Baltic hard bottom mesocosms unplugged: replicability, repeatability and ecological realism examined by non-parametric multivariate techniques

The general utility of large-scale artificial ecosystems for ecological and ecotoxicological research is evaluated by a case study of the replicability, repeatability and ecological realism of a Baltic Sea hard bottom littoral mesocosm (called BHB-mesocosm). The structure (species abundance and biomass listings) of the macrofauna community associated with bladder-wrack, Fucus vesiculosus L., in nine control mesocosms run during 4 separate years, is investigated by descriptive and analytical multivariate statistical techniques. Non-metric multidimensional scaling (NMDS) is used to visualise differences in community structure between mesocosm controls run during the same year (replicability), differences among mesocosms run during different years (repeatability) and differences between mesocosms and the field mother system (ecological realism). Analytically the community structure of parallel mesocosms is shown to differ significantly by use of one-way analysis of similarities (ANOSIM) tests, which demonstrates a poor replicability. Despite this high degree of variability between parallel mesocosms, the null hypothesis of no differences among years can in turn be rejected by a two-way nested ANOSIM for both abundance and biomass data, which is evidence of a poor repeatability. If it is assumed that the field samples taken in connection with the transplantation are representative of the initial situation in the mesocosms, the divergence of the mesocosms from the mother systems was considerable in all years. When all 4 years are used as replicates, i.e. independent observations on dissimilarities between the mother system and the mesocosms after 5 months of enclosure, in a two-way crossed ANOSIM, significant differences for both macrofauna abundance and biomass data are revealed, which is evidence of a poor ecological realism. The similarity percentage breakdown procedure (SIMPER) establishes the species principally responsible for these differences. Considering abundance data Theodoxus fluviatilis is always an important discriminator between groups, mostly accompanied by Mytilus edulis and either Idotea spp. (replicability) or Gammarus spp. (realism). For biomass data Mytilus and Idotea are the most important overall discriminators, followed by Lymnaea spp. (repeatability and realism) and Cerastoderma glaucum (realism). Finally the factors and processes restricting mesocosm performance are outlined and their consequences are briefly discussed. It is concluded that the degrees of replicability, repeatability and ecological realism are too low for straightforward use of these and probably most other mesocosms in predictive risk assessment or in extrapolation of results to natural ecosystems. (C) 1999 Elsevier Science B.V. All rights reserved.