60 R. C. Pais, W. C. McComb, and J. Phillips 



We used the modified minimum-area method to estimate home 

 ranges because it miminized the chance of including areas not used by 

 an individual (Harvey and Barbour 1965, Mooty et al. 1987). This 

 method is most useful with irregularly shaped home ranges, and those 

 of white-tailed deer are usually elongated (Marchinton and Hirth 1984). 

 The cumulative (total to date) locations recorded for each individual 

 were plotted against each estimated home range to determine if the 

 estimate was accurate. Locations were recorded 24 hours apart to 

 increase the independence among observations (Swihart and Slade 

 1985). Home ranges were plotted only for individuals for which an 

 asymptote was approached (Fig. 1). 



The habitat of the study area was characterized with the aid of the 

 Kentucky Department of Natural Resources' GIS. This computerized 

 system uses information from satellite and aerial photography, U.S. 

 Geological Survey 7.5-minute topographic maps, and site inspections to 

 categorize habitat features over large land areas. Four GIS files, each 

 consisting of a habitat category, were used in this study (Table 1). 

 Appropriate sets of categories were assigned to 0.4-ha (1-acre) polygons. 

 Fifteen 10-ha site inspections were conducted to ascertain the reliability 

 of the GIS data in Knott Co. No discrepancies were detected. 



Modified minimum-area home ranges for does were digitized onto 

 a computer map for each GIS file. Each home range was then 

 repositioned at random on the same 7.5-minute quadrangle map to 

 estimate the random availability of habitat in the vicinity of the actual 

 home range. The percentage of the total area represented by each 

 habitat category was calculated for each home range according to the 

 MAP model (Berry and Tomlin 1981). Because radio locations were 

 imprecise (error polygons averaged 50 ha) relative to some map features, 

 and because deer are highly mobile, we wished to identify home-range 

 selection rather than selection of patches within home ranges (second- 

 order selection of Johnson 1980). The percentages represented by each 

 habitat category in actual randomly available home ranges of both 

 resident and relocated does were compared by Student's /-tests. Data 

 approximated normality because percentages of habitat types had narrow 

 ranges of variability in this data set. Transformation was not necessary. 



RESULTS AND DISCUSSION 

 We recorded 925 radio locations from approximately 1,900 bearing 

 sets over a 3,600-km 2 area; 742 locations were of 1 1 relocated does and 

 183 were of six resident does at Robinson Forest. Most locations of 

 relocated does revealed that the does had dispersed less than 15 km 

 from the release site. 



