Sampling variation in macroinvertebrate data and implications for river quality indices

Abstract

SUMMARY 1. The EC Water Framework Directive (2000/60/EC) recognises the need for biological monitoring. Indices derived from standard samples of macroinvertebrates are frequently used for the appraisal of the ecological quality of rivers. However, information on the errors or chance variation that can influence the value of an index is also important.2. This paper describes a study to quantify the observed sampling variation in three ecological indices based on the Biological Monitoring Working Party (BMWP) score system across a wide range of river types and qualities. The indices are number of BMWP taxa, BMWP score and Average Score Per Taxon (ASPT).3. The study sites were selected to encompass the four major groups within the River InVertebrate Prediction And Classification System (RIVPACS) site classification for Britain. Within each group, four sites which differed in ecological quality grade were chosen (total of 16 sites). At each site three standard RIVPACS samples were taken in each of spring, summer and autumn by trained staff. In each season, two samples were taken by one biologist and the third by a different individual to allow for within and between‐operator variation.4. The effects of sampling variation within a season on the number of taxa, BMWP score and ASPT across all sites, irrespective of operator, could be represented by some simple parameters. We found that the sampling SD of the square root of the number of taxa, square root of BMWP score and the untransformed ASPT were roughly constant in each case, irrespective of site type or quality. For each index, SD for two and three seasons combined samples were smaller than for single season samples.5. Inter‐operator influences on sample values were negligible (4–12% of total sampling SD) in this study. This underlines the importance of adequate training for all staff involved in extensive monitoring programmes which use standard procedures from one year to the next, but may involve different staff.6. Indices for number of taxa, BMWP score and ASPT were all estimated with greater precision from combined season samples than from the averages of two or three seasons' samples.7. This study enables us to estimate confidence intervals for the values of the number of taxa, BMWP score and ASPT based on single season, two or three season combined samples collected using standard RIVPACS procedures for any river site in Britain. The results can also be used in simulation models which incorporate the effects of sampling variation into assessments of the ecological quality of river sites based on the ratio of observed to RIVPACS expected values of these BMWP indices.