Lake snow caused by the invasive diatom Lindavia intermedia can be discriminated from different sites and from other algae using vibrational spectroscopy. (31st May 2021)

• Record Type: Journal Article
• Authors: Ahmmed, Fatema; Fraser‐Miller, Sara J.; Garagoda Arachchige, Piumika Samanali; Schallenberg, Marc; Novis, Phil; Gordon, Keith C.
• Other Names: Kiefer Wolfgang guestEditor.; Colomban Philippe guestEditor.; Edwards Howell G. M. guestEditor.
• Is Part Of: Journal of Raman spectroscopy. Volume 52:Number 12(2021)
• Page Start: 2597
• Page End: 2608
• Publication Date: 2021-05-31
• Subjects: algae -- chemometrics -- classification -- Lindavia -- Raman spectroscopy; Raman spectroscopy -- Periodicals; 535.846
• DOI: 10.1002/jrs.6161
• Languages: English
• ISSNs: 0377-0486

Abstract

Abstract: Lake snow refers to mucilaginous suspended material in lakes comprising of extracellular polysaccharides (EPS) and other components. The present study employed Raman and infrared (IR) spectroscopy to discriminate lake snow produced by the diatom Lindavia intermedia from different lakes (Wānaka, Wakatipu and Hāwea) in New Zealand and from mucilaginous samples associated with other algae. It is possible to distinguish algal material including extracellular polymeric substances produced by L. intermedia and the genera Didymosphenia, Zygnema, Spirogyra and Nostoc . Furthermore, the study also explored the use of Raman spectroscopy for quantitative detection of lake snow suspended in a water column using partial least squares regression (PLSR). Thirty‐three (33) samples (lake snow, 21; Didymosphenia, 3; Zygnema, 3; Spirogyra, 3; Nostoc, 3) were analysed using Raman and IR spectroscopy. The data analysis was carried out through support vector machine (SVM) and principal component analysis–linear discriminate analysis (PCA–LDA)‐based classification methods. The SVM classification model provided better accuracy (100%) in species discrimination for both the calibration and full cross‐validation sets compared to the accuracy (92%) obtained by the PCA–LDA model. The PCA analysis separated lake snow based on both sampling location and sampling depth. A partial least squares regression (PLSR) model was constructed using different dilutions (0.0001 to 0.0284 mg/ml) of lake snow suspension with two different spectral preprocessing methods (PP1, smoothing + SNV transformation; PP2, smoothing + RBC + SNV transformation) to investigate the ability of 1064 nm Raman in the quantification of suspended algal loading in the water column. The PLSR analysis with PP1 (smoothing + SNV transformation) demonstrated a better correlation coefficient ( R 2 ) of 0.94 with lower RMSEcv of 0.2% compared to PP2 ( R 2, 0.71; RMSEcv, 0.5%). Overall, the present study demonstrated the potential for Raman and IR spectroscopy to detect and distinguish differences within lake snow samples. Abstract : Raman and infrared spectroscopy, with chemometric analysis, may be used to classify different algal growths in New Zealand lakes, which lakes the species came from and the depth of growth for the species.