Effects of taxonomic and numeric resolution on the ability to detect ecological patterns at a local scale using stream macroinvertebrates

With 3 figures and 3 tables Abstract: The increasing demand for methods of rapid stream bioassessment has stim- ulated the evaluation of data simplification. In particular, these studies have assessed how much power is lost when species/morphospecies identification is replaced by family identifications or use of EPT (Ephemeroptera, Plecoptera and Trichoptera) taxa only. A second simplifying factor commonly evaluated is the use of presence/absence data instead of density. These simplifications have provided valid results in most cases where differences among groups are large, particularly in studies comparing impacted vs. non-impacted stream sites and ecological studies involving large spatial scales. Here I evaluate whether data simplification, both in terms of taxonomic (families, mor- phospecies of EPT) and numeric (presence/absence) resolutions, is valid for ecological studies done at local scales, where differences among groups are subtle. Datasets used are derived from a five-year study of five stream sites situated in a catchment in south- east Brazil. Streams were sampled twice a year, in the rainy (summer) and dry (winter) seasons. I used Analysis of Similarity (ANOSIM) to evaluate if differences i) among stream sites and ii) between seasons within a stream site, revealed by using the full data set (morphospecies, quantitative data), were also detected when using the sim- plified datasets. The effect of taxonomic resolution was not significant; the two sim- plified levels of this factor (morphospecies of EPT, families) were able to recover the same groups revealed by the full dataset. However, the use of presence/absence data had a strong negative effect on the ability to distinguish groups, particularly when dif- ferences were small (between seasons within a stream site). The success in recovering groups using simplified taxonomic data agrees with previous evaluations done using datasets from applied fields and those from ecological studies involving large spatial scales. However, in contrast to results observed in applied and large-scale studies, use of simplified data quantification in local datasets resulted in significant loss of infor-