HOME-RANGE SOFTWARE

                     CALHOME: a program for estimating
                            animal home ranges
                                     
            John G. Kie, James A. Baldwin, and Charles J. Evans

Address for John G. Kie: U.S. Forest Service, Pacific Northwest Research Station, 1401 Gekeler Lane,
LaGrande, OR 97850, USA. 
Address for Charles J. Evans: U.S. Forest Service, Pacific Southwest Research Station, 2081 East Sierra 
Avenue, Fresno, CA 93710, USA. 
Address for James A. Baldwin: U.S. Forest Service, Pacific Southwest Research Station, 800 Buchanan Street, 
Albany, CA 94170, USA.

Key words: adaptive kernel, bivariate normal, harmonic mean, home range, minimum convex polygon,
utilization distribution

     Biologists have long sought methods to describe an animal's use of 2-dimensional space (often
called home range). The minimum convex polygon is an early example (Mohr 1947). In this method the
outermost set of data points representing an animal's locations are connected to form a polygon with no
concave sides. Developed before the advent of digital computers, it is easy to implement graphically, although 
it suffers from a number of disadvantages (White and Garrott 1990). However, the minimum
convex polygon has been widely used and is a useful method for comparative purposes with previous
studies.

     The bivariate normal assumes that an animal's locations are normally distributed in 2 dimensions
in the form of an ellipse (Jennrich and Turner 1969). However, in most cases, this oversimplifies an animal's 
pattern of spatial use.

     The harmonic mean calculates harmonic-mean values at grid points systematically overlaid on an
animal's area of use (Dixon and Chapman 1980), but it has many drawbacks. It does not produce a
probability density function, and hence it is difficult to interpret; it is sensitive to the size of the grid
arbitrarily chosen by the user; it produces misleading results when data points occur near grid-line
intersections; and it has been shown to be an inappropriate application of kernel methods (Worton 1987,
1989).

     The adaptive kernel is a non-parametric estimation procedure that is applicable to a variety of
home-range estimation problems where the assumptions of a parametric model such as the bivariate
normal cannot be met (Worton 1989, 1995). The adaptive kernel is free of the problems associated with
the harmonic mean. For example, although it requires that a grid structure be overlaid on the data points,
it is not sensitive to the size of the grid. However, the adaptive kernel requires intensive computation, and
not all home-range programs currently available offer the choice of this method (Larkin and Halkin 1994).
 We have developed program CALHOME (CALifornia HOME Range) to compute home ranges using X,Y
coordinates for a single animal. The user can choose any of the following home-range methods: minimum
convex polygon, bivariate normal, harmonic mean, and adaptive kernel. The advantages of CALHOME
are: it is menu-driven and easy to use (particularly in a classroom environment), it allows for adaptive-kernel 
analyses, and it writes the points needed to plot the home-range polygons to an output file for later
importation into a geographic information system. A detailed review of an earlier, beta-test version of CALHOME and 
other home-range programs is available (Larkin and Halkin 1994).

     CALHOME is executable on IBM-compatible microcomputers with >=500K free conventional
RAM. The source code is FORTRAN. CALHOME is designed for a color VGA monitor and graphics card.
Analyses from different home-range methods are displayed on the screen in 16 different colors at a
resolution of 640 x 480 pixels. Monochrome monitors are also supported. Hardcopy output is supported
on Hewlett Packard Plotters, Hewlett Packard Laserjet printers, and Epson dot-matrix printers. CALHOME
should be executed from a DOS prompt and not in a multitasking, windows environment.

     The user must specify a data file which contains the X,Y coordinates (and a full path if not the
current working directory), a name without an extension for the results files, and a display file (usually the
current result file base name, but which can also be results from a previous analysis). The user can
specify an optional overlay file (in ASCII format) that displays a base map showing the outline of a study
area, roads, streams, or key landmarks. CALHOME is limited to <=500 data points.

     For each home-range method the user can select up to 4 different utilization distributions (for
example, 30% and 60% core areas, and 95% and 100% home ranges). With adaptive kernel, the user
can choose a bandwidth or smoothing parameter (Worton 1989, 1995) or let the program estimate an
optimal bandwidth by assuming the data are normally distributed. An output file reports the bandwidth
used and a least-squares cross-validation (LSCV) score, a measure of how well the bandwidth fits the
data. If the data are multimodally distributed (for example, an animal using 2 separate core areas), then a
bandwidth smaller than the estimated optimum may result in a lower LSCV score indicating a better fit
(Worton 1989).

     With either adaptive kernel or harmonic mean, the user specifies a grid dimension or a grid-cell
size. The default is a grid of 30 x 30 cells overlaid on the data set, which can be increased to a maximum
of 50x50 cells resulting in a smoother fit at the expense of slightly longer computing time. The size of
individual grid cells can also be specified.

     For each analysis, CALHOME generates 3 output files: 1 containing a summary of the analysis
including the kinds, numbers, and sizes of the home ranges specified; 1 containing the X,Y coordinates
needed to plot the home-range polygons (in ASCII format, suitable with minor editing for importation into a 
geographic information system as a polygon coverage); and 1 containing a listing of distances between
data points.

