• Microbial fuel cells (MFC) can act as biosensors via electrochemical signal shifts. • Urea contamination reduces voltage output and diversity in soil-based MFCs. • Diurnal temperature cycles have minimal impact on voltage or community structure. • Bacterial diversity is higher in controls (no urea) and under cyclic temperatures. • Machine learning identified contamination in tMFCs with 99% accuracy. Soil based terrestrial microbial fuel cells (tMFCs), traditionally used as batteries, could also serve as biosensors whereby the addition of an anthropogenic compound changes their voltage output. The ability and rate of electron transmission at the anode surface shifts based on external stimuli (e.g., temperature or contaminant concentration and frequency). This study investigates microbial response to diurnal temperature cycles and whether the shifts obscure voltage changes induced by the addition of urea. Replicate tMFCs were constructed with either 70 mg/ml urea or water and incubated at either a 25°C static or 10°C to 25°C cycling temperature regime. Voltage was measured for 30 days and the tMFC electrodes were destructively sampled to determine bacterial taxonomy. Though voltage of all tMFCs increased throughout the incubation period, those exposed to urea had an output of 692 mV compared to 1011 mV observed in the controls. In both control and urea tMFCs, maximum voltage output during peak performance was significantly greater in the static temperature incubation than in cyclic (p-value = 0.004 and 0.037, respectively). Bacterial diversity was significantly higher (p-value < 0.0001) in the control tMFCs and was significantly higher (p-value = 0.003) at the cathode surface under cyclic temperature conditions. A previously trained machine learning model was able to accurately discern control vs. urea tMFCs despite the diurnal temperature conditions with 99% accuracy. These data show that diurnal temperature fluctuations had minimal impact on the bacterial community, and the presence of a contaminant was a strong predictor of tMFC community composition.
DiDominic et al. (2026) studied this question.