178 Michael R. Lusk, Michael J. Lacki, and Richard A. Lancia 
significantly different than zero (Kappa = 0.46, P <0.0001). Stan- 
dardized coefficients (Table 4) indicate trap success was directly 
related to percent canopy cover, percent greater star chickweed, and 
percent exposed rock, but inversely related to percent total herb 
cover, percent evergreenness, and percent spring beauty. The group 
centroid on the discriminant axis for successful traps (0.50) was 
significantly different (F = 9.74, P = 0.0001) from the group controid 
of unsuccessful traps (-0.56). This suggests that the discriminant 
function could separate successful and unsuccessful traps based on 
the habitat variables under consideration. 
We ran a series of one-way ANOVAs on the six variables 
that were included in the final model to determine if these vari- 
ables could separate rooted and unrooted sites (Table 5). None of 
the six variables was significantly different between rooted and un- 
rooted sites. 
In 1990, the multicollinearity test revealed that no variables 
had to be removed from the analysis because of interrelatedness. 
PROC STEPDISC chose five variables for the model (in order of 
entrance): distance to closest understory tree, percent canopy cover, 
percent total herb cover, percent spring beauty, and percent exposed 
rock (Table 4). As compared to the 1989 model, one new variable 
(understory tree distance) was added and two variables (percent ever- 
greenness and percent greater star chickweed) were deleted. We re- 
moved percent spring beauty from further analysis because of a 
scarcity of observations that resulted in a covariance matrix that 
Table 5. ANOVAs to separate rooted and unrooted sites, Great Smoky Mountains 
National Park, performed on variables chosen by discriminant analysis (Table 4). 
Variable 
Rooted 
mean 
Unrooted 
mean 
F 
P 
1989 
% herb cover 
1.18 
1.13 
0.16 
0.71 
% evergreenness 
0.04 
0.21 
4.04 
0.11 
% canopy cover 
0.99 
1.09 
3.84 
0.12 
% star chickweed 
0.17 
0.13 
0.34 
0.59 
% spring beauty 
0.001 
0.02 
0.84 
0.41 
% exposed rock 
0.04 
0.003 
4.19 
0.11 
1990 
understory distance 
1.38 
1.45 
0.08 
0.79 
% canopy cover 
1.01 
1.07 
0.73 
0.44 
% herb cover 
0.88 
0.94 
0.32 
0.60 
% exposed rock 
0.01 
0.008 
0.12 
0.74 
