Franziska Walther
(Helmholtz Centre for Environmental Research - UFZ, Physiological Diversity, Leipzig, Germany & German Centre for integrative Biodiversity Research iDiv Halle-Jena-Leipzig, Physiological Diversity, Leipzig, Germany)
Manual microscopic analyses are traditionally the gold standard for various palynological applications. However, the recent trend is towards automated, database-driven pollen analyses that are expected to be cheaper, less time-consuming and allow better reproducibility than traditional microscope-based methods. One such innovative approach is multispectral imaging flow cytometry combined with machine learning. This method rapidly records microscopic brightfield and fluorescence images, along with various pollen traits such as diameter, fluorescence, and shape. Based on these data, a convolutional neural network classifier can be trained to enable specific pollen identification.
This method was tested for main pollen type identification in three monofloral honey samples (Brassica napus, Helianthus annuus, and Castanea sativa). For this purpose, three different classifiers were trained: one with pollen collected from plants only, another with pollen measured from honey only, and a third with a mix of plant and honey pollen.
The results showed that honey pollen from different varieties could be accurately predicted using a classifier that included respective honey pollen.In contrast, pollen collected from plants alone was only partially suitable for detecting honey varieties. Combining plant and honey pollen did not improve identification accuracy; instead, it led to a reduction in accuracy.
Looking to the future, ongoing advances in automated pollen analysis techniques have the potential to significantly support palynological applications by improving the efficiency and accuracy of pollen type identification.
Keywords: honey, melissopalynology, multispectral imaging flow cytometry
| Status Group |
Doctoral Researcher
|
Franziska Walther
(Helmholtz Centre for Environmental Research - UFZ, Physiological Diversity, Leipzig, Germany & German Centre for integrative Biodiversity Research iDiv Halle-Jena-Leipzig, Physiological Diversity, Leipzig, Germany)
Susanne Dunker
(Helmholtz Centre for Environmental Research - UFZ, Physiological Diversity, Leipzig, Germany & German Centre for integrative Biodiversity Research iDiv Halle-Jena-Leipzig, Physiological Diversity, Leipzig, Germany)
Martina Janke
(Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit LAVES, Institut für Bienenkunde, Celle, Germany)
Selina Campbell
(Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit LAVES, Institut für Bienenkunde, Celle, Germany)
Silvio Erler
(Julius Kühn Institute (JKI), Federal Research Centre for Cultivated Plants, Institute for Bee Protection Braunschweig, Germany & Technische Universität Braunschweig, Zoological Institute, 38106 Braunschweig, Germany)
Thomas Hornick
(Helmholtz Centre for Environmental Research - UFZ, Physiological Diversity, Leipzig, Germany & German Centre for integrative Biodiversity Research iDiv Halle-Jena-Leipzig, Physiological Diversity, Leipzig, Germany)
Demetra Rakosy
(Helmholtz Centre for Environmental Research - UFZ, Community Ecology, Leipzig, Germany & German Centre for integrative Biodiversity Research iDiv Halle-Jena-Leipzig, Species Interaction Ecology, Leipzig, Germany)
Stan Harpole
(Helmholtz Centre for Environmental Research - UFZ, Physiological Diversity, Leipzig, Germany & German Centre for integrative Biodiversity Research iDiv Halle-Jena-Leipzig, Physiological Diversity, Leipzig, Germany)
Elsa Friedrich
(Universität Hohenheim, Landesanstalt für Bienenkunde, Hohenheim, Germany)
Annette Schroeder
(Universität Hohenheim, Landesanstalt für Bienenkunde, Hohenheim, Germany)
There are no materials yet.
