Application of a habitat model to define calving habitat of the North Atlantic right whale in the southeastern United States

Abstract

Spatially-explicit habitat models can impart a scientific basis for delineating critical habitats that relate species' distributions to physical and biological conditions, even in marine envi- ronments with vague and dynamic boundaries. We developed a habitat model of the relationship between the winter distribution of North Atlantic right whales Eubalaena glacialis, one of the most endangered large whales in the world, and environmental characteristics in its only identified calving ground, the waters off Florida and Georgia. Our objective was to provide a scientific basis for revising critical habitat boundaries in the southeastern USA (SEUS) and to predict potential habitat in the mid-Atlantic region north of the study area through a better understanding of the re- lationship of observed right whale distribution to environmental conditions. A long-term data set of right whale sightings from aerial surveys within the SEUS (conducted seasonally, December through March, from 1992/1993 to 2000/2001) was used in a generalized additive model to eval u - ate right whale distribution in relation to sea surface temperature, bathymetry, wind data, and sev- eral spatial variables. Model results indicated that sea surface temperature and water depth were significant predictors of calving right whale spatial distribution. The habitat relationships were unimodal, with peak sighting rates occurring at water temperatures of 13 to 15°C and water depths of 10 to 20 m. Model results indicated areas of potentially important calving habitat outside cur- rently defined critical habitat. Our semi-monthly predicted distributions, based on model results, provide managers with a range of scientifically based choices for revising critical habitat boundaries to achieve the desired level of protection. Predictions extrapolated through the mid- Atlantic suggested appropriate habitat features north of the study site, although analysis of data from more recent surveys in this region would be required to validate model results.