Environmental Engineering Reference
In-Depth Information
When the area sampled by a mark-recapture survey does not cover an entire popu-
lation, it is important to realise that an isolated mark-recapture survey is effectively
only a single replicate, and does not therefore provide a sound basis for inferring
anything about the wider population. If the intention is to estimate abundance
over a wider area, a set of replicate mark-recapture surveys should be carried out,
selecting survey locations according to the criteria set out in Section 2.1 in order to
ensure a representative sample.
The basic mark-recapture method described above requires a number of restrict-
ive
assumptions
to be satisfied if it is to give unbiased estimates of abundance:
●
All individuals are equally likely to be caught;
●
The population remains constant between capture occasions (i.e. it is closed);
●
There is no loss or misidentification of marks.
It is rare that all of these assumptions will be satisfied, and foolhardy to assume
so without good evidence, as uncorrected variation in capture probability leads to
biased abundance estimates. Fortunately there are many techniques implemented
in the available software that can allow violations of assumptions to be detected
and controlled for. In this section, we give an overview of the most common rea-
sons for violations, and the ways in which they might be avoided through good sur-
vey design and appropriate analysis. The analyses introduced below generally
require complete capture histories for all individuals to be known, and individual-
specific marks should therefore be used whenever possible.
2.3.4.1 Equal catchability
Common reasons for variation in capture probability are poor coverage of the
surveyed region, intrinsic variation between individuals in their susceptibility
to capture, changes in overall capture probability with time and changes in indi-
vidual behaviour as a result of capture.
Poor coverage
can arise if capture effort
is too thinly spread, leaving holes in the effectively sampled area in which some
individuals may have no chance of being caught (Karanth and Nichols 1998). This
problem can be avoided by spreading capture effort evenly across the sampled
region and ensuring that it is sufficiently concentrated. When captures are made by
traps at fixed locations, aim to place at least two traps per minimum home range
area of the target species. As a rule of thumb, this approximately translates to a
spacing of half the diameter of a home range.
Poor coverage can be one source of heterogeneity in capture probability
between individuals, which is a major problem for mark-recapture estimation
(Box 2.4). Heterogeneity may also be due to
intrinsic differences
between individ-
uals, for example, if larger individuals are easier to catch. If heterogeneity exists but
remains uncontrolled for, abundance will be underestimated, often substantially
so. This problem is minimised if a very large proportion of the population is
caught, but must otherwise be tackled at the data analysis stage. One way to do this
is to seek measurable individual characteristics that might correlate with capture
probability (Box 2.4). These characteristics, known as
covariates
in data analysis,
