Environmental Engineering Reference
In-Depth Information
finding out the cause of these declines, but a recent study by Wolf and Mangel
(2004) uses all the available spatially disaggregated population data to explore the
strength of support for a range of possible hypotheses involving interactions with
fisheries, food quality and quantity and predation. This approach is one of an
'ecological detective' (Hilborn and Mangel 1997), letting the data drive the mod-
elling process. They showed strong support for the hypothesis that reductions in
the quantity and quality of fish available for the sea lions have a strong effect on
fecundity and pup survival, and also that predation by killer whales has a moder-
ate effect on sea lion survival. In this case, exploitation has had an indirect effect
on sea lions through fisheries rather than through direct harvest of the sea lions
themselves.
Source
: Wolf and Mangel (2004).
●
Using indices such as
Akaike's Information Criterion
(AIC), which compares
between models, giving a measure of which of the models under test is most
likely to be true, given the data (Burnham and Anderson 2002). This can then
be used as the basis for management, or for understanding where our main
uncertainties lie, and so what more information we need. For example, we
may find that with the current data, we can not distinguish between the dif-
ferent potential CPUE : abundance relationships, because there are not
enough data at the extremes of low population size or catch rate. This is of
concern as it means there is a potential risk of population crash as catch rate
increases.
We revisit this approach in Sections 5.4.4 and 7.4.2.
As particular approaches to conservation are tried in a range of locations, data
become available that can be used to generalise about which approaches work best
where, and which factors predispose them to success (Box 4.6). Combining the
quantitative results from a number of studies statistically can produce a more
robust analysis than looking at each case individually, because the sampling error
that afflicts each individual case is reduced. This allows the estimation of the size of
the effect of one or more factors on the target variable. This type of analysis is
known as a
meta-analysis
(Gurevitch
et al
. 2001). Still rigorous but less quantita-
tive is a
systematic review
(CEBC 2006). Less formal reviews of the available evi-
dence are also useful, and may be more realistic in conservation (e.g. Kellert
et al
.
2000). But in all cases it is important to control for biased sampling of the popula-
tion of possible studies (for example, project implementers only reporting suc-
cesses, or reviewers using only examples that support their own world-view) and
