Assessing the severity of diatom deformities using geometric morphometry
Evaluation de l'intensité des déformations de diatomées à l'aide de la morphométrie géométrique
Abstract
Deformities in diatoms are increasingly used as an indicator of toxic stress in freshwaters. However, the percentage of deformities alone often fails at highlighting the magnitude of toxic exposure. An approach combining the assessment of deformities coupled with information on their severity could improve the sensitivity of this biomarker. With the aim of quantifying the deviation from the normal form, we tested the applicability of geometric morphometry to evaluate the degree of deformities in different diatom species. We used photomicrographs of normal and deformed specimens from laboratory cultures of Gomphonema gracile, Nitzschia palea, and of Achnanthidium minutissimum from field samples collected along a gradient of toxic contamination. The geometric morphometry approach is based on several landmarks positioned on the outline of the diatom valves. Statistical analyses were conducted based on the geometrical coordinates of the landmarks. This technique allowed to discriminate between normal and deformed individuals. The geometric morphometry approach revealed a gradient in the intensity of the deformities observed on Gomphonema gracile and Achnanthidium minutissimum, in-line with a priori, visually determined (subjective) classifications. A relationship between the degree of deformity in Achnanthidium minutissimum and a gradient of zinc contamination was found. In contrast, the approach failed to obtain good fit for Nitzschia palea individuals because deformities in this species were more variable in terms of their location on the valves. Geometric morphometry provided encouraging results to objectively quantify the intensity of diatom deformities affecting valve outline, and could easily be implemented in further automatic diatom identification developments.
Domains
Environmental SciencesOrigin | Files produced by the author(s) |
---|