A modified weighted mixture model for the interpretation of spatial and temporal changes in the microbial communities in drinking water reservoirs using compositional phospholipid fatty acid data

Abstract

The aim of this work was to check whether a methodology based on the analysis of data that contain the entire phospholipid fatty acid, PLFA, compositions of water samples can be successfully used to interpret spatial and temporal changes in the microbial communities in water reservoirs. The proposed methodology consists of the construction of a modified weighted multivariate mixture model for the PLFA profiles of the water samples collected in a given monitoring campaign and the identification of latent PLFA components through a comparison with the known PLFA profiles of some cultivated or non-cultivated microbial communities. A 16S rDNA analysis of some of the selected water samples in the monitoring campaign was performed in order to verify the results of the PLFA analysis. The results showed that the proposed methodology can be useful for a dynamic and sensitive evaluation of changes in the microbial quality of water before and after flash flooding and can help in taking a decision regarding further risk assessment.

Keywords: Beta diversity; Bray-Curtis similarity; Chemometrics; Compositional data; Missing elements; Multivariate curve resolution for compositional data.