Paolo Ciucci, Luigi Boitani, Monica Masi, Department of Animal and Human Biology, Universite di Roma "La Sapienza", Viale dell'Universite 32, 00185 Rome, Italy

Following the traditional design of habitat use studies (i.e., use vs availability, Neu et al. 1974), habitat use by wolves has been generally assessed by radiotracking data (e.g., Ciucci et al. 1997). Radio locations, however, are discrete in time and their inaccuracy can be quite large, especially when wolves are moving in mountain areas. However, snow-tracking is a method often adopted in wolf research (e.g., assessment of winter kill-rates, scent-marking, nutritional status, behaviour, etc.) and a significant sample can provide accurate data on movements and habitat use. We utilized snow-tracking data to assess use of different habitat categories at a local scale (i.e., within the home-range). We applied logistic regression to estimate a resource probability funcion (Manly et al. 1993).

In the winters from 1991 to 1995 we sampled 250 km of tracks in the snow from a pack of 2-5 wolves in the Orecchiella Natural Park in the Northern Apennines (Italy). The study area (approximately 100 km2) has been defined by connecting the outermost locations reached by wolf tracks in the snow, and corresponds to the core of the pack's territory where snow presence and conditions generally allow tracking from December through March. Wolf movements were recorded directly in the field on 1:10.000 aerial photo maps, and were subsequently transferred into a GIS (Arc/Info). Habitat variables stored in the GIS were cover type, altitude, slope, aspect, and the presence of roads and other human activity centers. The study area has been then converted into a grid (50x50 m cells) whose intersection with the habitat topologies made possible a census of all the resource units available to the wolves; the same resource units were also classified as used or unused according to the snow-tracking trajectories. Following a design I protocol (cf. Manly et al. 1993), we used logistic regression to model the probability of a given resource unit being used as a function of 7 habitat variables, among which cover type, altitude and aspect appeared to be the most significant. The resource selection function can also be utilized to test prediction of optimal movements through the territory and integrated into GIS modelling of wolf habitat suitability at large scale.

References:

Ciucci et al., 1997. J. Zoology, London 243:803-819
Neu et al., 1974. J. Wildl. Manage. 38:541-545
Manly et al. 1993. Resource selection by animals. Chapman and Hall, 177 pp.