Environmental Engineering Reference
In-Depth Information
without harvesting effort. If abundance is to be calculated over large, heteroge-
neous areas, an aggregated CPUE should be calculated as the mean of local values
weighted by area, not total catch over total effort (Walters 2003). This clearly
requires at least some harvesting effort in all parts of the area; no inference can be
drawn about any areas lacking significant harvesting effort. The
spatial scale
of
sampling must therefore be chosen on the basis of the timescale of sampling and
patterns of harvester movement within that timescale. The problem of
changing
harvest efficiency
might be avoided by restricting analysis to periods of constant
harvesting technology and behaviour. However, problems arising as a result of
either
harvester or animal behaviour
often prevent the reliable use of catch-effort
methods. For example,
Pteropus
fruit bats are commonly shot at colonial roosts on
Pacific islands (Brooke 2001; Brooke and Tschapka 2002), and because these
roosts can be large and stable, many bats might be harvested before any reduction
in CPUE can be detected. Caution is therefore needed unless a strong case can be
made for the absence of such behaviour-related problems.
Box 2.5
Estimating abundance from catch data, and the method in action:
pig harvest in a National Park.
Given two consecutive abundance indices,
a
1
and
a
2
, and catch
C
, initial popula-
tion size is:
N
ˆ
1
C
1
a
2
/
a
1
There are two possible ways to apply this model:
●
An independent abundance index might be used, for example, a sighting
rate from direct counts of live animals, or the density of fresh dung (but see
Section 2.3.6.1 for caveats on indirect sign indices).
●
The variation in harvesting effort might be quantified, for example, total
time spent searching, total distance travelled by harvesters, or the total num-
ber of traps or fishing lines set. In this case, CPUE is used as the abundance
index. In this approach, both catch and effort are usually quantified in full,
but this does not have to be the case; while the absolute total catch is
required, CPUE may be estimated from a representative sample of the har-
vest if this is more feasible.
In practice, results are much more robust when based on a series of several
harvesting periods over which the population becomes progressively depleted.
This approach requires a model of population size from each period to the
next (a
dynamic model
). A very simple model for this purpose defines fitted
population size at one point in time as the previous fitted value minus the
observed catch:
N
ˆ
t
1
N
ˆ
t
C
t
