No magic number: determining cost-effective sample size and enumeration effort for diatom-based environmental assessment analyses

Abstract

Aquatic microorganisms are commonly used as indicators in environmental assessments. Two key aspects of sampling design for these studies are the number of sites sampled and the enumeration effort of organisms within samples. However, there has been no rigorous determination of a cost-effective balance between them. Here, we use regional diatom data sets from 207 to 493 lakes to determine the influence of sample size (lake number) and enumeration effort (valve count) on the accuracy and cost-effectiveness of commonly used environmental assessment analyses. We find that both lake number and valve count can be considerably reduced from the original data sets. However, there is considerable variation among analyses, such that a single most cost-effective sampling configuration cannot be recommended. For weighted-averaging assemblage versus environment models, we recommend 70–100 lakes and 200 valves for retaining high accuracy. For other analyses, we recommend iteratively building up data sets, using initially high (300 or more) and consistent valve counts and using subsampling to determine when the influence of either lake number or valve count plateaus. This technique is applicable to other types of microorganisms used in environmental assessments.