Sieve mesh size and quantitative chironomid paleoclimatology

The minimum sieve mesh size for fossil chironomid analysis is usually set at 100 μm, to ensure adequate recovery of small species. Yet taking into account the labor intensity of sorting and identifying fossil chironomid remains, the large numbers of samples requiring processing in paleoclimate studies with high temporal resolution, and the increasing need to engage non-specialist analysts in this work, it seems appealing to sieve samples through a larger mesh size that would mainly retain easier-to-see, easier-to-extract and easier-to-identify fossil specimens. In this study we evaluated the influence of sieve mesh size on chironomid-based quantitative paleoenvironmental reconstructions in African lakes. We developed two chironomid-salinity calibration data sets based on either the >100 μm or >150 μm size fractions of surface-sediment fossil assemblages, and compared the performance of salinity-inference models derived from them. We find that, despite ∼35% additional fossil loss, restriction to >150 μm data did not appreciably affect individual taxon optima or tolerances, and resulted in only a modest reduction of mean taxon richness per lake. Parameters of statistical model performance were as good, or better than those of models based on the >100 μm data, albeit only after excluding two atypically dilute (<100 μS/cm) lakes from the calibration. Application of inference models based on >150 μm or >100 μm data to a 200-year fossil record from Lake Abiyata, Ethiopia, produced very similar trends, amplitudes and uncertainty ranges of inferred past salinity change. Restriction to >150 μm data reduced the mean fossil yield of core samples by ∼20% on average, i.e. fossil loss was markedly lower than in surface-sediment samples. Using the larger mesh size reduced sample processing time by up to 50%, partly by removing a significant proportion of visually obstructive organic debris, and partly by allowing a greater fraction of chironomid fossils to be identified directly in the sorting tray. The fraction of 1st instar group taxa in surface-sediment samples was reduced from 13% to 3%, increasing the mean taxonomic resolution of fossil assemblages, and thus their ecological specificity.