Some pages with notes and ideas

Reconstructing past vegetation dynamics

One of the most detailed methods available to date to reconstruct past vegetation dynamics is the analysis of fossil pollen. Unfortunately, fossil pollen analysis does not distinguish beyond family or generic level in most cases, while its spatial resolution is limited amongst others by windblown dispersal of pollen, affecting the accuracy of pollen-based vegetation reconstructions. To overcome these limitations, I developed together with Boris Jansen the VERHIB model, which can exploit the information in of plant-specific groups of biomarkers preserved in suitable archives (e.g. peat deposits). VERHIB is based on an explicit description of an accumulation process of vegetation biomarkers in a marine or lacustrine sediment, peat deposit or soil. VERHIB inverts this forward model to reconstruct paleovegetation on the basis of the observed accumulated biomarker signal. Tikhonov regularization is used to cope with non-uniqueness resulting from the combination of fine-scale output resolution and limited sample numbers. Hard as well as soft constraints, derived from biological principles and dynamical models are included to ensure model results are always within physically and biologically realistic limits. More details can be found in this publication.

We are continuing and broadening this work a bit. A major challenge is to combine qualitative and quantitative information from distinct sources (e.g. pollen as well as biomarkers) in the inversion process. In addition, it will be crucial to build-up a database of species or genus-specific biomakers and collect knowledge about the degradation of these biomarkers in different environments over long time spans.