Sublittoral macrofauna was sampled along a putative pollution gradient at 6 sites in Frierfjord/Langesundfjord. Norway, during the GEEP Mlorkshop. Data were subjected to a variety of multivariate statistical analyses which discriminate between sites on their faunistic attributes, and univariate measures of community stress were determined. Multivariate analyses produced generally slmilar results. Univariate stress measures in combination ranked the sites in order of increasing disturbance. Measured levels of pollutants in the sediments correlated poorly with the multivariate 2-D configurations and with the univariate measures of disturbance. It was concluded that water depth was the overriding factor controlling community structure, and that this masked any possible effects of pollution. The 3 deepest sites d~splayed the most obvious signs of stress, and this was attributed to seasonal anoxia in the deeper parts of Langesundfjord. In order to establish cause and effect relations between measured levels of pollutants (copper and hydrocarbons) and community responses, boxcores of sublittoral sedlment were subjected to 4 levels of contamination in a mesocosm experiment. Although some of the 'classical' intuitive methods indicated possible con~munity responses, the objective methods largely failed to reveal clear-cut differences in community structure between treatment levels. The relative merits of the 'classical' and objective approaches are discussed.

Isolation of groups of species indicative of pollution-induced effects on benthic communities has traditionally been based on subjective ecological expertise. A review of more general methods covering the use of diversity indices, computer based multivariate analyses is glven. The conclusion drawn is that there are many drawbacks with these methods too and their so-called 'objectivity' is questioned. Using a modification of the log-normal distribution of individuals among species a new method is proposed which can be used to isolate objectively groups of species sensitive to pollution effects. An example is drawn from a time-series study on effects of organic enrichment on a soft-sediment community. The method suggested has a sound theoretical base, is simple to plot, utilizes all the data on individuals and species in a community and identifies groups of species that can be used in reduced scale monitoring programmes.

An objective technique for identifying potential indicator species using analyses based on the distribution of individuals among species is presented. Benthic community data from 6 different areas in N. W. Europe are analysed. In each case species indicative of environmental change in the area are identified, even from areas where such changes are small and have resulted only in minor perturbations in the local community structure. It is shown that particular species groups are characteristically indicative of each area, and that ubiquitous indicator species comnlon to many areas do not occur. The method is rapid and involves little computation; it is robust and can be used across a range of sample sizes; it is theoretically sound and allows the objective selection of species useful as indicators of pollutant effects.