Abstract :
A meta-analysis of more than 150 model stream ecosystem studies employed in hazard assessment was conducted to assess the effect of model ecosystem size on biological complexity and experimental design. Test system length was highly correlated to duration of colonization and chemical exposure period; however, size was generally unrelated to taxa richness, diversity, abundance, or biomass for algal, protozoan, and invertebrate assemblages. Fish were usually caged and not free-roaming, even in the relatively large (>10 m) systems. Experimental designs, especially in recent years, have been less concerned with pseudoreplication, resulting in studies emphasizing numerous treatments with few or no replicates. Test systems have evolved that emphasize flow through exposures, use of natural source waters, and regression-style experimental designs. These factors collectively reduce problems historically associated with pseudoreplication. Larger model ecosystems (>6 m) are advantageous for increasing the ability to sample more intensively through time and space without imposing sampling artifacts. Fish (caged and free-roaming) are also more often used in larger systems. Several model ecosystem studies that were compared with natural field-based communities demonstrated a high degree of correspondence for biological complexity attributes. Well-designed model stream ecosystems have a high degree of biological complexity and reasonable levels of variability for ecologically relevant endpoints. The use of small application factors (≈1) is applicable to ecotoxicological results from these studies for risk assessment.
