Abstract The analysis of fossil biotic remains in lake sediment records allows for the reconstruction of past ecosystem dynamics and as such forms a powerful tool for understanding ecosystem processes and environmental change on a range of spatial and temporal scales. However, when producing palaeoecological datasets, analysts are often not able to assign all specimens to morphotypes at the highest taxonomic resolution. As a result, datasets containing unprocessed or raw counts usually include categories of identification across multiple different taxonomic levels (e.g. species morphotype, genus, tribe or even family level), also for fossil remains that may actually have originated from the same taxon (e.g. the same species or species morphotype). Whilst different strategies to deal with this problem have emerged over the years, it is rarely described in papers how analysts dealt with this issue, and how datasets were processed from raw counts to a final dataset used for palaeoecological interpretation, numerical analysis or quantitative inference. Using chironomid identifications as an example, we here describe the four main strategies for dealing with such multi-level identifications that can be applied to unprocessed count data with different levels of taxonomic detail: combining, retaining, deleting and assigning. We discuss advantages and disadvantages of each of these strategies and illustrate their impacts on palaeoecological analyses using both theoretical and practical examples. We conclude that there is no one optimal way to deal with the issue of incomplete or multi-level taxonomic identifications of fossils that may have originated from the same taxon, but that analysts will have to determine the strategy that best befits their project on a case-by-case basis, taking into account the type and quality of data they are working with, as well as the overall aims of their research project. We recommend that the choice of strategy (or combined strategies) is clearly described in the form of a brief statement in the Methods section of manuscripts where primary data are presented and that ideally both the unprocessed raw count data and the processed data used for ecological and numerical analyses are archived where publication outlets allow for this. Whilst we use chironomid research as the primary example throughout this manuscript, the proposed strategies and recommendations are relevant to a wide range of microfossil groups.
Heiri et al. (Mon,) studied this question.