ABSTRACT
Over the past five years many field aquatic mesocosm studies have been conducted or are under way to meet regulatory requirements of the U.S. Environmental Protection Agency (EPA). Mesocosm studies as an ecosystem-level test are still developing, and, consequently, considerable variations are evident in study design and data analyses. In addition, experimental design often does not properly address crucial points at the study outset, e.g., defining estimated entry rates (EER) for spray drift and/or runoff into aquatic environments and what are the ecologically important effects. Study endpoints need to be carefully selected relative to the test chemical, to avoid pointless and tedious measurements of a generic list of parameters that are not pertinent to the study objectives. This paper delineates criteria to be considered in mesocosm study design and then discusses the uses of one-way analysis of variance (ANOVA), where specific hypotheses are tested, and non-hypothesis testing regression methods. Conventionally in hypothesis testing, data have been tested against the null hypothesis of “no difference between treatment and control means” using Student’s t-test. A new method has been proposed by the EPA that reverses this approach and tests a null hypothesis of “a difference between treatment and control means which is greater than a specific amount”. These two methods of statistical analyses are compared.
