ABSTRACT. Multivariate statistical analyses of data on environmental variables and benthic fauna from 14 oil and gas fields obtained from 24 surveys collected between 1985 and 1993 are presented. At all fields oil-based drilling mud was used. The purpose of this study was to investigate contamination gradients, assess effects on benthic fauna both spatially and temporally and to evaluate measures such as diversity indices, indicator species and multivariate analysls techniques in assessment of pollution. Results from analyses of baseline surveys of environmental variables and fauna were charactensed by a lack of distinct gradients in station placement, having a typical shot-gun pattern in PCA-, DCA-and MDS-ordination analyses. Likewise there was no consistency in which environmental variables correlated with the fauna. Contamination was assessed using all the physical and chemical data in classification and PCA-ordination analyses Clear patterns were found using 4 categories, conveniently termed initial, moderate, severe and gross. The categories were usually apparent as rings radiating from the platform. Initial contamination of the outermost areas at most fields was shown as elevated levels of barium and total hydrocarbons (THC) and sometimes also by elevated levels of zinc, copper, cadmium and lead. Three fields were studied in particular and showed contaminated areas of over 100 km2 (Valhall), over 15 km2 (Gyda) and over 10 km2 (Veslefrikk). After a period of 6 to 9 yr contamination had spread, so that nearly all of the outermost stations 2 to 6 km away from the platforms showed evidence of contamination. Thus, the existing sampling design is no longer suitable for assessment of the area contaminated. Effects on the fauna showed, as with contamination, 4 categories. Analyses hnking fauna and environmental variables indicated that the effects were mainly related to THC, barium and strontium, but also to metals like zinc, copper, cadmium and lead, which are all discharged in drill-cuttings. Effects on the fauna closely followed the patterns of contamination with only a few stations at each field that were contaminated not showing effects. Thus the areas showing effects were only slightly less than the areas contaminated. Subsequent to cessation of discharges biodegradation of oil and reduced concentrations of THC were observed. Yet there was an extension of areas where the fauna was affected several years after cessation of dnll-cutting discharges This may indicate that barite and related compounds associated with the discharges also have an environmental impact. However, preliminary results from fields using only water-based mud clearly indicate a reduction in environmental contamination and biological Impact, compared to effects reported here, for oil-based drill-cuttings. Diversity indices appl~ed to the data did not show the extent of effects and such Indices alone should not be used to interpret changes. The consistent patterns that the multlvariate techniques were able to detect showed that these methods...

Although analyses of macrobenthic infaunal communities traditionally require identification of the organisms to the species level, there is still only a limited number of studies which have examined the use of higher-level taxa in monitoring surveys. These generally indicate that identification of organisms to the lowest possible taxon may not always be necessary to enable description of spatial patterns in routine environmental and pollution monitoring programs. In other fields of marine ecology, such as studies of biodiversity, comparative investigations over large geographical areas, and the development of rapid-assessment techniques, the use of identification to higher taxonomic levels is often necessary. Detailed comparative studies of faunal patterns at various taxonomic levels are therefore important. This study uses multivariate analyses of macrofauna and environmental data from 20 separate investigations in the Skagerrak and North Sea to examine faunal patterns at different taxonomic levels. Data are analysed at the levels of species, genus, family, order, class and phylum. Transformation of data is also considered since degree of tlansforrnation is as important in determining the outcome of subsequent analyses as the taxonomic level to whlch organisms are Identified. Correlations between the underlying similarity matrices at the species level and higher taxonomic levels show highest values in polluted areas, lower values in less disturbed areas and lowest values in pnstine areas, indicating that the faunal patterns for the various taxonomlc levels become more similar as the degree of disturbance increases. For those areas where time-senes data are available it is shown that, as contamination increases, correlations between the faunal patterns at all taxonomlc levels, but especially the highest levels, and the environmental variables tend to increase, and after several years of contamination all the correlations are high, independent of taxonomic level. Again, the faunal patterns at the different taxonomic levels tend to become more s~milar as a result of increased pollution For all 20 investigations, highest correlations between faunal patterns (i.e. the underlying similarity matrices) and environmental variables were mainly found at the levels of species, genus and family, and often there was a distinct drop in correlation value between family and order In most cases, ~ndependent of the level of pollution, there is only a minor reduction in correlation between species and family, suggesting that identification to the level of family may be satisfactory in many routine monitonng surveys. The debate about the level of taxonomic resolution required for routine environmental monitoring becomes relatively unimportant if the effects of choosing different transformations are not also considered.

Although surveys of soft-bottom macrofauna are an important tool in marine pollution monitoring, the high costs involved have often been criticised. Species identification is time-consuming, and one solution is to identify organisms to a taxonomic level higher than species. This study, using data from a survey in the vicinity of the Valhall oil field in the Norwegian sector of the North Sea, examines the effects of using abundances of different taxonomic levels, and of using different data transformations (used to adjust the relative weightings of rare and abundant taxa) in subsequent multivariate analyses of faunal patterns, and relates the environmental variables to the observed faunal patterns from the different analyses. The study area has a constant water depth, homogeneous bottom sedlment and a uniform benthic community. At the time of the survey the platform had been active for 9 yr, and strong gradients in environmental contaminants and related faunal changes were found. Data from 27 stations around the oil platform were analysed at the levels of species, genus, family, order, class and phylum, using a range of data transformations and multivariate techniques. The data matrix contained 156 species grouped into 138 genera, 102 families, 4 2 orders, 18 classes and 10 phyla. Matrices derived from species, genus and family abundances constructed using the same transformation are very similar, and even at higher taxonomic levels the gradient of change in community structure is still detectable. As the taxonomic level increases the effects of transformations become stronger, so although both taxonomic resolution and transformation affect the results of analyses, the effects of each are different and, to a large extent, unrelated. The highest correlations between matrices derived from measured environmental variables and biotic matrices are between environmental variables related to drihng activity and mildly transformed family abundances, suggesting that analyses of higher taxonomic levels are more likely to reflect a contamination gradient than are analyses based on species abundances. Before any general recommendations regarding taxonomic levels are given for future macrobenthic surveys, there is an obvious need for studies of weak, intermediate and strong gradients in various types of contamination in homogeneous and heterogeneous environments, using the same sampling and analysis methods.

No abstract

Most generalisations about marine benthic diversity (species richness) are derived from few studies, few samples and low specles numbers. It is questionable whether the data on whlch most parad~gms, espec~ally about the deep sea, are based truly represent general patterns of marine diversity. Available information from deep-sea studies are summarised and compared with some extensive data sets from the shallow coasts of Norway and Australia. We show that species richness per unit area is as hlgh, if not higher in shallow sedimentary habitats a s was reported for the deep-sea data by Grassle & Maciolek (1992; Am Nat 139 313-341). Apparent high diversity in the deep sea may be explained, In part, by the vast area of this environment. All surveys in both the deep-sea and coastal habitats are shown to traverse a variety of microhabitats and thus sediment heterogeneity is not a n explanation for the high species richness in coastal environments.