A new tool for formalised vegetation reconstruction from (sub)fossil records – the FEVER Index

Plant macro-remains provide valuable environmental information of the past, but reconstruction of past vegetation is challenging, because a macrofossil sample may include material from various habitats and also because its species composition is biased and incomplete. Therefore, we aimed to propose, test and evaluate an objective tool for data reconstruction in archaeobotany and palaeoecology. Our Fossil assEmblage VEgetation Reconstruction Index (FEVER Index) indicates relative probabilities that particular taxa in a fossil assemblage come from respective vegetation types. In contrast to the Frequency Positive Fidelity Index (FPFI) used for modern vegetation classification, the FEVER Index emphasises the importance of diagnostic species. The comparison between the FEVER and FPFI indices, when they are applied to a large dataset of modern vegetation plots, has shown that the FEVER Index has greater classification accuracy. In the case where taxonomic data were reduced to genera only, the efficiency of the FEVER Index was even higher than FPFI. This shows that the FEVER Index is more accurate when applied to incomplete fossil data, but only when there are some diagnostic species still present. We also examined the similarity between modern vegetation and corresponding seed bank data. Wetland habitats, such as calcareous fens and periodically exposed riverbeds showed high similarity between the vegetation and the seed banks because of the local origin of the seed bank material. Lower similarity was, however, detected in the case of small pools in the upper reaches of the river Lužnice, the seed bank of which included not only aquatic vegetation but also plants from terrestrial habitats nearby, transported by flowing water. Finally, we provide two examples of applying the FEVER Index to fossil data.