Evaluating the precision of kick-sampling in upland streams for assessments of long-term change: the effects of sampling effort, habitat and rarity fig: 5 tab: 5
Bradley, David C.; Ormerod, S. J.
Archiv für Hydrobiologie Volume 155 Number 2 (2002), p. 199 - 221
published: Dec 9, 2002
ArtNo. ESP141015502001, Price: 29.00 €
Kick-sampling is widely used as a standard method in qualitative studies of benthic macroinvertebrates, but its precision has been evaluated only for a limited range of sites and taxa. Upland streams are particularly poorly represented among the available data. With biomonitoring now expanding under the Water Framework Directive to encompass conservation evaluation, and also to assess the ecological status of a wider range of water bodies, fuller assessment is now timely. Here we appraise the efficiency of timed kick-sampling in riffles (2 minute) and margins (1 minute) of upland steams in mid-Wales for different abundance classes of taxa. From replicate samples drawn simultaneously from 5 streams, we assessed whether individual samples could reliably detect differences between sites, habitats, and years. We considered also the effect of rare taxa (taxa with < 4 individuals in total over five sample replicates). Surveys from 1990 and 2001 were compared to account for differences between years, seasons, and operators. On average, 69.4 % and 66.2% of taxa were captured in one combined riffle and margin sample replicate out of five in 1990 and 2001, respectively. Single margin samples generally contained a lower proportion of the taxa collected in total over five sample replicates (54.1 % and 61.6 %) than riffle samples (71.5 % and 61.8 %). These data illustrate the importance of sampling more than one stream habitat. Taxon accumulation required increased sampling effort for successively rarer abundance classes. As a result, the rarest taxa (< 4 individuals in total over five sample replicates) were a large source of sampling error in sample replicates. Extrapolation using a predictive function suggested that up to ten sample replicates would be required to adequately census rare taxa in riffles and margins if compiling taxon inventories. Sample replicates were, nevertheless, more similar in composition within habitats and streams than between them so that even individual samples differentiate sites and habitats. Composite samples from riffles and margins, and samples from riffles alone, adequately detected year-to-year faunal changes within sites, irrespective of whether rare taxa were deleted from analyses. We conclude that in upland streams typical of NW Europe, single kick-samples provide reliable and reproducible samples capable of detecting differences in assemblages between sites and years. These results are comparable to other assessments, providing further support for the widespread application of the kick-sampling method in biomonitoring studies of lotic macroinvertebrates. However, conservation evaluation and compilation of taxon inventories will require more exhaustive sampling within and across habitats. We suggest that those developing sampling programmes to satisfy the Water Framework Directive should look carefully at the effects of sampling effort on any measure where rare taxa are involved.