Multispectral imaging flow cytometry for spatio-temporal pollen trait variation measurements of insect-pollinated plants

This data repository contains pollen trait data and code for data analysis supplementing the study:

Walther F, Plos C, Deilmann T, Lenk A, Römermann C, Rakosy D, Harpole WS, Hofmann M, Hornick T, Dunker S (2025): Multispectral imaging flow cytometry for spatio-temporal pollen trait variation measurements of insect-pollinated plants.

Background: AI outperforms humans in object recognition but faces challenges in pollen identification due to high trait variation. Traditional pollen databases are often too small to support robust AI models. In our study, we analysed the traits of 64,001 pollen grains from four plant species using multispectral imaging flow cytometry (MIFC) and examined their spatial and temporal variation. The variation found affects not only ecological aspects but also the accuracy of AI-based classification.

Data:

code:

R_Code_pollen_trait_variation.R

commented R script with all packages used

data:

README.txt: Description of datasets

metadata.csv: Metadata description of all measurements used in study

dataset1.txt: pollen trait values measured with MIFC (only fully developed pollen)

dataset2.txt: pollen trait values measured with MIFC (developed & undeveloped pollen)

data_spiderplots.txt: pollen trait values measured with MIFC prepared for spiderplot plotting

data_AI.txt: machine learning results