Management of terrestrial vertebrate pests - Invasive Species Management - page 175

brown tree snakes ( Bioga irregularis ) (Russell et al . 2008). h e tools for dealing
with this problem include: reducing the ability of the animals to embark at the
source (e.g. for brown tree snakes in Guam; Engeman and Linnell 1998); control
on the vessel (e.g. the 'de-rat' certifi cation requirements under international and
national maritime laws); and prompt reaction to eliminate any that do arrive (e.g.
the prophylactic control against invasive rodents on high-risk islands using long-
life bait stations; Morriss et al . 2008).
More broadly, however, there is a greater risk of new vertebrate species establish-
ing in the wild as a result of the escape or release of exotic animals held legally or
illegally in the pet trade. Managing these risks is not simple. Australia, for example,
has catalogued all the exotic terrestrial species believed to be present (Vertebrate
Pest Committee 2002), but the accuracy of the list is not known. A question
is whether inaccurate or partial lists are better or worse than no lists at all. h e
former might give a false sense of security, while the latter might induce delays
in responses. Other species imported or held illegally are unlikely to be reported.
Despite these problems, such a catalogue allows some rational consideration of
new species proposed for importation and management of any discovered to be
present but not listed.
12.3 Tools to manage established wild populations
The same set of tools may be variously used to detect, contain, eradicate, or control
established populations of vertebrates, although which tools or combination of
tools and how they are applied are obviously strategy-specifi c.
12.3.1 Detection tools
With wild animals it is usually diffi cult to quickly detect initial incursions, or
fi nd the few survivors of an attempt at eradication, or to accurately delimit dis-
tributions. To be absolutely sure no animals are present we would have to search
everywhere with an infallible detection device; this is seldom, if ever, possible. We
therefore conduct sample surveys with devices that do not always detect animal
presence, but when no animals are found it is unclear whether that is because no
animals were present or because they were present but simply not detected. If,
however, the probability of an animal being detected when it is actually present is
known (e.g. Ball et al . 2005), we can apply the Bayesian methods originally devel-
oped for military search-and-destroy tactics or civilian search-and-rescue oper-
ations (Haley and Stone 1979). These estimate the probability and confi dence that
at least one individual remains in a search area despite the lack of detection, using
the known detection probabilities of the search devices or systems (e.g. for the suc-
cessful eradication of feral pigs on Santa Cruz Island in California; Ramsey et al .
2009, or the unsuccessful eradication of musk shrews on Ile aux Aigrettes; Solow
et al . 2008). These methods also allow an explicit consideration of how much
monitoring or control is required to achieve any desired level of confi dence that