Environmental Engineering Reference
In-Depth Information
Shouldered
Spiked
Heaped
Humped
Distance
Fig. 2.4
Examples of line transect data with fitted detection functions. The ideal
form has a broad shoulder of high detectability close to the line (shouldered).
The other forms commonly observed in data make robust analysis difficult, and
should be avoided through appropriate field methods as far as possible.
Spiked data usually arise from excessive attention to the centre line or point.
Humped data are usually the result of animal movement away from the observer.
Heaped data are usually the result of approximating distances to the nearest
round number.
More problematic are cases where animals
respond to observers
by moving
away from or, less commonly, towards them (Box 2.3). When such movement is
either ignored or not detected, it effectively results in an anomalous density around
the observer. In the case of animals fleeing, this leads to under-estimation of den-
sity. In practice, it will usually become quickly obvious to observers whether this is
a problem, although occasionally movement may take place in thick cover and
remain undetected. In this case, the problem might show up only as a humped
detection function (Figure 2.4). This problem cannot be fixed at the analysis stage,
and every effort should therefore be made to ensure that distances are measured to
the animal's initial location, before any movement in response to observers has
taken place. Tricks for helping to ensure this might include:
●
Using multiple observers to define the extent of responsive movements,
keep track of animal locations and increase the chances of early detection
(although this can be counterproductive if the additional observers increase
disturbance);
●
On line transects, looking well ahead to detect responsive movements as
early as possible;
