ANALYSING INTERACTION FROM RADIO-TRACKING
DATA: PROGRAMS FOR PCs
Three programs can be downloaded here, for analysis of
interaction between concurrently tracked pairs of individuals (DYNAMIC), overlap
and concordance in the utilisation distributions of pairs of home-ranges
(STATIC), and definitive shifts in the location of a range over periods of
weeks or months (DRIFT). For more background information, see:
The
programs run on any IBM-compatible computer using MS-DOS. For a Mac version,
use Wildtrak.
The
programs will write to, and read from, temporary files created on the default
drive. The analyses are performed most quickly if the programs are called up
from a hard disk, where the data files are also stored.
Data
should be contained in ASCII files, one file per animal with co-ordinates in
sequential order. Each data line should contain the following variables: time (hrs and minutes or decimal hours), date (with 4 digits for
the year), x- and y-co-ordinates (integers), activity (optional). For nocturnal
tracking data, remember that the date changes at 00:00 hrs. These variables can
be positioned in any order along the line, but they must occupy the same column
positions throughout the file. This is because the programs identify the
variables on the basis of the columns they occupy. The first time you run one
of the programs you will be able to create a separate file of these column
positions. This can then be called up on subsequent runs of this or the other
programs.
Try
running the programs on the sample data (ANIMAL01.TXT and ANIMAL02.TXT with COL.DEF
to identify column positions). Just type the name of the program you want to
use and then press the enter key ("DRIFT<enter>" etc.). The
programs are self-explanatory. When running DRIFT on ANIMAL02.TXT, look at
weekly, monthly, or yearly changes in range size and location (and the relation
of cumulative fix total to range size) using 5x5m grid-cells with the option of
influences and a 15-mins independence interval. When running dynamic on these
two sample files, use a time-block of 15 mins and a
critical separation distance of 10m.
The
programs will ask you to specify an independence interval. If you give 10 mins (say), then the selection will be done for you so that
only fixes spaced apart by at least 10 mins are taken
(for example, skipping every other fix where you have 5-min intervals between
fixes).
The
main function of these programs is to help you explore your data and to give
numeric values to relationships between individuals that you already know about
in a qualitative way.
Beware
of apparent dynamic interaction resulting from two animals regularly coming and
going at similar times from the same or different sleeping positions. Use only
active fixes or data from the middle of the night to get around this problem.
Note
also that instantaneous (or nearly so) fixes may be easier to obtain when the
pair of individuals are close together than when they are far apart. If so, you
could end up with a greater proportion of paired fixes at close separation than
is representative of the actual time they spend together, and the extent of
positive dynamic interaction will be over-estimated.
The
correlation coefficient for static interaction is best used in comparisons
between pairs of animals. If you want to compare the concordance in utilisation
distributions of a pair to random expectation, then it might be useful to
simulate model territories of similar size and overlap but containing random
fixes. Several runs of the model data through the static program will give a
mean + confidence intervals for the correlation coefficient under the
null hypothesis of random utilisation (see for example Taber, A.B. et al. 1994. Mammalia, 58: 61-71;
Doncaster, C.P. & Macdonald, D.W. 1997. J.
Zool. Lond., 241:
73-87).
Centre
for Biological Sciences
University
of Southampton
Life
Sciences Building 85
Southampton
SO17 1BX, U.K.
e-mail:
cpd@soton.ac.uk
Tel:
+44 (0)23 80594352