112 Anne M. King, et al. 



virginianus; squirrels, Sciurus carolinensis; voles, Microtus spp.; and 

 birds were frequently identified food items (Progulske 1952; Davis 1955; 

 Kight 1962; Fritts 1973; Buttrey 1974; Miller and Speake 1978). 



MATERIALS AND METHODS 



Skinned bobcat carcasses were provided by fur dealers and trappers 

 during the 1978-79 and 1979-80 trapping seasons. The carcasses were 

 weighed, sexed, and necropsied. Adult and kitten (< 8 months) age 

 classes were assigned using lacteal tooth replacement criteria (Crowe 

 1975). 



Stomach contents, prepared following Korschgen (1980), were 

 identified to order and, when possible, to species. Data were recorded 

 by percent frequency of occurrence because not all material in each 

 stomach could be identified, hence percent volume could not be meas- 

 ured precisely. Food items were identified macroscopically or, when 

 necessary, mammals were identified by microscopic examination (at 

 lOOx and 400x) of sample hairs using hair keys (Spiers 1973; Moore et 

 al. 1974). 

 Statistical Analyses and Interpretation 



The state was divided into three physiographic regions (Stuckey 

 1965), and the data were examined at statewide and regional levels. 

 Comparisons of carcass weights were made using least-squares fit to a 

 fixed-effect linear model using the Scientific Analysis System (SAS; 

 Barr et al. 1979). Weights of kittens and adults were analyzed sep- 

 arately. Food items were ranked by frequency of occurrence, and differ- 

 ences in ranks were evaluated using Wilcoxon's Rank Test (Wilcoxon 

 1945). Chi-square tests were used to evaluate differences in prey selec- 

 tion. Calculation of expected frequencies of prey items for chi-square 

 tests were based on an assumed equal probability of occurrence in a 

 particular sex and age group. 



Caution must be observed when making inferences about the pro- 

 portion of food items present in the diet based solely on stomach anal- 

 yses because: 1) smaller prey species are frequently under represented in 

 stomach analyses due to differences in digestibility (Weaver and Hof- 

 fman 1979; Merriwether and Johnson 1980); 2) some food items may 

 have been eaten incidentally as trap bait, garbage or carrion; and 3) 

 frequency of occurrence data cannot be used to evaluate food preferen- 

 ces without estimates of both prey abundance and availability. 



RESULTS AND DISCUSSION 



From October through March 1978-79 and 1979-80, 505 bobcats 



(229 $, 276 5) were collected. Weights between samples collected each 



year were not different (p>0.05); therefore, the samples were pooled for 



subsequent analyses. Among the three physiographic regions, weights 



