Identification of pollen types of beekeeping interest by non-targeted mass spectrometry

Background The identification of pollen is important in the field of beekeeping for the determination of the botanical origin of bee products and investigations of bee diet. Until now, it has been performed by melissopalynology, the microscopic examination of pollen grains. However, this technique has some limitations, such as the necessity of experienced analysts and identification restricted to the family level for some pollen types. Although many techniques have been proposed as alternatives or complements to melissopalynology and omics techniques have been explored to gather information on the botanical origin of honey, no study has yet been conducted on a large set of pollen types. Results The study dataset consisted of 34 different pollen types of pellets collected by honeybees in Switzerland and analyzed in multiple biological replications, leading to 150 observations. The pollen samples were analyzed after tryptic digestion using a non-targeted mass spectrometry-based method. Liquid chromatography coupled with mass spectrometry (LC–MS) was employed to identify pollen, and melissopalynology was used as a reference method for the identification. We built an OPLS-DA prediction model for the 34 pollen types. The model clearly identified new samples in their membership group (Acer sp., n = 10) and a new pollen type at the species-specific level for Quercus sp. Less predictable results were achieved for Composita H and pollen collected directly from the plant. Conclusion The use of a non-targeted mass spectrometry-based method and chemometrics resulted in a promising tool for pollen identification as a replacement/supplement method to traditional melissopalynology.