Toward a new rapid method to assess plant species in a bee pollen diet

Honey bee diet surveys require an accurate knowledge of the pollen supply and a rapid counting method for microscopic analyses. Referring to the literature, we aimed at reducing the number of pollen grains counted in order to decrease the time spent for each sample. For this, we compared counts on whole transects to alternative methods of counting by sub-sampling, using the non-acetolysed palynological method. For each pellet sample, a drop of a homogenized suspension was dried, degreased, and mounted onto a slide dressed with glycerol gelatin. The counting was performed through a digital camera fitted on a microscope. The study was carried out in two phases. Phase 1 was a test of slide homogeneity, carried out on a 10-slide set with one replicate. For each slide, we compared an exhaustive count of each present taxon on a whole transect divided in four quarters. Phase 2 was a search for an alternative counting method. The work was performed over the first six samples with replicates retaining exactly the same transect, but counting only one microscope field in three, then four, and finally one in five. Data analyses were carried out with Chi-square tests on taxa that were present in more than 5% of the sample. Phase 1 showed some cases of heterogeneity in microscope slides, leading to an error hazard when counting in a continuous mode up to a defined grain number. In phase 2, the comparison of alternative countings to full transects verified the linearity of the counting results. Suitable results were given by the reading method one in three. We improved the accuracy of this method for the particular case of samples containing large pollen grains, e.g., Zea mays, which are unevenly disseminated on the slide.