Stochastic modeling of bacterial transport and retention during aquifer artificial recharge

Managed aquifer recharge (MAR) sytems based on water filtration allow to improve recharged water quality and quantity by retaining suspended particles and microorganisms. However, the periodic detection in groundwater of pathogens and other microorganisms that represent a significant risk for human health makes it necessary to study the mechanisms affecting the propagation and fate of microbial populations during the process.

In this work a series of column experiments were performed to characterize bacteria transport in porous media. Two type of columns were built. One using only sand and another using a combination of sand, compost and wood chips. In each column a punctual injection of tracers (rhodamine and amino-G acid) and bacteria consortium collected from the effluent of a wastewater treatment plant were injected. Samples of column outflows were collected to obtain breakthrough curves of the tracers and the different amplicon sequence variants (ASVs) of bacteria to determine the material influence on the retention of bacteria. Bacteria displayed a strong anomalous behavior with late arrival peaks and longer tails than those obtained with the tracers.

A continuous time random walk (CTRW) transport model was developed to interpret the experimental results. The model characterizes transport in terms of mobile-immobile domains. Bacteria are transported with the mean flow and experience transitions from and to low mobility zones with a certain frequency. Transport is described in terms of four parameters, namely, the mean flow velocity, the dispersion coefficient, the trapping rate, and the mean residence time in the immobile zones. The model was able to reproduce satisfactorily the observed breakthrough curves of over 470 measured ASVs. The analysis of the breakthrough curved determined that bacteria form two clusters. The breakthrough curve of one cluster has heavy tails and it is formed by small, motile, gram-negative bacteria. The other cluster displays strong peaks and a relatively weaker tailing. CTRW parameters are able to predict the cluster in which a certain bacteria belongs.