We report on the in situ characterization of tree pollen
molecular composition based on Raman spectroscopy.
Different from purification-based analysis, the nondestructive
approach allows (i) to analyze various classes of
molecules simultaneously at microscopic resolution and
(ii) to acquire fingerprint-like chemical information that
was used for the classification of pollen from different
species. Hierarchical cluster analysis of spectra from fresh
pollen samples of 15 species partly related at the genus
level and family level indicates separation of species based
on the complete Raman spectral signature and yields
classification in accord with biological systematics. The
results have implications for the further elucidation of
pollen biochemistry and also for the development of
chemistry-based online pollen identification methods.