180 Michael R. Lusk, Michael J. Lacki, and Richard A. Lancia 
Validation of the discriminant analysis models by replication 
of the study is highly recommended (Taylor 1990), but proper ap- 
proaches to validation are usually not performed in wildlife habitat 
studies (Rexstad et al. 1990). One advantage of our study is that 
we had two years of data, and thus we attempted to confirm the 
importance of variables selected by the discriminate function analy- 
sis. Of the three variables that appeared in both years and that 
produced standardized coefficients, percent herb cover was inversely 
related in both years, percent exposed rock was directly related in 
both years, and percent canopy cover was directly related in one 
year and inversely related the next. 
The failure of the ANOVAs to detect any significant 
difference between microhabitat variables at rooted and unrooted sites 
confirms what is to be expected. Assuming deer mice are selecting 
certain microhabitat features as preferred habitat, and if there are 
no differences in mice populations between rooted and unrooted sites, 
then it seems reasonable that there would be no differences in the 
key microhabitats between rooted and unrooted sites. 
It is difficult to compare our research with other discriminant 
analysis studies of small mammal habitat because the majority of 
other studies concentrated on separating the preferred habitat of two 
or more small mammal species within a homogeneous or heteroge- 
neous habitat (Dueser and Shugart 1978, Kitchings and Levy 1981, 
Vickery 1981, Buckner and Shure 1985). We feel that our study is 
unique in that it focuses on the use of discriminant analysis to 
predict trap success for one species within a generally homogeneous 
habitat type. 
Two microhabitat variables appeared in both the 1989 and 
1990 overall models and maintained the same directional relation- 
ship both years. Therefore, these variables seem to be particularly 
important to deer mice. Mice seem to be more likely to orient 
toward traps with greater exposed rock in the area and less herba- 
ceous cover. Open areas very close to the ground, as the result of 
exposed rock and thin ground cover, would provide deer mice with 
a wide view of the terrain. This open view might be advantageous 
to the mice for two reasons. First, perhaps foraging would be fa- 
cilitated in that seeds, fruits, and insects would be more visible. 
Second, the open ground could make it easier for mice to detect 
and avoid terrestrial predators such as the long-tailed weasel 
(Mustela frenata). Open ground cover would not necessarily expose 
foraging mice to avian predators if the midstory was thick enough 
to compensate. 
