Wild Hog Rooting 
181 
The understory distance variable is perhaps the easiest to 
confirm and interpret. Trap success decreased as distance to under- 
story trees increased, which might be a reflection of available 
refuges or a measure of midstory canopy. Vickery (1981) found 
that deer mice used areas with heavy midstory cover which may be 
related to predator avoidance. The heavy midstory would provide 
cover from aerial predators while at the same time providing close- 
ly accessible escape routes from terrestrial predators. Lockard and 
Owings (1974) hypothesized that bannertail kangaroo rats ( Dipodomys 
spectabilis) seasonally vary foraging patterns to avoid exposure on 
moonlit nights because of increased predation pressure. The prefer- 
ence of deer mice for a heavy midstory cover, coupled with the 
importance of a wide view at ground level, suggest a habitat pref- 
erence that allows protection from both terrestrial and aerial preda- 
tors while at the same time providing optimal foraging opportunities. 
The inverse relation of trap success to percent evergreenness 
agrees with the findings of Kirkland and Griffin (1974) that deer 
mice avoid coniferous areas. Most of the individual trap sites with 
a high ratio of evergreenness were located near the edges of the 
trap grid where beech forest faded into spruce-fir forest. The ap- 
pearance of percent evergreenness as an important variable prob- 
ably represents marginal deer mouse habitat. Avoidance of these 
areas may be related to the preference of the more dominant 
Clethrionomys gapperi (Crowell and Pimm 1976) for coniferous 
areas in the Great Smoky Mountains National Park (Linzey and 
Linzey 1971). Competition may be a factor in microhabitat selec- 
tion only along the fringes of primary deer mouse habitat. 
Other variables in the models are difficult to interpret. It is 
doubtful that the presence of spring beauty is of any biological 
importance. This vernal herb was very difficult to detect, and the 
few times it was detected it did not seem to be contributing any- 
thing to the habitat requirements of the mice. Greater star chickwood 
is a fairly ubiquitous herb in the beech forests, and there is no 
obvious explanation for its appearance as a variable. Perhaps, chick- 
weed had set seed by our trapping dates, thus mice were orienting 
toward this plant as a food source. We have no ready explanation 
as to why these variables appeared in the model. 
Neither can we offer an explanation for canopy cover being 
directly correlated to trap success one year and inversely correlated 
the next. This might simply be a stochastic artifact, or it could be 
related to the nature of the statistics themselves. Dr. Gerig (per- 
sonal communication) stated that it was not surprising that a vari- 
able appearing in one year’s analysis might switch signs if it appeared 
