tracking every hour to collect location data from 5–6 animals (i.e. a minimal
sample for statistical tests).
For a study over more than 1 year, pilot work should also involve recording
locations at 3 to 4 day intervals. This is convenient for recording dispersal and
survival data, and for indicating the seasons in which animals tend to settle.
6.4.2Recording locations
Location data reveal how and when animals move, disperse, and interact with
resources, such as food and cover. Locations close in time also show how animals
interact with each other, and hence space themselves or transmit disease. The data
may be recorded automatically, or by close approach to animals and by triangula-
tion techniques (details in Kenward 2001).
Continuous recording of locations in a pilot study builds a trajectory, which
gives movements of dispersers in detail and gradually defines a home range if the
animals are settled. However, it takes time to record each location, and a whole
trajectory is not needed to define a home range. Indeed, it is a mistake to focus on
collecting large numbers of locations, unless you need travel distances, because
locations are not statistically independent records. Not only are adjacent records
constrained by how far an animal can travel in a given time, but animals also tend
to use the same roosts, routes, and foraging sites repeatedly, often at similar times
of day. Comparisons between categories of animal, or the same animal in differ-
ent seasons, are therefore based on a single index that represents, for example, the
proportion of foraging locations in a particular habitat or the area encompassing
a selected proportion of the locations.
When analyses are based on indices that represent each set of locations, the
efficiency issue becomes “how few locations adequately estimate each index”?
This standard number of locations should be established during the pilot study,
by estimating with increasing numbers until the index becomes acceptably
stable. For example, the area of a convexpolygon plotted round 100% of the loca-
tions (X 100 -defined for brevity by unique letterswith a percentage inclusion)
initially increases rapidly as locations are added, but eventually tends to an
asymptote if the animal is settled in a seasonal home range (Figure 6.3(a)). If
animals were monitored continuously, there may be hundreds of locations
to reach, say, 95% of the maximum area. However, if locations are sampled at
intervals of 1, 2, 3 h etc. a protocol can be developed to estimate X 100 efficiently.
With 2–5 locations sampled at intervals throughout the day, 30–50 locations are
usually enough. If fewer locations are available for some animals, but the rate at
which the asymptotic areas is approached is found to vary little among animals,
152 |Radio-tagging