Recognising subtidal foraminiferal assemblages: implications for quantitative sea‐level reconstructions using a foraminifera‐based transfer function

Microfossil‐based transfer function models are increasingly used to provide decimeter‐scale sea‐level change reconstructions. In this paper I demonstrate that in the tropical location of northern Australia problems arise in selecting the appropriate elevation range for the modern training set used to calibrate fossil calcareous foraminiferal assemblages. Most calcareous foraminiferal species found in cores occur in both modern intertidal and shallow subtidal environments. A lack of independent measures to indicate whether fossil assemblages come from intertidal environments forces use of a training set that includes intertidal and subtidal environments. This results in decreased precision compared to using a training set solely from intertidal environments. The widely used method of assessing model fit to fossil assemblages (modern analogue technique) often fails to discriminate between acceptable and unacceptable reconstructions. It is important to investigate a number of different measures including modern analogue technique, canonical correspondence analysis and changing bootstrapped sample specific transfer function errors to fully understand the level of similarity between modern and fossil foraminiferal samples, to judge the reliability of transfer function‐predicted sea‐level reconstructions. Copyright © 2008 John Wiley & Sons, Ltd.