Pyrosequencing analysis yields comprehensive assessment of microbial communities in pilot-scale two-stage membrane biofilm reactors.

We studied the microbial community structure of pilot two-stage membrane biofilm reactors (MBfRs) designed to reduce nitrate (NO3(-)) and perchlorate (ClO4(-)) in contaminated groundwater. The groundwater also contained oxygen (O2) and sulfate (SO4(2-)), which became important electron sinks that affected the NO3(-) and ClO4(-) removal rates. Using pyrosequencing, we elucidated how important phylotypes of each "primary" microbial group, i.e., denitrifying bacteria (DB), perchlorate-reducing bacteria (PRB), and sulfate-reducing bacteria (SRB), responded to changes in electron-acceptor loading. UniFrac, principal coordinate analysis (PCoA), and diversity analyses documented that the microbial community of biofilms sampled when the MBfRs had a high acceptor loading were phylogenetically distant from and less diverse than the microbial community of biofilm samples with lower acceptor loadings. Diminished acceptor loading led to SO4(2-) reduction in the lag MBfR, which allowed Desulfovibrionales (an SRB) and Thiothrichales (sulfur-oxidizers) to thrive through S cycling. As a result of this cooperative relationship, they competed effectively with DB/PRB phylotypes such as Xanthomonadales and Rhodobacterales. Thus, pyrosequencing illustrated that while DB, PRB, and SRB responded predictably to changes in acceptor loading, a decrease in total acceptor loading led to important shifts within the "primary" groups, the onset of other members (e.g., Thiothrichales), and overall greater diversity.

Removal of multiple electron acceptors by pilot-scale, two-stage membrane biofilm reactors.

We studied the performance of a pilot-scale membrane biofilm reactor (MBfR) treating groundwater containing four electron acceptors: nitrate (NO3(-)), perchlorate (ClO4(-)), sulfate (SO4(2-)), and oxygen (O2). The treatment goal was to remove ClO4(-) from ∼200 µg/L to less than 6 µg/L. The pilot system was operated as two MBfRs in series, and the positions of the lead and lag MBfRs were switched regularly. The lead MBfR removed at least 99% of the O2 and 63-88% of NO3(-), depending on loading conditions. The lag MBfR was where most of the ClO4(-) reduction occurred, and the effluent ClO4(-) concentration was driven to as low as 4 µg/L, with most concentrations ≤10 µg/L. However, SO4(2-) reduction occurred in the lag MBfR when its NO3(-) + O2 flux was smaller than ∼0.18 g H2/m(2)-d, and this was accompanied by a lower ClO4(-) flux. We were able to suppress SO4(2-) reduction by lowering the H2 pressure and increasing the NO3(-) + O2 flux. We also monitored the microbial community using the quantitative polymerase chain reaction targeting characteristic reductase genes. Due to regular position switching, the lead and lag MBfRs had similar microbial communities. Denitrifying bacteria dominated the biofilm when the NO3(-) + O2 fluxes were highest, but sulfate-reducing bacteria became more important when SO4(2-) reduction was enhanced in the lag MBfR due to low NO3(-) + O2 flux. The practical two-stage strategy to achieve complete ClO4(-) and NO3(-) reduction while suppressing SO4(2-) reduction involved controlling the NO3(-) + O2 surface loading between 0.18 and 0.34 g H2/m(2)-d and using a low H2 pressure in the lag MBfR.