[Science News] - Identifying fossilised moss spores thanks to Meise Botanic Gardens' SEM imaging expertise

Scanning electron microscopy (SEM) is an important tool, and the SEM-team at Meise Botanic Garden has many years of expertise in imaging the tiniest particles, such as unicellular algae, pollen grains, and spores of mosses and fungi. This expertise is essential for gathering historical data and learning from past events will help us better understand climate change. 

When core sediment samples from the Quaternary age (2.58 million years ago to present) are studied for reconstructing the vegetation of the past, the data collected is almost exclusively based on pollen grains produced by conifers and flowering plants. This excludes other components of the land flora, such as spores from bryophytes (mosses). Bryophytes comprise a significant taxonomic diversity of about 22 000 species thriving across all ecosystems except marine. 

The absence of identification keys for bryophyte spores has caused them to be a largely unexploited resource in palaeoecology. This is why Prof. Dr. Alain Vanderpoorten, a bryophyte specialist at the University of Liège, wanted to accumulate SEM images of spores of all known bryophytes in order to create a digital identification tool. His idea has taken an inventive turn when he considered the possibility of creating an AI-based application.

His PhD student Ms. Alix Milis has now conceptualised the idea and is coordinating a multi-institutional research presented in a recent paper in the journal Trends in Plant Science. Meise Botanic Garden plays an important role in this project on two counts. Firstly, its herbarium holds a large collection of 420.000 bryophyte specimens and secondly, the institute is renowned for its high quality SEM-imaging. 

The SEM-team of Meise  Botanic Garden has collaborated with Prof. Vanderpoorten since 2015 and has produced the images of spores collected from herbarium specimens that have been “fed” to the machine learning algorithms, in this case deep neural networks (DNNs) with astoundingly good results. Once the DNNs can recognise enough taxa the door is opened towards the identification of fossilised bryophyte spores in core samples. This tool will generate a higher insight in past climatic impact, and consequently lead to a better understanding of our future climate. 

Reference

Milis, A., Mäder, P., de Haan, M., Ballings, P., Van der Beeten, I., Goffinet, B., & Vanderpoorten, A. (2025a). Time to spice-up paleoecological records with bryophyte spores. Trends in Plant Science.