M. Kansiz et al., Fourier Transform Infrared microspectroscopy and chemometrics as a tool for the discrimination of cyanobacterial strains, PHYTOCHEM, 52(3), 1999, pp. 407-417
Fourier Transform Infrared (FTIR) microspectroscopy, in combination with ch
emometrics, was investigated as a novel method to discriminate between cyan
obacterial strains. In total, 810 absorbance spectra were recorded from one
eukaryotic and five cyanobacterial taxa spanning three genera and includin
g two strains of one species, Microcystis aeruginosa. Principal Component A
nalysis (PCA) based classification techniques such as Soft Independent Mode
lling of Class Analogy (SIMCA) and K-Nearest Neighbours (KNN) were investig
ated. Different spectral regions using derivative spectra were investigated
to find the best combinations for classification. The highest rate of corr
ect classifications (99-100%) was achieved using first derivative spectra w
ith a spectral region of 1800-950 cm(-1) for both the SIMCA and KNN. A dend
rogram constructed using averaged spectra of the six taxa studied showed th
at the two strains of Microcystis aeruginosa exhibited the highest degree o
f similarity, while the eukaryotic taxon was the most dissimilar from the p
rokaryotic taxa. (C) 1999 Published by Elsevier Science Ltd. All rights res
erved.