Animal Movement - Mobility Data - page 279

100 m

(b) (c)
Figure 13.1 Discretization of animal trajectories. (a) The path of an animal (in the example
a foraging deer) is perfectly recorded. (b) Path sampling at high resolution, and (c) sampling
at low resolution. Sampling is done at fixed time intervals.

(a)

biological details might be lost. Clearly the appropriate sampling depends on the
process of interest and there are not general guidelines. Thus, in order to plan
the sampling design, researchers need some preliminary understanding of the
dynamics of the process of interest. In the example of Figure 13.1 , sampling (b)
cannot be sufficient to investigating the selection of food items, while sampling
(c) can be appropriate to investigate the annual home range.
The parameterization of a discretized path is represented in Figure 13.2 a.
Till now we have considered a fixed time sampling. The distance among spatio-
temporal positions is referred as a “step.” The step itself is arbitrary and does
not represent a significant behavioral feature of the studied organism. In the
example path, during the first four steps the organism may have maintained the
same motivation and behavioral tactics so that the small observed differences in
speed and direction of steps may represent environmental or sampling nuisances
(such as precision of a GPS collar, irregularities of the terrain or presence of
obstacles for terrestrial organisms, or drifts due to wind or streams in flying or
swimming animals). On the contrary, the spatio-temporal position at t + 1-
where a sharp directional shift occurs - may represent a changing of motivation.
A move represents the distance between biologically meaningful variations in
the path. It is thus appropriate to use a different discretization of the trajectory
with variable temporal intervals (Figure 13.2 b). This approach is more natural
than the use of steps but it is difficult to identify turning points where meaningful
behavioral changes occur. The basic idea to identify the moves is that within
each move the α t are strongly correlated (i.e., are very close to 0) while the β i
are uncorrelated.

13.2.6 An Example: Foraging and Social Behavior of the Fallow Deer

To exemplify methods used to investigate animal movement we review some
studies about foraging and social behavior of fallow deer ( Dama dama ). In