Effect of fixed-fraction subsampling on macroinvertebrate bioassessment of rivers.

We investigated the effect of different subsample fractions on the variability of benthic invertebrate metrics. The results of six fractions 1/12, 1/6, 1/4, 1/3, 5/12 and 1/2 were compared to the results of the whole samples. Over 120 metrics were tested using five datasets: ecoregion Alps and four river types. In general, variability of metrics decreased with increasing subsample size, but variability varied greatly with the selected metric group and river type. Independent of river type, the highest variation was observed for the composition/abundance group metrics and the richness metrics, whereas it was low for the diversity indices and for the metrics of the sensitivity/tolerance group and intermediate for the functional metric group. For all metric groups independent of river type, the main decrease in variability occurs up to 1/4 subsample. We suggest that the effect of subsample size on variability of metrics should be tested prior to selecting potential assessment metrics.