124 William C. McComb and Robert L. Rumsey 



plant taxon on each 4 m 2 plot. Total cover, total density, plant species 

 richness, and Shannon-Weaver plant species diversity were calculated 

 for each station during both sampling periods. Habitat characteristics 

 on treatments and aspects are described in detail by McComb and 

 Rumsey (1981). Correlation and stepwise regression were used to iden- 

 tify microsite characteristics important to the capture of each species. 



Since stepwise regression selects variables in a progressive series, 

 some characteristics potentially important to small mammal captures 

 may not have been selected if they were highly correlated with a pre- 

 viously selected variable. Correlation used in conjunction with stepwise 

 regression ensured that any highly correlated microsite characteristics 

 important to small mammal capture were not overlooked. A compari- 

 son of PRESS (predicted residual sum of squares) statistics was used to 

 objectively select the best model produced by the stepwise procedure. 

 Each model was entered into the General Linear Models (GLM) proce- 

 dure with the print (P) and confidence limits (CLM) options specified 

 within the Statistical Analysis System (SAS 79) to provide the PRESS 

 statistic (Helwig and Council 1979). The strongest model was assumed 

 to be the one with the lowest PRESS statistic. Microsite characteristics 

 that were distributed linearly with respect to small mammal capture 

 were identified with this method. 



RESULTS 

 Small Mammal Captures 



We captured 385 mammals in 3,250 trapnights (Table 2). Peromys- 

 cus leucopus accounted for 76.4% of the captures, followed by Blarina 

 brevicauda (6.8%), Sorex fumeus (6.5%), Microtus pinetorum (3.9%), 

 Ochrotomys nuttalli (3.6%), and Tamias striatus (2.9%). Species rich- 

 ness and species diversity were similar among treatments (x=4.3, 0.467, 

 respectively) and among aspects (x=3.0, 0.833, respectively) (Z^O.05). 

 With the exception of T68, more mammals were captured on clearcut 

 and herbicide-treated plots than on control plots (/ > <0.05) (Table 2), 

 and more mammals were captured on northfacing slopes than on south- 

 facing slopes (P<0.05) (Table 3). 



More P. leucopus, O. nuttalli, and S. fumeus were captured on 

 clearcuts than on control plots (/K0.05) (Table 2). Kirkland (1977) 

 reported an increase in S. fumeus abundance, and only a slight increase 

 in P. leucopus abundance, after clearcutting at West Virginia sites. We 

 found no differences in relative abundance of B. brevicauda or S. 

 fumeus between herbicide-treated plots and control plots; however, with 

 the exception of T46, 100% more P. leucopus were caught on herbicide- 

 treated plots than on control plots (Table 2). More M. pinetorum were 

 caught on M91 plots than on any other except T23 plots (/K0.05). 



