm for correcting geolocation position estimates using surface temperature (ST) and dive depth of the study animal, both of which are routinely collected by GLTDRs. We validated the accuracy and precision of geolocation estimates by fitting free-ranging adult gray seals (Halichoerus g t y p w ) with both GLTDRs and satellite-linked data recorders (SDRs) and then compared geolocation estimates with the location estimates calculated by Service Argos. We then assessed the performance of our algorithm by comparing corrected geolocation positions with those calculated by Service Argos for the same test animals. To examine the effect of factors other than the behavior of the diving animals on the performance GLTDRs, we also compared geolocation position estimates with GPS positions using fixed sites on land. METHODS The study was conducted between October 1996 and January 2000 on Sable Island (43”55’N, 6O0O0’W), a partially vegetated sandbar approximately 300 km southeast of Halifax, Nova Scotia, Canada. Sable Island is the largest haul-out location for gray seals in the Northwest Atlantic population. Animals congregate in large numbers on the island in May and June to molt, in late December and January to rear offspring and mate, and in smaller numbers throughout the year between foraging trips. Adult gray seals were captured onshore using handheld nets (see Bowen et al. 1992) following the spring moult or in the fall (late September and early October). Seals were weighed to the nearest 0.5 kg on Salter spring balances suspended from an aluminum tripod and anesthetized using Telazol (equal parts of teletamine and zolazepam). Males and females received an average dose of 0.45 rng kg-’ body mass and 0.90 mg kg-‘ body mass, respectively (Bowen et al. 1999).
942
MARINE MAMMAL SCIENCE, VOL. 18, NO. 4, 2002
We used two models of GLTDRs (Mk3e and MkS) developed by Wildlife Computers (Redmond, WA) to assess geolocation position estimates for individual free-ranging seals. Instruments were secured to netting and then the netting was attached to the pelage of the anesthetized animals using a 5-min epoxy. GLTDRs were positioned between the shoulders along the midline of the individual and weighed between 65 and 300 g (
