Freshwater quality has been changing due to the ever greater use of water resources and the contamination load resulting from human activities. Management of these systems, thus, requires constant diagnose of water quality with fast and efficient methodologies. The conventional methods adopted are, however, time-consuming, often very expensive, and require specialised expertise. Raman spectroscopy (RS) is a simple, fast and label-free technique that can be applied to environmental diagnosis using diatoms. Here, we developed a diagnostic method based on Raman spectroscopy applied to freshwater diatoms. For this, Raman spectra were recorded from diatoms of three lakes of a natural city park. The data acquired was analysed by chemometrics methods to describe the data (Partial Least Squares Regression), infer relationships in the dataset (Cluster Analysis) and produce classification models (Artificial Neural Network). The classification models developed diagnosed the lakes with excellent accuracy (89%) without requiring taxonomic information about the diatom species recorded. This study provides a proof-of-concept for the application of diatom Raman spectroscopy to diagnosing water quality, laying an important foundation for future environmental studies aiming at assessing freshwater systems, to be replicated at larger scales and to varied geographic settings.
Keywords: Environmental health; Frustule; Lipids; Pigments; Raman spectra.
Copyright © 2021 Elsevier B.V. All rights reserved.