In Environmental science & technology ; h5-index 132.0
Stability as evaluated by functional resistance and resilience is critical to the effective operation of environmental biotechnologies. To date, limited tools have been developed that allow operators of these technologies to predict functional responses to environmental and operational disturbances. In the present study, 17 Microbial Fuel Cells (MFCs) were exposed to a low pH perturbation. MFC power dropped 52.7 ± 35.8% during the low pH disturbance. Following the disturbance, 3 MFCs did not recover while 14 took 60.7 ± 58.3 h to recover to previous current output levels. Machine learning models based on genomic data inputs were developed and evaluated on their ability to predict resistance and resilience. Resistance and resilience levels corresponding to risk of deactivation could be classified with 70.47 ± 15.88% and 65.33 ± 19.71% accuracy, respectively. Models predicting resistance and resilience coefficient values projected postperturbation current drops within 6.7-15.8% and recovery times within 5.8-8.7% of observed values. Results suggest that abundances of specific genera are better predictors of resistance while overall microbial community structure more accurately predicts resilience. This approach can be used to assess operational risk and is a first step toward the further understanding and improvement of overall stability of environmental biotechnologies.
Lesnik Keaton Larson, Cai Wenfang, Liu Hong