Reduced count size can provide a robust and more efficient diatom assessment of environmental conditions

Abstract

Abstract Bioassessment indices routinely used in freshwater assessments are calculated using enumeration methods that are often time‐intensive. For diatom assessments that are mainly based on time‐consuming valve counts, little attention has been devoted to optimizing current counting methods, despite the need for cost‐effective bioassessments and biomonitoring. With a focus on count size sufficiency, we examined diatom samples from >3000 rivers and streams across the United States to assess the effects of reducing the standard count size of 600 diatom valves on the recently published, genus‐level, trait‐based multimetric diatom indices (MMIs) and its component metrics. The diatom data were collected by the United States Environmental Protection Agency's National Rivers and Streams Assessment (NRSA). The MMIs, based on the NRSA data, were designed to assess river and stream ecological condition for three large ecoregions across the United States—East, Plains and West. We found that reducing the standard 600 valve counts introduced variability in component metric values, especially for % taxa metrics, but the variability had little effect on the performance of the actual MMI. More than 90% of all subsamples from the Plains and East ecoregions, and over 80% of West subsamples, were placed in the same condition class regardless of whether a 600‐ or 100‐count size was used. Moreover, the ability of each component metric and MMI to discriminate least‐ from most‐disturbed sites remained the same at 100 valves as that obtained using 600 valves. Synthesis and applications. Contrary to the general perception that counting a high number of diatom valves will lead to a more responsive and robust index, our research shows that a good index performance can still be obtained by counting considerably fewer diatom valves. A smaller count sum will not just reduce the analysis effort but also the quality assurance/quality control effort. For large‐extent bioassessment programmes, reducing the count size could lead to substantial savings in resources that could potentially allow the assessment of more sites. Our subsampling approach to determine an optimal count size is an effective way to achieve time‐ and cost‐effective sample processing that can be applied to any group of organisms.