Conservation of endangered species - Ecological Applications: Toward a Sustainable World - page 121

Fig. 5.5 Observed koala
population trends (solid
circles) compared with
predicted population
performance (open
circles, ±1 SD) based on
1000 repeats of the
VORTEX modeling
procedure at (a) Oakey
and (b) Springsure.
Real population
censuses were not
performed every year.
(After Penn et al.,
2000.)

(a)

60

50 Observed

VORTEX

40

30

20

10

0

Year

(b)

60
50
40
30
20
10
0

Year

Table 5.2 Sensitivity
analysis of the effect of
varying fertility or
mortality at different
life stages on the
average population
growth rate

Oakey Springsure

Scenario λ E λ E

Standard 0.930 0.464 1.034 0.084

− 10% fertility 0.915 0.608 1.015 0.117

and the
probability of extinc-
tion (E) during the
modeled period (1000
repeats). The 'Standard'
values are for the
scenarios shown in
Table 5.1 and Figure
5.5. (From Penn et al.,
2000.)

λ

+ 10% fertility 0.943 0.321 1.052 0.046

− 10% newborn mortality 0.936 0.386 1.042 0.053

+ 10% newborn mortality 0.923 0.504 1.026 0.089

− 10% juvenile mortality 0.933 0.418 1.038 0.068

+ 10% juvenile mortality 0.927 0.488 1.031 0.086

− 10% adult mortality 0.936 0.323 1.041 0.042

+ 10% adult mortality 0.923 0.526 1.028 0.099

(i.e. 380 out of 1000 population models with the stated demographic values went
extinct), while that for Springsure is only 0.063.
Next, a sensitivity analysis was performed to explore the effects of small alterations
to demographic values (mortality, % breeding, etc.) (Table 5.2). This allows managers
to gauge the likely effects of measurement errors, or to estimate consequences of a
management change designed to alter a demographic parameter in a benefi cial way.
The declining Oakey population, as you would expect, has an average population
growth rate of less than 1. The population growth rate of the Springsure popula-

tion is slightly greater than 1 for all scenarios but note that this does not necessarily
lead to a zero risk of extinction, as would be expected if

λ

were greater than 1 and
constant. This is because the random elements in the model mean that some of the

λ