Diatom enumeration method influences biological assessments of southeastern usa streams

Abstract

Current fixed-count enumeration methods for benthic diatoms are likely inadequate for most research and monitoring objectives. These methods underestimate taxa richness and may fail to detect losses of species caused by human impacts. Consequently, the full potential of diatoms is not realized in current assessments of biological integrity or species diversity. In this study, we hypothesize that alternative enumeration methods differ in their ability to quantify species composition. Furthermore, we hypothesize that an alternative to the traditional fixed-count method will improve both performance of observed/expected (O/E) indices derived from River Invertebrate Prediction and Classification System models and the discrimination of reference-quality and human-modified sites by other standard metrics used in biological assessments. To test these hypotheses, we assessed 1) how well 3 counting methods characterized species richness in a subset of 15 samples of stream benthic diatoms and 2) how counting method affected the performance of O/E indices and metrics by comparing the traditional fixed-count method against the best-performing alternative method. These latter comparisons were based on samples collected from 68 reference-quality streams and 20 streams located along an urban disturbance gradient. We demonstrate that traditional fixed counts failed to detect >½ of species present in most of the 68 reference-quality sites. Instead, timed-presence data produced the O/E index with the best performance and a level of precision similar to published invertebrate O/E indices. Furthermore, the O/E index based on the timed-presence data allowed us to determine which species are most often lost with urbanization. We found that traditional fixed-count and alternative timed-presence data produce metrics that are nearly equally able to discriminate between reference and disturbed sites. This study demonstrates that alternative counting methods improve species detection and require up to ∼30% less effort.