Environmental characteristics in oligotrophic waters: Data evaluation and statistical limitations in water quality studies

Abstract
Standard statistical tests of nutrient variability in the euphotic layer of an oligotrophic system in the S.E. Aegean Sea were performed. Practical problems resulting from data handling, such as high errors associated with low concentrations, nonlinearity and interaction among variables were examined. The practical problems in the analysis of environmental data, arising from statistical limitations were considered: linear correlations between chlorophyll α and nutrients were found to be significant only for ammonia, and the multiple regression model was found to be of limited value as a tool for a description of the system. By contrast, principal component analysis contributed significantly to the interpretation of each variable. The difficulties arising from the use of statistical methodology, and particularly the limited goodness-of-fit of the regression models, are discussed in relation to the suitability of oligotrophic waters as ‘control sites’ in eutrophication studies.