220 Monitoring Threatened Species and Ecological Communities
The notion that some species may act as indicators for other species is
controversial in ecology, due to concerns that subtle differences between species
can lead to categorically different responses to threats or management
(Lindenmayer et al. 2000). However, acknowledging that monitoring every species
of interest in a particular region or in response to a particular management may
not always be feasible, there is some appeal to the prospect of learning something
about a species that is difficult to monitor by monitoring one that is easier to
observe and is sensitive to similar environmental changes (Box 16.3).
The concept of surrogacy may extend beyond similar or related species to
environmental attributes or even threats. Lindenmayer et al. (2000) argued that
vegetation structure could be monitored as a surrogate for species that were
sensitive to change in that particular attribute. Environmental, landscape and
threat variables could be evaluated for their efficiency as surrogates of threatened
species in a way similar to the approach described in Box 16.3. For example, if an
assemblage of threatened small mammals and reptiles could be indirectly
monitored by measuring habitat attributes (e.g. vegetation structural classes), or
alternatively, by monitoring the relative density of predators (e.g. cats), the
approach described in Box 16.3 could be adapted to help determine which
surrogate should be chosen.
What could be achieved by an extra dollar spent on monitoring?
Statistical power is the probability that one would reject the null hypothesis of ‘no
change’ if a change actually exists. In Chapter 20, statistical power analysis is used
to evaluate the effectiveness of the Kakadu biodiversity monitoring program in
understanding the state and trends in threatened species populations. The chapter
demonstrates how the power of a monitoring program to detect declines depends
on: the size of the decline that is important to detect; the detectability of the
species; the frequency and intensity of the monitoring relative to the nature of
changes being observed; the background ‘natural’ variation in the system; and the
frequency of false alarms (falsely declaring a significant decline when there isn’t
one) that can be tolerated. A key result from this work is that, for a given species
and effect size of interest, the power of the monitoring design varies non-linearly
with budget (Fig. 16.4). Yet, the statistical power of monitoring programs is rarely
analysed. Consequently, for most monitoring programs, it is unclear whether they
would be in the pink, yellow or blue region of Fig. 16.4. In most of the pink region,
they are probably wasting money by looking too hard for an effect they could
detect with less effort. In the blue zone, they are wasting money because they have
such a low chance of detecting a change of interest, they may as well not bother.
The yellow zone is where you want to be. In many instances, monitoring
programs may be operating just below the yellow zone, such that, with just a little