The perils of taxonomic inconsistency in quantitative palaeoecology: Experiments with testate amoeba data

Abstract

A fundamental requirement of quantitative palaeoecology is consistent taxonomy between a modern training set and palaeoecological data. In this study we assess the possible consequences of violation of this requirement by simulating taxonomic errors in testate amoeba data. Combinations of easily confused taxa were selected, and data manipulated to reflect confusion of these taxa; transfer functions based on unmodified data were then applied to these modified data sets. Initially these experiments were carried out one error at a time using four modern training sets; subsequently, multiple errors were separately simulated both in four modern training sets and in four palaeoecological data sets. Some plausible taxonomic confusions caused major biases in reconstructed values. In the case of two palaeoecological data sets, a single consistent taxonomic error was capable of changing the pattern of environmental reconstruction beyond all recognition, totally removing any real palaeoenvironmental signal. The issue of taxonomic consistency is one that many researchers would rather ignore; our results show that the consequences of this may ultimately be severe.