Parataxonomy vs. taxonomy in biodiversity studies - pitfalls and applicability of 'morphospecies' sorting

Parataxonomic sorting of samples to recognizable taxonomic units (RTUs, morphospecies, morphotypes or, as proposed here: parataxonomic units [PUs]) is generally considered to be a sufficiently reliable and conservative approach in ecological biodiversity studies or conservation biology. It is obviously time-saving because it avoids the burdens of taxonomy. However, evaluations of parataxonomic sorting by taxonomic resorting show many overestimations of species numbers. Hence, RTU sorting is not necessarily conservative. Sorting errors can be more than 100% (median in the present compilation: 22%). Even if the cumulative results for diverse groups like beetles have a very low overall error, the error rate in the single families is generally much higher. This pattern is likely to cause severe problems in multivariate analyses. The presumable error rate in sorting does not depend only on the group to be sorted, but also on the sorter and the sample. Therefore, the sorting error is not predictable. Since PUs are generally neither described nor assigned to existing names, the sorting results are difficult to check and it is mostly not revealed why the samples are sorted as they are. Since parataxonomy does not use existing biological knowledge, creates typological units and does not disclose its sorting criteria, inter-subjective testability and falsifiability of the sorting results are more difficult than of taxonomic identifications (or are even impossible). Parataxonomy does not fulfil the criteria of a scientific method, but is propedeutic and can be a heuristically valuable tool to find out patterns in taxonomically neglected groups. However, it is only the first step in sorting and identifying samples in biodiversity studies. PUs are useless for inventories and area selection in conservation evaluation, biogeographical and autecological studies; they provide only uncertain data for studies in species turnover and overlap, but they can be used quite reliably for global comparisons of gross species richness, non-comparative descriptions of species richness of single sites or for comparisons of sites without species overlap. If results of parataxonomic sorting show clear and biologically meaningful patterns, the sorting is likely to be reliable. Weak or no detectable patterns may easily be caused by erroneous sorting.