Consider the Community: Developing Predictive Linkages between Community Structure and Performance in Microbial Fuel Cells

Graduation date: 2017The complex, dynamic nature of microbial communities in both natural and engineered environments complicates the work of scientists and engineers who wish to channel microbial interactions for societal good. The successful management of these communities towards engineering goals is dependent on developing predictive linkages between community structure and functional outputs. The performance of microbial fuel cells (MFCs), an emerging environmental biotechnology, is driven by a diverse microbial community capable of converting the chemical potential energy contained in waste streams to electrical energy. This technology stands to benefit greatly from an increased understanding of the microorganisms contained within as it transitions from the laboratory to practical application. MFCs also offer a controlled environment in which new approaches to developing predictive understandings of microbial communities can be developed.\ud Revolutions in molecular science over the past decade paved the way for the rapid increase in genomic data available for microbial communities from a wide range of environments. Increases in computing power and accessibility over the same period provide a means in which the amassed community data can be mined for potential interactions and linked to functional outcomes. One of the methods through which this can be done is the use of artificial neural networks (ANNs). ANN-based models can be used to generate accurate microbial assemblage predictions across a variety of environments, but have never been applied to the microbial communities of environmental biotechnologies.\ud In the present dissertation, MFC biofilms are analyzed over time, across reactor designs, under varying environmental conditions, and following pH disruption to identify core community membership. Results demonstrated that deterministic interactions shaped consistent community structures characterized by the formation of highly conductive anodic biofilms. The core MFC community is defined by a high abundance of anode-respiring Geobacter sulfurreducens. and biomass fermenting Aminiphilus circumscriptus along with other syntrophic bacteria. Community structure shifted into repeatable formations following the introduction of various substrates and wastewaters. Under changing conditions reactor performance in terms of power generation, treatment rates, and coulombic efficiencies was repeatable and linked to community composition using ANN models. ANN models that incorporated community predictions performed significantly better than those solely based on environmental parameters and predicted all performance metrics within 6% providing the first evidence for the value of including community data into ANN-based MFC models. Community composition could also be linked to biofilm stability following exposure to low pH solutions. Through the first quantitative evaluations of biofilm\ud resilience in MFCs a correlation between the relative abundance of Geobacteraceae and process stability was observed, however, ANN models that considered relative abundance of other bacteria predicted stability more accurately. Further development of these models can be used in practical settings to determine and avoid risk of deactivation during operation.\ud This dissertation characterizes a single MFC community over a variety of conditions and represents the first attempt to use machine-learning based approaches to connect community structure to performance in environmental biotechnology applications. The further development of these and other similar artificial intelligence data-mining tools will improve the management of microbial communities that drive environmental biotechnologies like MFCs and spur them towards practical application. Strengthening linkages between community, structure, interactions, and function in these technologies may be applied across industries, inspiring new applications and innovations involving microbial communities