Use of similarity measures for quality control of benthic diatom samples

A method for evaluating the similarity of replicate benthic diatom samples based on the Bray–Curtis similarity measure is described. This technique may be useful as part of quality control procedures where objective performance measures are required. Levels of similarity >60% typically indicate good agreement between the primary analyst and auditor. However, an evaluation of 57 comparisons indicated that achievable levels of similarity were dependent upon the species diversity of the sample, with samples with high species diversity typically having lower levels of similarity than samples with low species diversity. Whilst a threshold value of 60% is adequate for most samples, a stiffer threshold of 70% should be applied to samples with very low levels of diversity.