The Home Range: A New Nonparametric Estimation Technique

A new technique for estimating home range size from a sequence of independent observations of animal positions is introduced. This method describes home range in a probabilistic sense, makes no assumption about the underlying distribution, and therefore is nonparametric. The procedure is based upon an existing nonparametric technique for density estimation which uses the Fourier transform (Tarter and Kronmal 1970). A comparison is made with two commonly used methods; the minimum convex polygon method (Southwood 1966) and bivariate normal method (Jennrich and Turner 1969), as well as another recent nonparametric method developed by Ford and Krumme (1979). All these methods have certain problems, making none of them suitable for every situation. The trade—offs among these methods and some fundamental limitations of home range estimation are discussed. Evidence is given which indicates that in some situations the proposed method is preferable to earlier methods.