Environmental Engineering Reference
In-Depth Information
consideration required to achieve representative samples is that the geographical
location of captures or observations should be well spread throughout the popula-
tion of interest.
Satisfying the fourth assumption (
instantaneous sampling
) needs careful
consideration of the duration of recapture or resighting events. Survival estimates
are likely to be biased if the recapture period is extended and coincides with a
period of substantial mortality, particularly if survival rates vary strongly over
time. The ideal is to restrict recaptures or re-sightings to a very short period of time
relative to the intervening periods. Directed recapture or resighting effort should
therefore be used in preference to opportunistic observations. In practice,
extended capture periods may be unavoidable, since robust survival estimation
also requires a reasonable recapture probability (as a rule of thumb, at least 0.2),
and this may be achievable only through prolonged effort. In some cases, particu-
larly when survival rate is relatively constant over time, bias caused by extended
capture periods may be minimal (O'Brien
et al
. 2005). Alternatively, for seasonal
species, the risk of bias can be minimised by timing recapture events to occur over
a part of the year in which mortality is thought to be minimal.
A final implicit assumption to note is that the estimated survival rate reflects
mortality and not
emigration
. In practice, the rate will often reflect a combination
of these processes, and it is often referred to as
apparent survival
to reflect this
uncertainty. Methods that can be used to separate true survival from migration are
introduced in Section 2.5.1.
Most commonly, mark-recapture survival analyses use
annual spacing
of
captures; however, it may be sensible to adjust this. First, it may be useful to
estimate survival rates during different parts of year, for example, comparing hunt-
ing and non-hunting seasons in order to assess the impacts of hunting. In this case,
the gaps between recapture events need not be of equal length. Second, the model
fitting procedures work best when survival rates are reasonably high (greater
than around 0.4) but not too close to 1, and this may require the spacing of recap-
tures to be adjusted so that survival rates in each interval meet this requirement. As
an extreme example, in very short-lived species, no individuals may survive from
one year to the next, in which case several recapture events will be required during
the year.
Dead recoveries
Just as the mark-recapture method enables survival rate to be separated from recap-
ture rate, the analysis of marked individuals
recovered dead
separates survival rate
from the probability of dying and being reported (the recovery rate). Imagine that a
group of 1000 animals is marked, and that 100 and 50 of these are recovered in each
of the following two years. The fact that the numbers recovered halved from 100 to
50 indicates a survival rate of 0.5. Patterns of marking and recovery can thus be used
to extract information about survival rate. The basic data required for this analysis
are the timing of marking and (if it happens) recovery for each individual.
