Delineating Canadian and Greenland polar bear (Ursus maritimus) populations by cluster analysis of movements

Within their circumpolar range, polar bears (Ursus maritimus) are not subject to absolute barriers. However, physiographic features do cause discontinuities in their movements. These discontinuities in distribution can be used to delineate population units. Based on satellite telemetry of the movements of female polar bears carried out in 1989-1998, we used cluster analysis to identify 6 regions within the Canadian and western Greenland Arctic in which movements appear to be restricted enough to identify distinct populations. These regions generally correspond to management units that have been previously identified as Viscount Melville Sound, Lancaster Sound, Norwegian Bay, Kane Basin, Baffin Bay, and Davis Strait. A north-south substructure was identified for the Baffin Bay population, but it was weaker than the structure identified for the 6 primary units. The 6 units were consistent with genetic information, except for the Baffin Bay - Kane Basin separation, and with mark-recapture observations and the traditional knowledge of Inuit hunters. Only 2 of 65 bears that provided telemetry information for more than 1 year were classified in different populations in different years. However, annual rates of exchange, measured as the percentage of locations outside the population boundary, ranged from 0.4 to 8.9%. Analysis of mark-recapture movements indicated no difference in large-scale movements between the sexes or long-term movements with age. Although our validation criteria for demographic closure were satisfied, the observed rates of exchange between adjacent populations suggest that population dynamics in adjacent populations may not be completely independent.