Tools and Technology Article A Critical Review of Home Range Studies PETER N. LAVER, Mogalakwena Research Centre, P.O. Box 60, Alldays, 0909, South Africa MARCELLA J. KELLY, 1 Department of Fisheries and Wildlife, 210B Cheatham Hall, Virginia Polytechnic and State University, Blacksburg, VA 24061-0321, USA ABSTRACT No consensus currently exists for the methods of estimation of home range size or for reporting home-range analysis results. Studies currently employ a variety of disparate methods or provide inadequate information for reproducing their analyses. We reviewed 161 home range studies published in 2004, 2005, and 2006 to assess what methods are currently employed and how results are reported. We found that home range reporting was generally inadequate for reproducing studies; that the methods employed varied considerably; that home range estimates were often reported and analyzed using inappropriate methods; and that many comparisons were made between studies that may produce spurious results. We urge for minimum editorial standards for reporting home range studies and we urge researchers to follow a unified methodology for estimating animal home ranges. We supply recommendations for such reporting and methodology. These recommendations will increase the reproducibility of studies and allow for more robust comparisons between studies. ( JOURNAL OF WILDLIFE MANAGEMENT 72(1):290–298; 2008) DOI: 10.2193/2005-589 KEY WORDS asymptote, core, estimators, home range, kernel density, methodology, reporting, software. While informal home-range estimation has existed for some time, the home range as a formal concept (Seton 1909, Burt 1943) has had a relatively short and tempestuous published lifespan. Several authors have reviewed methods of home range estimation (Hayne 1949, Stickel 1954, van Winkle 1975, Worton 1987, Harris et al. 1990) and home range software (Larkin and Halkin 1994, Lawson and Rodgers 1997, Seaman et al. 1998, Larson 2001, Horne and Garton 2006). Worton (1987) provided a historical review of the development of home range estimators and compared the properties of several estimators. Since Worton (1987) and Harris et al. (1990), many studies have also made descriptive or statistical comparison of home range estimators and their various implementations. Most of these results have been summarized by the reviews of Powell (2000) and Kernohan et al. (2001). Harris et al. (1990) provided the only synthetic review of home range studies published in peer-reviewed journals (over 4 yr from 1984 to 1988) and they found that most studies were on mammals and used minimum convex polygon (MCP) home range estimation. Harris et al. (1990) provided recommendations for improved data collection and analysis but they focused largely on data collection, and they reviewed only studies which employed radiotelemetry. In the interim there have been extensive developments in the field of home range estimation and important issues in performing robust home range estimation have emerged. We reviewed recent home-range studies based on newer criteria such as the analysis of whether data collection has been adequate, the reporting of estimators, and the implementation of advances in estimators. One of the recent advances in estimators, kernel density estimation (KDE), has become prevalent and it requires several choices regarding the parameters used; thus, much of our review was focused on this subject. Adequate data collection has been a contentious issue with reference to serial autocorrelation (Swihart and Slade 1985a, Hansteen et al. 1997, Otis and White 1999, Blundell et al. 2001, Fieberg 2007), site fidelity (Spencer et al. 1990, Swihart and Slade 1997), and number of location estimates per animal (Stickel 1954, Bekoff and Mech 1984, Hansteen et al. 1997, Seaman et al. 1999, Bo ¨rger et al. 2006). Sensitivity of home range size to the number of location estimates has even led Gautestad and Mysterud (1993) to raise questions about the validity of the asymptotic home- range concept. Another contentious issue has been the delineation of home range boundaries for MCP and associated point-peeling techniques (Robertson et al. 1998). Methods for delineating the core areas of proba- bilistic home-range models are relatively well-described (Samuel et al. 1985, Samuel and Green 1988, Seaman and Powell 1990). A great body of literature makes empirical or simulated comparisons of home range estimators and these studies include Boulanger and White (1990), Worton (1995), Robertson et al. (1998), and Kenward et al. (2001). Of these estimators, KDE has been the most influential since its introduction into home range studies by Worton (1989a). Kernel density estimation presents a problem for consistency amongst studies because it has a multitude of possible implementations. Many studies from both the statistical and ecological literature have made suggestions for the optimal implementation of KDE. For instance, Bow- man (1985) and Silverman (1986) reviewed the type of smoothing (fixed or adaptive) and the method of bandwidth selection, and they reported both issues as highly influential in KDE estimates. Worton (1989a, b) recommended adaptive over fixed smoothing, but conceded that this choice had far less influence on estimates than the method of bandwidth selection (Worton 1995). Worton (1989a) recommended least-squares cross-validation (LSCV) for bandwidth selection (or some multiple thereof; Worton 1995) as did Seaman and Powell (1996) and Bo ¨rger et al. (2006; but see Hemson et al. 2005, Horne and Garton 2006). Gitzen and Millspaugh (2003) showed that the 1 E-mail: makelly2@vt.edu 290 The Journal of Wildlife Management  72(1)