Development of a new pan-European testate amoeba transfer function for reconstructing peatland palaeohydrology

In the decade since the first pan-European testate amoeba-based transfer function for peatland palaeohydrological reconstruction was published, a vast amount of additional data collection has been undertaken by the research community. Here, we expand the pan-European dataset from 128 to 1799 samples, spanning 35 degrees of latitude and 55 degrees of longitude. After the development of a new taxonomic scheme to permit compilation of data from a wide range of contributors and the removal of samples with high pH values, we developed ecological transfer functions using a range of model types and a dataset of similar to 1300 samples. We rigorously tested the efficacy of these models using both statistical validation and independent test sets with associated instrumental data. Model performance measured by statistical indicators was comparable to other published models. Comparison to test sets showed that taxonomic resolution did not impair model performance and that the new pan-European model can therefore be used as an effective tool for palaeohydrological reconstruction. Our results question the efficacy of relying on statistical validation of transfer functions alone and support a multi-faceted approach to the assessment of new models. We substantiated recent advice that model outputs should be standardised and presented as residual values in order to focus interpretation on secure directional shifts, avoiding potentially inaccurate conclusions relating to specific water-table depths. The extent and diversity of the dataset highlighted that, at the taxonomic resolution applied, a majority of taxa had broad geographic distributions, though some morphotypes appeared to have restricted ranges. (C) 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).