A manipulative experiment to evaluate predicted changes in intertidal, macro-faunal assemblages after contamination by heavy metals

Concentrations of heavy metals (zinc, copper and lead) were manipulated experimentally to test the hypotheses about effects on intertidal, soft-sediment assemblages of animals in two sand-flats in Port Hacking, Australia. Hypotheses about changes in the structure and composition of whole assemblages and changes in mean abundance and variability of individual taxa were tested. Specific hypotheses were derived a priori from repeated observations of assemblages in urban and non-urban areas of Port Hacking. After manipulation, concentrations of metals were similar to those in sediments near urban areas. Nevertheless, responses of assemblages to the increased concentrations of metals were weak. Polychaetes, spionids and amphipods responded to experimental treatments. Changes were, however, not consistent among times and places and generally not in agreement with what had been predicted. Significant spatial and temporal variability were detected for all variables investigated. Increased concentrations of metals did not affect variability or overall structure of assemblages. Thus, there was little evidence that increased concentrations of metals caused benthic assemblages in pristine areas to become more similar to those in areas contaminated by human activities. Several potential explanations for the discrepancy between previously observed correlative patterns and the results presented here are discussed. These include a critical assessment of different aspects of the experimental study, such as lack of statistical power, insufficient basis for prediction, artefacts due to experimental procedures and issues to do with the difference between evidence based on correlation and manipulation. Explicit comparisons showed that there were significant effects of physical disturbance due to repeated sampling and that assemblages of animals at the start of the experiment were different from those previously observed in uncontaminated areas. These observations were surprising and made interpretation more difficult. Nevertheless, it is possible that the metals investigated really contribute only marginally to previously observed differences between urban and non-urban areas. Repeated comparisons between observations of patterns and manipulative experiments like these can only improve the basis for prediction and the power of current mechanistic models.