Chemometric discrimination between streams based on chemical, limnological and biological data taken from freshwater fishes and their interrelationships

The VALIMAR project aims at identifyingbiomarkers in fish that are suitable to detectand predict environmental stress from chemicalpollution or from limnological parameters inthe field. For two small streams in SouthernGermany, concentration values of 31contaminants in water and sediment and 12 limnological parameters as well as 27 biomarkersmeasured in brown trout and stone loach werecollected. All these physicochemical andbiological parameters have been analysed forpatterns that discriminate between the streams,using discriminant analysis (DA), analysis ofvariance (ANOVA) and of covariance (ANCOVA), and principal component analysis (PCA) asmultivariate statistical techniques. Moreover,the biological data were analyzed with respectto species-specific patterns, and the partialleast-squares regression method (PLS) was usedto study the impact of chemical and limnological data on the health status of the targetspecies as characterized by the biomarker data.Abiotic as well as biotic data yielded goodseparations between the streams, with theultrastructure of gill (US-gill) being thestrongest discriminator variable among all 27biomarkers tested. With regard to the two fishspecies, the biomarker data from brown troutshow significantly greater differences betweenthe two streams than the biological responsesin stone loach. Application of PLS yieldssignificant regression models for only fewbiomarkers including US-Gill, which can bepartly traced back to significant noise levelsin the data set as quantified by permutationtests.