During the last couple of decades, an increasing number of studies use sequence clusters as units for taxonomic diversity. It is well known that such molecular operational taxonomic units (MOTUs) do not necessarily correspond to species, but they are treated as such when measuring diversity and testing theories. Here, I show that data from studies of molecular evolution and species diversification of fungi indicate that commonly used cut-offs are likely to lump species in many cases. At the same time, empirical studies show that the mean within-species variation is close to these cut-offs. That the within-species variation estimates are plausible is supported by coalescence modelling under a range of parameter settings. In addition, studies using crossing tests to delimit species show that there often is an overlap in within- and between-species distances. The available data therefore indicate that sequence clusters are likely to misrepresent species. However, to keep a biological relevance, MOTUs should be kept in close agreement with species. Studies using them should therefore asses how sensitive the results are to differences between MOTUs and species--something that is rarely done. An even better solution is to directly include the uncertainty in species delimitation in the analyses, but in many cases, we need to increase our knowledge of taxonomy and evolution to do this accurately. Even if the empirical data referred to here pertain to the "barcoding" region of rDNA in fungi, there is nothing indicating that the situation is substantially better for other taxa or genes.
Keywords: barcoding; community ecology; fungi; molecular operational taxonomic units; species delimitation.
© 2015 John Wiley & Sons Ltd.