Measures of central tendency of aquatic habitat parameters: application to river macrophyte communities

The aim of this paper is to assess usefulness of different measures when determining central tendencies which characterise the environmental requirements of living organisms. Mathematical analyses were made of the environmental parameters of river macrophyte communities which were taken as a representative pattern of different levels of biological structures. To deliver a representative dataset, botanical surveys were carried out on a range of British rivers together with environmental assessment and the plant communities groups were identified based on characteristic species according to eslished phytosociological criteria. The mean values and standard univariate medians of the revealed associations were compared with means calculated on the basis of advanced transformation and also with the rarely calculated multivariate L median. Due to high variance and asymmetrical distribution, the analyses based on the mean-value appeared to be limited in application. To avoid this disadvantage transformation to obtain normality of the dataset standardisation was proposed although even this did not fully reach a satisfactory symmetry. It was concluded that each environmental variable for each single biota must be individually treated by a suile transformation to obtain approximately normal distributions. The univariate median was very resistant to the effects of outliers but gave a flattened output of the environmental dataset making the partitioning of biological units very difficult. The multivariate L median appeared to be unaffected by outliers. It enabled to obtain considerable ordering of communities against individual environmental parameters without data transformation.