Original paper

An update on the quantification of stream drift

Matthaei, Christoph D.; Werthmüller, Daniel; Frutiger, Andreas

Archiv für Hydrobiologie Volume 143 Number 1 (1998), p. 1 - 19

138 references

published: Sep 22, 1998

DOI: 10.1127/archiv-hydrobiol/143/1998/1

BibTeX file

ArtNo. ESP141014301001, Price: 29.00 €

Download preview PDF Buy as PDF


The reliable quantification of stream drift is of vital importance for all drift studies. Although variation between drift samples is smaller than between benthic samples, some six to seven replicates have been considered necessary by ALLAN & RUSSEK (1985) to obtain a precision of 95 %-confidence limits within 50 % of the mean. While a review of drift studies published during the last twelve years showed that many researchers have probably used too few replicates to allow a reliable quantification of stream drift, it may be often unnecessary to use as many as six or seven. In the present study conducted in a Swiss river, we analysed data from 24 drift samples taken on six different occasions during three studies conducted in winter, spring, and summer. In each season, drift was sampled twice. The 95%-confidence limits obtained with just four replicates were within the 50 %-boundary on all six sampling occasions for the total number of invertebrates and the five relatively most common taxa in spring and summer. In the two winter collections, taxon-specific confidence limits were outside 50 % of the mean in five cases out of ten. However, overall drift densities in winter were low, and all but one of the taxa with high variation between replicates were quite rare in actual numbers. A similar or even better statistical precision may be possible in other rivers, because the higher number of replicates recommended by ALLAN & RUSSEK (1985) may have been partly due to collecting some of their replicate samples during periods of low overall drift. When studying the drift of rare taxa, however, actually more than six to seven replicates may be required to generate reasonable estimates of their drift densities. We suggest that researchers conduct more pilot studies in their individual study systems to determine the appropriate number of replicates needed, since this has been rarely done in the past.


quantificationstream drift