Abstract/Summary

Many methods have been devised to count pollen grains automatically; however, few combine speed, reliability, inexpensiveness and user friendliness. This study describes a combination of simple, glycerine-based extraction, digital imaging and free particle counting software configured to achieve semi-automated processing of a large volume of images. Pollen grains were extracted from anthers of 10 common perennial grass (Poaceae) species, all implicated in pollinosis in Europe, and samples, illuminated on slides and digitally imaged. ImageJ algorithms were designed to remove significant extraneous content and count just the pollen grains, then applied in batch mode on multiple images. Accuracy was assessed by comparing a sample of automated software counts to manual, visual counts of the same images and found to be high. Total pollen production per anther and per inflorescence was estimated by counting the number of pollen grains per anther and the number of florets per inflorescence. Methodological and natural variation in pollen counts is discussed. Results were compared to published pollen counts of the same species; new pollen production figures are published for Cynosurus cristatus. This method is portable to other plant species, and requires only readily available reagents, equipment and software, it is quick, reliable, inexpensive and user friendly.