Stream and river typologies - major results and conclusions from the STAR project

    loading  Checking for direct PDF access through Ovid

Abstract

The EU Water Framework Directive uses abiotic variables for classifying streams and rivers into types. For rivers, the EU Water Framework Directive fixed typology i.e. ‘System A' typology are defined by ecoregions, size based on the catchment area, catchment geology and altitude. Within any given part of the WFD typology, it is assumed that biological communities at undisturbed sites will be broadly similar and will therefore constitute a type-specific biological target and a way to stratify the spatial variability in stream and river monitoring and assessment. The data collected for the STAR project cover 13 countries and include 22 stream types. A total of 233 sites were fully sampled for all biological quality elements (fish, macrophytes, benthic macroinvertebrates, and diatoms) in the study. Analysing the STAR macroinvertebrate dataset in relation to environmental and biogeographical variables resulted in three major groups of stream types that correspond to three major landscape types in Europe: Mountains, Lowlands and Mediterranean. Similar results were found when analysing all four biological quality elements (fish, macrophytes, benthic macroinvertebrates, and diatoms) sampled in the STAR project. The studies also showed that the stream types using the WFD ‘System A' descriptors are probably less useful at finer scales and it is suggested that a stream typology should take three main parameters as a starting point, i.e., climate (temperature), slope (current velocity) and stream size. Existing site-specific multivariate RIVPACS-type predictive models were also compared to both null models and the WFD ‘System A' physical typology as methods of predicting macroinvertebrate reference conditions. It was concluded that the multivariate models are more effective in predicting reference conditions primarily because they make use of continuous rather than categorical predictor variables and because the multivariate RIVPACS-type models are not constrained by the use of a limited number of variables.

Related Topics

    loading  Loading Related Articles