34 J. P. Phelps and R. A. Lancia 



sampling effort for coverboards and PVC pipes in the clearcut, edge, 

 and control. Due to the lack of replication, the results and conclusions 

 from this study apply to the particular site we studied. Generalization 

 to other sites is risky. 



Data analysis — Capture data were analyzed using the Shannon- 

 Weaver Diversity Index (Poole 1974). This allowed statistical com- 

 parisons among drift fence arrays in the clearcut, on the edge, and in 

 the control; and between coverboards in the clearcut and in the control. 

 Variance for each treatment was calculated based on the number of 

 captures in each treatment, and was used to calculate the ^-statistic (a 

 = 0.05) for each comparison (Poole 1974). The formulas used are as 

 follows: 



Var (//') = [r MPl ln 2 p,] - [g. lf >, lnp,] 2 + s-1 



N 2N 2 



H\-H\ 



[Var(/f ,) + Var(//' 2 )] 1/2 



d.f. = [Var(/f ,) + Var(/r 2 )] 2 

 [VarCtf',)] 2 + [Var(tf' 2 )] : 



N, N 2 



where i represents each species in the sample, pi = proportion of 

 species i in the sample, s = number of species in the sample, N = 

 number of captures in the sample. 



Assumptions of the Shannon-Weaver Index were that all species 

 present were sampled, and all were equally catchable. These assumptions 

 were not met because of the differing ability of the traps to catch 

 various species (Gibbons and Semlitsch 1981). Also, two species caught 

 by hand were never captured in a trap of any kind and were therefore 

 excluded from calculations of H' . These were the brown water snake 

 (Nerodia taxispilota) and the timber rattlesnake (Crotalus horridus). 

 Their size explains the lack of pitfall captures, as no larger snakes 

 were caught in pitfalls. Other large snakes were captured under 

 coverboards. 



Poole (1974) states that "no great error" will occur in calculating 

 Shannon-Weaver as if all the species available are present in the sample, 

 even when, as in this case, some species are not represented. Unequal 

 catchability of species should still allow relative diversity comparisons. 