     Output from CALHOME includes a screen display and a variety of hardcopy outputs (Fig. 1).
Choosing any of the hardcopy outputs will not result in immediate output but will cause a line of
instructions to be written to a queue file. After exiting CALHOME, a separate hardcopy printing program
generates the requested output. CALHOME allows several output options: plot original data points; plot an
overlay grid on the output; define a non-default scale for hardcopy output; display units in metric or English 
format; and scale axes to include all the home range polygons, all of the points in the overlay file, or both.

     Several data-preparation utilities are also distributed with CALHOME. SELECT allows the user to
select subsets from larger data sets for analysis by CALHOME. For example, the user may have a data
set containing the locations of several animals and can use SELECT to generate files for each animal.
LOCATE allows the user to convert telemetry data consisting of simultaneous azimuth readings taken at
2-6 fixed-antenna sites into estimated X,Y coordinates for the animal's location. LOCATE uses the
algorithms presented by Lenth (1981a, 1981b). LATLONG allows for conversion of latitude and longitude
(lat-longs) to UTM's used by CALHOME. SELECT, LOCATE, and LATLONG are all executed from the
DOS prompt, and process all of the records in one input file in batch mode.

     MADTRAN (U.S. Defense Mapping Agency, Fairfax, Va.) is included and allows the user to
convert between coordinate systems such as latlongs and UTM's, as well as between geographic datums
such as WGS-84 used by aircraft LORAN and GPS satellites and NAD-27 (the basis for most topographic
maps for the United States published by the U.S. Geological Survey). MADTRAN only works in interactive mode 
(the user must type each record from the keyboard), although it will save multiple outputs to a file.

     Copies of CALHOME and the accompanying utilities can be obtained by sending a blank, formatted
3.5-inch, high-density diskette to the senior author. They are also available from Illinois Natural History
Survey (FTP nhsbig.inhs.uiuc.edu, world wide web http://nhsbig.inhs.uiuc.edu) and Bird Monitor (FTP
ftp.im.nbs.gov, dial-up access 301-497-5831). Users obtaining CALHOME from these sites are
encouraged to send their names and addresses to the senior author for inclusion in a database.

     Acknowledgments. We thank J. Moore for programming the user interface for CALHOME and
Alan Prowell for programming the data preparation utilities. We also thank the California Department of
Fish and Game for funding, and the beta-test users, including Larkin and Halkin (1994), who provided
valuable suggestions on an earlier version of the program.

Fig. 1. Sample hardcopy output from CALHOME generated on a Hewlett Packard Laserjet printer using
the sample data set included on the distribution diskette, requesting a 60% and 95% adaptive kernel
analysis, specifying bandwidth = 188 m and a grid of 50 x 50 cells. The sample overlay file shows a road,
several streams, and a square, 260-ha pasture not quite perfectly aligned on a north-south axis. The
legend on the right appears on both hardcopy and screen displays. Screen displays are available in 16
colors on a VGA display and in black and white on a monochrome monitor.

                                Literature cited
                                        
DIXON, K. R., AND J. A. CHAPMAN. 1980. Harmonic mean measure of animal activity areas. Ecology 61: 1040-1044.

JENNRICH, R. I., AND F. B. TURNER. 1969. Measurement of  non-circular home range. J. Theoretical Biol.   22:227-237.

LARKIN, R. P., AND D. HALKIN. 1994. A review of software packages for estimating animal home ranges. Wildl. Soc. 
Bull. 22:274-287.

LENTH, R. L. 1981a. Robust measures of location data. Technometrics 23:77-81.

LENTH, R. L. 1981b. On finding the source of a signal. Technometrics 23:149-154.

MOHR, C. O. 1947. Table of equivalent populations of North American mammals. Am. Midland Nat. 37:223-249.

WHITE, G. C., AND R. A. GARROTT. 1990. Analysis of wildlife radiotracking data. Academic Press, San Diego, Calif. 383pp.

WORTON, B. J. 1987. A review of models of home range for animal movement. Ecol. Modelling 38:277-298.

WORTON, B. J. 1989. Kernel methods for estimating the utilization distribution in home-range studies. Ecology  70:164-168.

WORTON, B. J. 1995. Using Monte Carlo simulation to evaluate kernel-based home range estimators. J. Wildl. Manage.  
59:794-800.

John G. Kie is with the U. S. Forest Service as a Research Wildlife Biologist in the Pacific Northwest Research Station, 
LaGrande, Oregon, and Associate Staff Specialist in the San Dimas Technology and Development Center, both in California. 
He is also an Affiliate Associate Professor of Wildlife Biology at the University of Alaska Fairbanks. He received his 
Ph.D. at the University of California and has taught courses there and at Arizona State University. His current interests 
include ungulate ecology, modeling animal movements in heterogeneous landscapes, and new space-based technologies for 
tracking wildlife. James A. Baldwin is a Mathematical Statistician with the U.S. Forest Service, Pacific Southwest Research 
Station, Albany, California. Charles J. Evans is a Computer Specialist with the U.S. Forest Service, Pacific Southwest 
Research Station. Fresno. California.

Software Editor: Smith