Environmental Engineering Reference
In-Depth Information
●
On line transects, instead of attempting to measure the perpendicular
distance on the ground, measuring the direct distance
d
, together with the
angle
between the transect and the line of sight to the animal. Perpendicular
distance can then be calculated later using trigonometry:
d
perp
.
This allows the animal's location relative to the transect to be pinpointed
rapidly, minimising the difficulties of trying to keep track of moving animals
while attempting to move into position level with them on the transect.
d
direct
·sin
Accuracy of measurements
Given a rigorous approach to measurement and the appropriate tools, this assump-
tion is easy to satisfy. Whenever possible, accurate measurements should be made
with the aid of tapes, range finders, graduated markers attached to observation
platforms, clinometers or sighting compasses, as appropriate to the environment,
the form of transport and the target species. However, where limited resources
mean that suitable equipment is unavailable, it may be possible to
estimate
distances
by eye. As well as being cheap, this approach has the benefit of being
quick and easy to apply in the field. However, you should be alert to two serious
problems with estimation by eye:
●
There is likely to be consistent bias in distance estimates, leading to biased den-
sity estimates (if distances are underestimated, densities will be overestimated).
●
Estimates are liable to be approximated to the nearest round figure. This leads to
heaping of the distance frequency data around the more popular numbers
(Figure 2.4), giving a poorer fit of the detection function and reducing precision.
If distances must be estimated by eye, it is essential to spend time training
observers to do this accurately in relevant habitat and weather conditions.
Observers should be aware of the need to avoid rounding distances, and surveys
should not proceed until all are able to make unbiased estimates of distance.
In some cases, it might be easier to assign sightings to
distance categories
in the
field, rather than recording full distance information. This is acceptable, although
accurate distance measurements are still important in this approach; assignment of
sightings to the wrong categories has the same effect as biased recording of dis-
tances. It is also necessary to use a reasonable number of distance categories—an
absolute minimum of three categories is required to make meaningful modelling
of the detection function possible.
Independence of observations
For species that form close-knit
groups
(particularly social animals), individuals
are not observed independently of one another. This can be ignored without giv-
ing rise to biased density estimates, but the non-independence of individuals will
give rise to artificially narrow confidence intervals. Realistic estimation of error
therefore demands that the unit of observation is the group. In this case, distance
analysis provides an estimate of the density of groups, and this is multiplied by the
