11. Variability and Measurement Error in Extinction Risk Analysis
173
values as surrogates for measurement errors and, using a method described below,
investigated their effect on our ability to estimate the population’s viability. We
should emphasize that our use of the reported standard errors as estimates of
measurement errors is somewhat arbitrary and may be subject to controversy.
However, the standard errors probably underestimate the overall measurement
error (K. Burnham, personal communication); it is well known that there is a
strong tendency to underestimate uncertainties in empirical data (Shlyakhter
1994; and references therein). A careful analysis of the data collection protocol or
perhaps comprehensive comparison among the results from independent data
collection teams would provide far better estimates, but neither has been conducted. The values we use are adequate for our primary purpose in this chapter,
which is to explore the effects of measurement error on quantitative analyses of
extinction risk rather than to make precise statements about the Spotted Owl.
Number of Territories
Although the number of breeding territories strongly affects the population’s
susceptibility to extinction, this number is not known with much precision for the
Northern Spotted Owl on the Olympic Peninsula. Estimates range from more than
300 to about half this number (Hanson et al. 1993; Forsman 1994; Kenney 1994;
Holthausen et al. 1994). Hanson and co-workers (1993, Table 1) report 163 known
breeding pairs on the Olympic Peninsula. A recent study reports the number of
confirmed breeding pair sites to be at least 155 but estimates the total number of
sites with territorial pairs on the Olympic Peninsula to be between 282 and 321
(Holthausen et al. 1994, Table 2).
If the current number of territories includes sites that have already sustained
some kind of environmental impact, it may overestimate the real number of
territories on which owls can persist and breed over the long term. A criterion
proposed by the U.S. Fish and Wildlife Service for a rule under section 4(d) of the
Endangered Species Act suggests that territories must have more than 40% suitable habitat within a 2.7-mile radius of each owl site (USFWS 1993). Kenney
(1994) estimated in a graphical analysis that less than 10% of extant territories are
currently below this criterion, although there is still considerable uncertainty with
respect to what the criterion actually implies about the number of territories (cf.
Holthausen et al. 1994, Figure 8).
Because the number of available territories is both important and poorly
known, we investigated a range of assumptions about its value so as to bracket the
variation in risk of extinction that could result. For the high value, we chose 300
territories, which may be an optimistic estimate on the number of territories (cf.
Holthausen et al. 1994). For the low value, we chose 200 territories, following an
estimate in USDI (1992), as a pessimistic estimate of the number of territories. If
this interval includes the true number of suitable breeding sites, then the actual
probability of extinction is likely to fall somewhere between the values estimated
for these bounds.
173
values as surrogates for measurement errors and, using a method described below,
investigated their effect on our ability to estimate the population’s viability. We
should emphasize that our use of the reported standard errors as estimates of
measurement errors is somewhat arbitrary and may be subject to controversy.
However, the standard errors probably underestimate the overall measurement
error (K. Burnham, personal communication); it is well known that there is a
strong tendency to underestimate uncertainties in empirical data (Shlyakhter
1994; and references therein). A careful analysis of the data collection protocol or
perhaps comprehensive comparison among the results from independent data
collection teams would provide far better estimates, but neither has been conducted. The values we use are adequate for our primary purpose in this chapter,
which is to explore the effects of measurement error on quantitative analyses of
extinction risk rather than to make precise statements about the Spotted Owl.
Number of Territories
Although the number of breeding territories strongly affects the population’s
susceptibility to extinction, this number is not known with much precision for the
Northern Spotted Owl on the Olympic Peninsula. Estimates range from more than
300 to about half this number (Hanson et al. 1993; Forsman 1994; Kenney 1994;
Holthausen et al. 1994). Hanson and co-workers (1993, Table 1) report 163 known
breeding pairs on the Olympic Peninsula. A recent study reports the number of
confirmed breeding pair sites to be at least 155 but estimates the total number of
sites with territorial pairs on the Olympic Peninsula to be between 282 and 321
(Holthausen et al. 1994, Table 2).
If the current number of territories includes sites that have already sustained
some kind of environmental impact, it may overestimate the real number of
territories on which owls can persist and breed over the long term. A criterion
proposed by the U.S. Fish and Wildlife Service for a rule under section 4(d) of the
Endangered Species Act suggests that territories must have more than 40% suitable habitat within a 2.7-mile radius of each owl site (USFWS 1993). Kenney
(1994) estimated in a graphical analysis that less than 10% of extant territories are
currently below this criterion, although there is still considerable uncertainty with
respect to what the criterion actually implies about the number of territories (cf.
Holthausen et al. 1994, Figure 8).
Because the number of available territories is both important and poorly
known, we investigated a range of assumptions about its value so as to bracket the
variation in risk of extinction that could result. For the high value, we chose 300
territories, which may be an optimistic estimate on the number of territories (cf.
Holthausen et al. 1994). For the low value, we chose 200 territories, following an
estimate in USDI (1992), as a pessimistic estimate of the number of territories. If
this interval includes the true number of suitable breeding sites, then the actual
probability of extinction is likely to fall somewhere between the values estimated
for these bounds.