Combining remote sensing and AI for algae bloom analysis in NB
Catherine Evans (Dalhousie)

The biological structure and function of microbial communities drive ecosystem services in many under-explored environments such as water supply lakes and engineered water systems. Drinking water utilities, distribution networks, and wastewater treatment plants WWTP are designed to facilitate removal of contaminants like nutrients ammonia, nitrogen, phosphorus, metals lead, manganese, harmful metabolites cyanotoxins, and organic carbon under highly variable influent characteristics. Changes in microbiomes can produce both desirable and undesirable performance outcomes in such systems.

Microbial community datasets are large and complex. An accessible microbiome data analysis toolset for dynamic water and wastewater environments will be critical to determining microbial interactions and biochemical mechanisms that mediate process outcomes. The availability of such an analysis toolset will be essential to bridge operator observation and experience with modern monitoring tools. The objective of this presentation is to demonstrate and describe a novel microbiome data analysis toolset and use it to provide simplified interpretation of biotic and abiotic factors affecting water utility performance, and guide targets of quantification genes and microbes for further evaluation and forecasting.

The primary analysis tool uses Poisson Principal Component Analysis PPCA to cluster similar microbiomes, then visualizes using colours and labels how the clustered communities relate to informative functions derived from metabolic activity. This will demonstrate how monitoring samples can be interpreted for biologically distinguishing treatment process outcomes. Key microorganisms and genes that explain the variation between clusters in PPCA will be identified and described according to their putative metabolic functions and predicted effects on water quality. Further, we show this toolset generates corroborative and novel insights with an emphasis on how the methods are appropriate for the inherent challenges and assumptions of microbiome data. Specifically, PPCA corrects for measurement error, DNA sequencing depth noise, and Next Generation Sequencing NGS workflow bias.