Small Mammal Response to Clearings 127 



As habitat heterogeneity increased, the capture site for each species 

 became less predictable. Models derived from captures from all plots 

 (habitat-wide) had low R 2 values (5.6 to 26.1). These models selected 

 the habitat characteristics that described the plot(s) with the most cap- 

 tures. Models derived from captures on separate treatments or aspects 

 (habitat-specific) had higher R 2 values (25.6 to 62.1) than habitat-wide 

 models. These models selected characteristics useful in predicting cap- 

 ture sites within a treatment or aspect. In the following discussions, 

 standardized partial regression coefficients are indicated parenthetically. 



Sorex fumeus. — A habitat-wide model selected number of logs 

 (SPRC = +0.18) and distance to a rock (-0.12) as important to capture 

 of S. fumeus, and both characteristics were correlated (/ > <0.01) with 

 capture (r = +0.20 and -0.16, respectively) (Table 4). Captures were pre- 

 dictable on T46 (R 2 = 47.0), clearcuts (R 2 = 25.6), and southfacing slopes 

 (R 2 = 27.2). Log characteristics (number, diameter, and length) were 

 significant (P<0.05) in habitat-wide and in habitat-specific correlation 

 and multiple regression analyses. Sorex fumeus was most apt to be 

 caught on clearcuts or heavily treated plots on northfacing slopes with 

 sparse understory (> 8 understory species per 4m 2 , < 45 understory 

 stems per 4 m 2 ), within 4.4 m of a rock, and within 1 .7 m of an 8 to 14 

 m long log. 



Blarina brevicauda. — A habitat-wide model selected seven charac- 

 teristics as important to capture of B. brevicauda and four of these vari- 

 ables were also correlated (P<0.05) with B. brevicauda capture: number 

 of logs (SPRC=+0.31), number of trees (SPRC= -0.16), slope (SPRC = 

 +0.21), and basal area (SPRC = -0.09) (R 2 = 26.1). Relatively many logs, 

 high understory density, and high understory cover were important in 

 habitat-specific models. Gottschalk and Shure (1979) reported high leaf 

 decomposition rates and high microarthropod populations on herbicide- 

 treated forest floors, but Getz (1961) suggested that leaf cover is impor- 

 tant in maintaining high humidity within soricid tunnels. Logs may pro- 

 vide moisture-maintaining cover for B. brevicauda and S. fumeus in 

 areas with low leaf cover and high food availability. Blarina brevicauda 

 was most likely captured at the edge of a plot on a northfacing slope > 

 47%, with > 21ogs per 4 m 2 , and< 1 tree per 4 m 2 (R 2 = 27.5). Geier and 

 Best (1980) also reported low plant species richness as important to B. 

 brevicauda abundance. 



Tamias striatus. — Capture sites of T. striatus were the least pre- 

 dictable (R 2 = 5.1) of those for the mammals captured. Tamias striatus 

 was most likely caught on the edge of a plot within 1.4 m of a log in an 

 area with basal area > 21 m 2 /ha, < 8 understory species per 4 m 2 , and 

 > 55 understory stems per 4 m 2 . Dueser and Shugart (1978) and Geier 

 and Best (1980) reported high understory density or cover as important 



