Publication details

Mollusc assemblages in palaeoecological reconstructions: an investigation of their predictive power using transfer function models



Year of publication 2011
Type Article in Periodical
Magazine / Source Boreas
MU Faculty or unit

Faculty of Science

Field Ecology
Keywords Molluscs; spring fens; palaeo-reconstructions; transfer functions; environmental factors; predictions
Description Fossil mollusc assemblages are commonly used to reconstruct past environments, as their shells are abundant in various types of calcium-rich deposits. However, it is impossible to exactly evaluate estimates derived from fossil data using directly measured factors. To assess the accuracy of environmental variables derived from mollusc species composition, two modern data sets (training and test), each consisting of 73 samples of treeless fen communities, were constructed along with known local and climatic variables. The main predictors of species composition were isolated using canonical correspondence analysis and forward selection with the Monte Carlo permutation test. The accuracy of prediction for those factors that were significant in the forward selection was studied via four commonly used transfer function models. Three independent gradients of species composition driven by calcium content, moisture and temperature were detected. The best predictions were found for variables that correlated with the main changes in species composition. The strongest correlation between the predicted and measured values of the test data set was observed for water conductivity (r=0.86), a good proxy of calcium content. The locally weighted–weighted averaging transfer function model performed best out of all the models for the majority of variables tested. Mollusc assemblages were found to be useful for estimating local environmental variables based on a given species composition. Along with the specific advantages of mollusc fossil material, there is much potential for the use of their fossil assemblages to reconstruct palaeoenvironmental variables using transfer function models calibrated from recent compositional data and directly measured factors.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.

More info