Variation in detection probability of polar bear maternity dens

Abstract

In surveys recording polar bear maternity dens, the proportion of actual dens in an area detected has usually been unknown. In this study, I evaluate how the proportion of dens that are observable varies with time in spring based on three 3–4 weeks ground surveys and observations of 25, 13, and 5 dens in Svalbard, Norwegian Arctic, during 2009, 2011 and 2012. The data indicate that few dens will be detected in surveys before early April, as few bears will have emerged at that time. Dens became unobservable after periods of drifting snow but are subsequently reopened within one or 2 days by occupying bears. Most bears departed dens by late April, and abandoned dens were quickly concealed by drifting snows. As such, I suggest that the optimal period for conducting an aerial survey is early to mid-April on a day with low winds and good visibility, because about half of the dens could be expected to be observable. Because emergence date, duration at den site, and abandonment dates varied considerably, uncertainty around the proportion of polar bear maternal dens detected during a survey will be large.