Pollen identification and quantification are essential in ecological and evolutionary research to address plant-pollinator relationships, pollination services, and plant reproduction. Research into pollen transfer patterns has been mainly based on traditional light microscopy, however, it is time consuming, labour intensive, and requires taxonomic expertise. High-throughput methods allow automated pollen identification across large temporal and spatial scales. This study compares the accuracy of species identification and quantification of their relative pollen abundance using traditional microscopy, and two high-throughput methods, namely multispectral imaging flow cytometry (MIFC) and metabarcoding. Method performance was tested using artificial samples with known pollen species composition and pollen samples from pollinators with unknown pollen composition. After checking the agreement against the line of identity, the coefficient of determination (R2) values of linear models between the method estimates were compared. Metabarcoding performed best at identifying the taxa from artificial mixtures, while the two other methods assessed the relative abundance most accurately when there was information about species identity. Comparability between methods was overall low when assessing pollen composition on pollinators. To acquire both pollen identity and relative abundance, we recommend a metabarcoding-guided MIFC analysis that can be used as a high throughput approach for pollen research at large spatial and temporal scales.
Švara et al. (Thu,) studied this question.