Ecological monitoring and assessment of pollution in rivers - Ecology of Industrial Pollution - page 132

Table 6.1. Calculating the expected probability of capture of a taxon at a test site.
See Fig. 6.1 for determination of probable end group membership

End group Probability site

belongs to end
group

% of reference sites
in end group with
taxon i present

Contribution of the end
group to the likelihood of
capture of taxon i

A 0.6 30 0.18

B 0.3 40 0.12

C 0.1 60 0.06

Note: Expected likelihood of capture (P E ) of taxon i 0.36.

to provide an overall predicted probability of occurrence for the test site. In the
example illustrated in Table 6.1 and Fig. 6.1b , the test site has a probability of
0.6 of being in biological end group A and taxon i occurs in 30% of group
A reference sites, giving a contribution of 0.18 (0.6 0.3) to the expected

probability. Using the same process, groups B and C contribute 0.12 (0.3 0.4)

and 0.06 (0.1 0.6) to the expected probability, respectively, giving an

overall predicted probability of taxon i occurring at the test site (P E )of0.36
(0.18 0.06; see Table 6.1 ). This calculation is done for each of the 642 taxa

in turn. The expected abundance is also calculated in a similar manner, using
the average observed abundance of the taxon for the reference sites in each end
group. Using the same approach and site group probabilities ensure consist-
ency in the predictions of the expected probability of occurrence and abun-
dance of taxa (Moss et al. 1987 ; Clarke et al. 2003 ). Thus, as part of the basic
output from the RIVPACS software, the macroinvertebrate taxa are listed in
decreasing order of expected probability of occurrence. The expected probabil-
ities of occurrence and expected abundance are then used to derive estimates of
expected values for a wide range of indices, which are compared to the values
derived from sampling the test site to give an observed/expected measure of site
quality.
An essential feature of the RIVPACS approach is the classification of reference
sites using the invertebrate fauna. Thus, sites with similar invertebrate assem-
blages are brought together and no assumptions are made about the environ-
mental features that influence species occurrence. Furthermore, as rivers
represent a continuum of variation, RIVPACS does not assume exact matching
to a discrete classification of reference site types. This differs from other
reference condition approaches, such as the Rapid Biological Assessment
method used by the United States Environmental Protection Agency (US EPA;
Barbour et al. 1999 ), in which environmental attributes are used to a priori
define the typology of reference sites and it is assumed that the biological
assemblages of these sites are similar within types. Norris and Hawkins ( 2000 )
provides a thorough critique of the relative merits of the RIVPACS and a-priori

þ 0.12 þ