122 William C. McComb and Robert L. Rumsey 



Our objectives were to compare the relative abundance of small mam- 

 mals among herbicide-created forest clearings, clearcuts, and uncut 

 areas on ridgetop, northfacing, and southfacing sites, and to identify the 

 habitat characteristics selected by each species. 



STUDY AREA AND METHODS 



Snag Ridge Fork watershed, in the University of Kentucky's 

 Robinson Forest, Knott County, Kentucky, contains a second-growth 

 mixed mesophytic forest typical of much of the central Appalachians 

 (Carpenter and Rumsey 1976). Ridges are dominated by shortleaf pine, 

 Pinus echinata; pitch pine, P. rigida; chestnut oak, Quercus prinus; and 

 scarlet oak, Q. coccinea. Southfacing slopes are dominated by hickories, 

 Carya spp.; white oak, Q. alba; black oak, Q. velutina; and sourwood, 

 Oxydendrum arboreum. Northfacing slopes are dominated by northern 

 red oak, Q. rubra; cucumbertree, Magnolia acuminata; and yellow- 

 poplar, Liriodendron tulipifera. 



Eighteen 0.4-ha square plots in the watershed were sampled. Four 

 plots on each of a northfacing slope, southfacing slope, and ridgetop 

 randomly received one of the following hand-broadcast herbicide treat- 

 ments in May 1976: 23 kg/ ha TORDON 10K (T23), 46 kg/ ha TOR- 

 DON 10K (T46), 68 kg/ha TORDON 10K (T68), or 91 kg/ha M-3864 

 (M91) (mention of trade names is for identification and does not imply 

 endorsement by the Kentucky Agricultural Experiment Station, Lexing- 

 ton, Kentucky). TORDON 10K is a pelletized picloram-based (4-amino- 

 3, 5, 6-trichloropicolinic acid) herbicide, and M-3864 is a 5% picloram 

 pellet. A fifth plot on each aspect was clearcut; felled trees were not 

 removed. A sixth plot on each aspect was established in the untreated 

 forest at least 75 m from any treated plot. Treated plots were 15 to 50 m 

 apart. 



Fifteen stations were established 4.3 m apart perpendicular to the 

 contour through the center of each plot. Two Museum Special snap- 

 traps, baited with peanut butter, were set within 2 m of each station for 

 one night each month on each plot, January to June 1980. Treatments 

 on each aspect were sampled simultaneously, and aspects were sampled 

 on consecutive nights. Stations were grouped into upper edge (5), plot 

 center (5), and lower edge (5). Analysis of variance and Duncan's New 

 Multiple Range Test were used to compare mean captures per station of 

 each species among treatments, aspects, and plot edges versus center. 

 The reciprocal of Simpson's Index (I/SPi 2 , where Pi = proportion of 

 captures in the ith treatment) was used as an index to each species toler- 

 ance (TI) to habitat changes (Geier and Best 1980). The maximum TI 

 for treatments was 6.00, and for aspects was 3.00. 



Twenty-one of twenty-nine microsite characteristics, chosen on the 

 basis of previous studies (Dueser and Shugart 1978; Geier and Best 



