A preliminary stochastic model for managing microorganisms in a recirculating aquaculture system

Predicting the growth of key microorganisms is essential to improve the efficiency of wastewater treatment of recirculating aquaculture systems (RAS). We have developed a stochastic model to assess quantitatively the microbial populations in RAS. This stochastic model encompassed the growth model into the Monte Carlo simulation and was constructed with risk analysis software. A modified logistic model combined with the saturation growth-rate model was successfully developed to regress the growth curves of six microorganisms. Monte Carlo simulation was employed to model the effects of chemical oxygen demand (COD) on the maximum specific growth rate. Probabilistic distributions and predictions under the different COD ranges were generated for each simulated scenario. The coefficient of determination (R²) and bias factor (Bf) were used to assess the performance of an established model. Logistic model produced a good fit to the growth curve of Flavobacterium sp. (R² = 0.9511), Acinetobacter baumannii (R² = 0.9970), Sphingomonas paucimobilis (R² = 0.9086), Vibrio natriegens (R² = 0.9993), Lutimonas sp. (R² = 0.9872) and Bacillus pumilus (R² = 0.9816). Bacterial population structure was determined by the construction of 16S rRNA gene libraries. A regular variation trend was observed for the dominant groups during the entire process, with a decrease of Cytophaga–Flavobacterium–Bacteroidetes from 37.6 to 18.7 % and an increase in Gammaproteobacteria from 8.5 to 30.6 %. The predicted model agreed well with observed values except for Flavobacterium sp., and the results can be applied to predict key microorganisms in actual environments. The results of this study provide a method to monitor the dynamics of key microorganisms, which can also help to evaluate the impacts of microorganisms on the operations of RAS.