The hidden biodiversity of pollen

The expertise needed to perform the laborious task of pollen analysis is rapidly disappearing. Moreover, many plant taxa produce highly similar pollen that cannot be distinguished beyond genus, family or even order level. This prevents detailed information to be gained from pollen analysis, as different species may have diverse ecological preferences or allergenic profiles. Information from pollen is of high societal relevance and is used in a multitude of research fields including allergology, taxonomy, forensics, biostratigraphy, apiology, paleoecology and aerobiology. Therefore, there is a high need for new techniques to help transform palynology. In this thesis, innovative microscopic and molecular techniques are used. The aim is to unravel hidden pollen biodiversity.

The main findings are that 1) neural networks can distinguish pollen types that cannot be confidently distinguished by palynologists 2) DNA metabarcoding can be used to accurately identify pollen to the species level and 3) DNA metabarcoding provides a reliable semi-quantitative measure of pollen grain abundance in mixed samples.