Geoscience Reference
In-Depth Information
1966). Calculating back to any time, a fox's home range included the cells that
had been visited before that time and that would be visited again. This
approach allowed Doncaster and Macdonald to follow foxes' home ranges as
they drifted across the landscape. More sophisticated survival estimators could
be applied to estimate the rates at which cells were lost from home ranges and
new cells added (Doncaster and Macdonald 1996). With this approach, occa-
sional sallies are easily identified as cells visited only once.
Vandermeer (1981) cogently discussed how choosing the size of cells is a
major problem for most analyses using grids. For data on animal locations, cell
size should incorporate, in some objective way, information about error asso-
ciated with location estimates for telemetry data, information about the radius
of attraction for trapping data, information about the radius of an animal's
perception and knowledge for all location data, and knowledge of the appro-
priate scale for the hypotheses being tested. For some comparisons, cell size
must be equal for all animals; for others, cell size relative to home range size
must be equal. However, changing cell size can change results of analyses
(Lloyd 1967; Vandermeer 1981), often because cell size is related to the scale
of the behaviors being studied.
MINIMUM CONVEX POLYGON
The oldest and mostly commonly used method of estimating an animal's
home range is to draw the smallest convex polygon possible that encompasses
all known or estimated locations for the animal (Hayne 1949). This minimum
convex polygon is conceptually simple, easy to draw, and not constrained by
assuming that animal movements or home ranges must fit some underlying
statistical distribution. However, problems with the method are myriad
(Horner and Powell 1990; Powell 1987; Powell et al. 1997; Seaman 1993;
Stahlecker and Smith 1993; White and Garrott 1990; van Winkle 1975; Wor-
ton 1987). Minimum convex polygons provide only crude outlines of animals'
home ranges, are highly sensitive to extreme data points, ignore all informa-
tion provided by interior data points, can incorporate large areas that are never
used, and approach asymptotic values of home range area and outline only
with large sample sizes (100 or more animal location estimates; Bekoff and
Mech 1984; Powell 1987; White and Garrott 1990). Because all information
about use of a home range within its borders is ignored using a minimum con-
vex polygon, most analyses using this method implicitly assume that animals
use their home ranges evenly (use all parts with equal intensity), which is
clearly not the case. One can calculate a minimum convex polygon using the
