in the habitat offering the greatest cover, whereas older, invulnerable fish foraged in
the open, where energy gain was highest (Werner et al. 1983).
Sensitivity to predation risk also underlies patterns of habitat use by many large
herbivores. For example, white-tailed deer in the boreal forests of Wisconsin and
Minnesota tend to concentrate in the no-man’s land between wolf pack territories
(Hoskinson and Mech 1976). Wolves tend to avoid going out of the area defended
by their pack, because of a pronounced risk of being attacked by hostile neighbors
(Lewis and Murray 1993; Mech 1994). This effectively creates refuges in between
territories in which individual deer are relatively safe.
One of the major difficulties in testing for risk-sensitive habitat use is finding a
sensitive way of measuring risk, ideally from the animal’s point of view. Brown (1988)
suggested that the giving-up density at feeding trays could be used as a field mea-
sure of habitat attractiveness, which should be sensitive to both predation risk and
alternative foraging opportunities in the surrounding habitat. This technique has been
successfully applied in a large number of field studies. For example, different species
of granivorous rodents in the Negev desert have different assessments of risk in the
same habitat, demonstrating interspecific differences in their perception of the risk
of predation versus energetic gain (Brown et al. 1994).
A useful way to evaluate such decision-making, which balances trade-offs among
competing risks and benefits to fitness, is known as dynamic state variable model-
ing(Mangel and Clark 1986; Clark and Mangel 2000). Although this approach is
beyond what we can cover here, it offers a powerful means of evaluating the con-
sequences of alternative behavioral activities that have complex trade-offs among energy
gain, reproduction, and mortality risk. Indeed, it may be the only way to link com-
plex sets of behaviors into a life-history framework. The monograph by Clark and
Mangel (2000) offers an introduction to the techniques of dynamic state variable
modeling, as well as describing a wide set of applications.By now it should be apparent that there are good reasons for wildlife species to choose
habitats carefully, to enhance the opportunities for feeding, while reducing the risk
of being eaten. Moreover, most species have a suite of other needs to meet, includ-
ing obtaining shelter from inclement weather, gaining access to water, or locating
suitable breeding sites, such as cavities in dead trees or burrows. Quantification of
specific habitat needs is known as habitat assessment, and this is an important area
of wildlife ecology. Much of this interest derives from practical benefits: knowing
precisely which wildlife habitats are essential allows appropriate management
decisions regarding alternative forms of land use. Moreover, good understanding of
habitat requirements can improve the odds of success when wildlife species are rein-
troduced to areas from which they were extirpated.
There are many ways to quantify wildlife habitat use. We shall focus on a recent
approach, the resource selection function(Manly et al. 1993; Boyce and McDonald
1999). Resource selection functions offer a flexible means of quantifying the degree
of habitat preference. Complex combinations of categorical and continuous variables
can be readily accommodated using this method. Moreover, the method can use a
Geographic Information System (GIS) to locate, manipulate, and analyze habitat data
of interest.
GIS is a means of linking complex geographical information on physical structure,
topographic relief, biological features, and human-made landscape elements into70 Chapter 55.5Quantifying
habitat preference
using resource
selection functionsWECC05 18/08/2005 14:42 Page 70