Abstract
CLAMP (Climate Leaf Analysis Multivariate Program) has been used for the past 17 years to estimate palaeoclimatic conditions. The reliability and applicability of this method, based on leaf physiognomic characters of fossil woody dicots, has been widely discussed over the same period.
The present study focuses on some technical aspects of CLAMP, mainly on its robustness in the context of the theoretical unimodal requirements of Canonical Correspondence Analysis, and introduces “correction coefficients” for these aspects of the statistical approach as a new way of interpreting and improving on CLAMP estimates. This tool was tested on datasets derived from 17 European fossil floras ranging in age from the Late Oligocene to the Pliocene.
Additionally, an objective statistical method for the selection of the best-suited modern vegetation dataset from 144 site (Physg3br) or 173 (Physg3ar) extant biotopes is proposed.