RESEARCH ARTICLE

Abstract Movements and behavior of top marine pre-dators are often closely linked with productive oceanic fronts or regional prey aggregations. Consequently, it is of interest to quantify habitat needs and preferences, which can facilitate predictions of conditions favoring persistence and success. Multivariate habitat models of movements and dive behavior of narwhals (Monodon monoceros, Linnaeus) in the eastern Canadian high Arctic and West Greenland were developed using data collected from satellite telemetry studies on three sepa-rate sub-populations. Twenty-six narwhals were captured between 1993 and 2000 and fitted with satellite-linked time–depth recorders. Geographic positions of whales at 24-h time steps were linked to dive behavior variables compressed on a daily scale, including num-bers of dives to different target depths or durations, time near the surface, daily dive rate, and travel speed. Whale movements and behavior were linked to biophysical variables in a raster format using a GIS (bathymetry, bottom topography, bottom temperature, and distance from the coastline) and analyzed using linear and gen-eralized linear mixed models, accounting for temporal autocorrelation and random variation among individu-als. Models suggested that several physical variables described narwhal behavior, predominantly bathymetry and distance from the coastline (particularly deep or long dives). Other descriptor variables, such as size category and sex, also explained portions of the model variability, particularly for shallow dives, surface time, and dive rates. Bottom temperature (C) was the strongest predictor of all dive behaviors when narwhal location data were restricted to periods with available temperatures. The bottom temperature range and gra-dient selected by narwhals on their wintering grounds often coincided with areas of concurrent high density of Greenland halibut (Reinhardtius hippoglossoides, Wal-baum) and predictable open water in winter pack ice in Baffin Bay. These quantitative habitat models made it possible to reconcile the behavioral traits of narwhals with dynamic environmental factors