Within-stream variability of benthic invertebrate samples and EU Water Framework Directive assessment results

The EU Water Framework Directive (WFD) demands for an integrated biological assessment of surface water bodies. The assessment should use biological Quality Elements as indicators and be based on comparing the observed biota with stream type specific reference condition biota to reflect the impact of various stressors. Moreover, the WFD demands that restoration measures must be taken whenever the Ecological Quality Class (EQC) is Moderate or worse. Therefore, it is important to know the reliability of assessment results in order to judge whether an expensive restoration measure is justified or not. The degree of uncertainty is linked with the variability of the data. The aim of the present study was to investigate the variability of benthic invertebrate samples of running waters. In 26 mid-sized streams in Germany, two samples were taken following a slightly modified STAR/AQEM protocol (N(Total) = 52). The STAR/AQEM protocol (Furse et al. 2006. STAE consortium 2003) is the result of two successive EU-funded projects, which aimed to develop a standardised macroinvertebrate sampling protocol and assessment system for running waters (Furse et al. 2006. Hering et al. 2003). STAR/AQEM multi-habitat sampling is based on 20 sampling units, which are distributed proportional to the estimated microhabitat covers at a sampling site. The two samples of each stream were taken at sites which were relatively close to one another (100 to 650m) and had similar structural parameters, hydromorphology and physical/chemical parameters. Thus, it was expected that biocoenoses at both sites and assessment results should be similar. However, the results showed that samples differed in terms of number of taxa and abundances. These differences caused deviations in metric results, including the Multimetric Index (MMI) and its constituent core metrics, such as the percentage of EPT-Taxa (EPT [%]) or number of EPTCBO-Taxa (EPTCBO). In 10 of 26 causes the differences in metric assessment results led to different ecological quality classes (EQCS). In order to quantify uncertainty in assessment results, the STARBUGS simulation software was used. By simulating metric scores based on the observed sampling variance in the present data set, this software calculates the confidence probability of a sample belonging to each of the five possible EQCs. For the MMI the average confidence probability that a stream site truly belonged to EQC estimated from the observed sample was 89%. This means that the probability that a stream site belongs to a different EQC than the assessment result indicates was quite low (approximately 11%). However, it is discussed that the given confidence probabilities can only be used to estimate the precision of the analysed samples - the results cannot be used for estimating the probability of future samples delivering the same assessment result. Moreover, the present study revealed that certain metrics are more susceptible to variability (EPT [%]m EPTCBO) than others (e.g. German Fauna Index GFI). The reasons for variability and consequently uncertainty inherent to benthic invertebrate samples are discussed. The results of the present study stress the importance of implementing standardised sampling and sorting protocols and quality control mechanisms in benthic invertebrate assessment schemes.