Environmental Engineering Reference
In-Depth Information
Box 2.10
Static life table analysis of survival from white-tailed deer harvest.
The table below shows the numbers of white-tailed deer in each age category
killed by hunters in one season in part of Michigan, USA (Eberhardt 1969).
Reasonably precise survival estimates are obtained for the first three age classes,
but the confidence intervals rapidly become very wide as age increases. Given that
there is no evidence of systematic changes in mortality rate after this time, it may
be sensible to calculate a single survival rate for deer aged over 5.5 and older, i.e.
19/35
0.54. Note that the youngest possible age category (0.5) does not
appear because this group is known to be under-represented in hunting offtake.
In order to estimate the full survivorship pattern, alternative methods would be
needed to estimate rates of mortality between 0 and 1.5 years of age. The standard
error is estimated using the approach described by Skalski
et al
. (2005b, p. 162).
Age
Number
Survival rate
Standard error
a
sampled
n
a
S
a
n
a
1
/n
a
SE(
S
a
)
1.5
425
0.64
0.05
2.5
274
0.54
0.06
3.5
149
0.36
0.06
4.5
53
0.32
0.09
5.5
17
0.47
0.2
6.5
8
0.75
0.41
7.5
6
0.5
0.35
8.5
3
0.33
0.38
9.5
1
1
1.41
10.5
1
harvested carcasses, the data are often irretrievably biased. For example, some age
classes may be more vulnerable (e.g. naïve young geese may be more prone to being
shot; Wright and Boyd 1983), or harvesters may show active preferences (e.g. fish-
ers adapting their gear to target larger fish; McClanahan and Mangi 2004). A com-
mon response to the latter problem is to estimate survival only for the ages where
there is assumed to be no selectivity (Box 2.10), but this assumption cannot usu-
ally be tested and is therefore risky.
Thus static life tables are effective only when high quality, unbiased data on age
structure are available, alongside supplementary information on population
trends. It is also important to have a
large sample size
, as small samples lead to
problems with counts that do not decline continuously with age (leading to sur-
vival rates greater than 1) and extremely imprecise estimates (Box 2.10). As a rule
of thumb, a basic life table analysis is likely to require a sample of at least a thou-
sand individuals to provide reasonably precise survival estimates (standard error
0.1 up to the fourth age category).
