New multivariate image analysis method for detection of differences in chemical and structural composition of chitin structures in tardigrade feeding apparatuses

Abstract

The feeding apparatus of the eutardigrade Paramacrobiotus richtersi was examined with confocal laser scanning microscopy using three different wavelengths, taking advantage of the autofluorescence of the chitin present in its structures. The main goal was to test new method based on multivariate image analysis for detection of differences in the chemical and structural composition of chitin structures and the relationships between chemical structure and physical properties of feeding structures. The collected images were analyzed using factor analysis under the self-developed MATLAB-based software ImageA. It was possible to observe small differences in chitin composition using the variations (shifts) of emission spectra caused by differences in the microenvironment of chitin fluorophore. Images of feeding apparatuses within the body and discharged by the animal during molt were recorded applying three excitation/emission (ex/em) sets and then analyzed using factor analysis with three methods for rotation of factor scores (without rotation, with orthogonal rotation, and oblique rotation). With this procedure, it was proven that chitin organization in the feeding apparatus of tardigrades is not homogeneous, but changes according to the mechanical and structural requirements of the structures. Besides the new knowledge about the tardigrade feeding apparatus, ImageA was also proven to be reliable and applicable tool for various problems in the analysis of multispectral images, as autofluorescence images are generally considered as difficult to analyze due to the small initial set of input data and high correlation among the images observed at different ex/em wavelengths.