Problems in the Variance Analysis of Nine Environmental Monitoring Variables: Determining the Number of Samples Needed to Detect a Change in Mean of 50

摘要

An analysis is conducted for nine environmental monitoring variables; the concentration of cadmium and PCB's in marine sediments, the incidence of fin erosion in dover sole and winter flounder, the incidence of black gill disease in rock crabs, the incidence of skeletal anomalies in Ammodytes, the incidence of genetic abnormalities in Atlantic mackerel embryos, a fecundity index for striped bass, an index of nesting success in osprey and brown pelicans and a mortality index for fish and shellfish larvae. The variables are classified as physical or biological monitoring variables and the number of samples expected to be required to detect a change in mean of 50% is estimated for each monitoring variable. The direction of change is that expected under the assumption of decreasing environmental quality. Problems in handling and acquiring encountered data are discussed and methods for calculating the appropriate variance estimate and determining the appropriate data transformation are presented. Our results show that fecundity in striped bass requires the least number of samples to detect a change in mean of 50%. We conclude that biological variables seem to have great potential as environmental monitoring variables, provided the relationship between environmental quality and the biological monitoring variable is understood.