Detailed model of shelter areas for the Cantabrian brown bear

Article history: Received 5 September 2006 Received in revised form 1 August 2007 Accepted 7 August 2007 We propose a methodology to determine critical areas for brown bear (Ursus arctos) in the Cantabrian Mountains (Spain), considered as shelter areas in which there are concentrations of winter dens and groups of resting sites. We assessed a function which enabled a precise spatial distributionmodel to be constructed. This function helps to explain the present division of the range of the bears into two isolated populations and to identify areas for protection with the aim of increasing their probabilities of survival or connection. This has been done using amultivariate analysis and by applying the results to variables modelled on a geographic information system. We took a location containing 161 bear dens and resting sites. These positions have been characterised bymodelling variables in a grid of 50-m cells derived from the topography (digital model of elevations, orientations, slopes, altitude variability in an area around each location). The variables include those that quantify distance weighting according to difficulties associated with access to different disturbance elements (infrastructures or buildings), variables that quantify distances to refuges (forests, scrubs, and rocky areas) and their shape (perimeter, thickness, eccentricity or shape of the forest and type of rocky areas). We constructed a logistic regression model with this data that locates and highlights shelter areaswith an 86% average rate of precision. This result reveals that the eastern population has shelter areas around the present range, indicating a possible expansion area; however, there is a gap measuring close to 30 km between both populations that is almost entirely lacking in shelters and that contains important infrastructures. Increasing refuge conditions in the gap to connect shelter areas will make it possible to recover the area as a corridor. © 2007 Elsevier B.V. All rights reserved.