Environmental Engineering Reference
In-Depth Information
probability of individual species being present at a site with particular physical
and chemical characteristics. The RIVPACS system makes these predictions for
the UK, allowing the fauna at a particular site to be compared with that
predicted for a site with those same physical characteristics (Clarke et al.
2003
). The methodology is transferable to other biogeographic regions if a
suitable training dataset from uncontaminated rivers is available. A site is
designated as polluted if there are appreciable numbers of absences. This
methodology is unlikely to detect changes in community composition not
accompanied by a reduction in diversity. But this phenomenon may not be
common in freshwater, as the depauperate nature of freshwater faunas means
that there will often not be a large species pool of more tolerant species that
can replace sensitive species eliminated by pollution. Recent attempts have
been made to construct similar indices of environmental quality in the marine
environment, driven in part by the need to be able to quantify 'good' ecological
status as part of the implementation of the EU Water Framework Directive. The
AZTI or AMBI index (Borja et al.
2000
,
2007
) is based on a classification of species
into five categories ranging from disturbance sensitive species to primary
opportunists. We do not yet have a clear picture of how this index relates to
other measures of pollution impact, but some studies have found that the
value of the AZTI index is not always closely related to other measures of
disturbance (e.g., Labrune et al.
2006
).
These methods may make some further progress, but dealing with the effects
of environmental differences other than pollution presents a fundamental
barrier to progress, and one that is difficult to solve using multivariate statis-
tical analyses of community composition. A potential solution may come from
the examination of relationships between pollutant concentrations and the
abundance and/or occurrence of individual species. Some studies have applied
the concept of species sensitivity distributions to understand the relationship
between contaminant concentrations and reduction in the abundance of indi-
vidual species (Leung et al.
2005
; Kwok et al.
2008
). The result of these studies is
thresholds below which ecological effects are unlikely to be occurring, rather
than assessment of whether or not effects are occurring at any particular
location. Inter-correlation of pollutant concentrations with other environmen-
tal variables is a difficulty in these studies, but recent developments in dealing
with co-linearity in multiple regression, particularly hierarchical partitioning
(Chevan & Sutherland
1991
; Mac Nally
2002
), may provide a solution to this
(see Brown et al. in preparation, for a worked example in a different marine
context).
Direct toxicity testing of environmental samples
Given these apparent limits to ecological monitoring, are there alternative
approaches which might be more sensitive or might be less affected by
