Deciphering the core fouling-causing microbiota in a membrane bioreactor: Low abundance but important roles.

Currently, membrane biofouling in membrane bioreactors (MBRs) is normally attributed to the occurrence of abundant bacterial species on membranes, whereas the roles of low-abundance bacteria have not been paid sufficient attention. In this study, the linear discriminant analysis (LDA) effect size (LEfSe) algorithm was used to identify active biomarkers, determining 67 different phylotypes among Bulk sludge, low-fouling Bio-cake (10 kPa), high-fouling Bio-cake (25 kPa) and Membrane pore in a membrane bioreactor with NaOCl backwash. Interestingly, a large proportion of the active biomarkers in bio-cake samples, such as Methylophilaceae, Burkholderiaceae, Paucibacter and Pseudoxanthomonas, did not fall within the abundant taxa (i.e., <0.05% relative abundance), indicating the preferential growth of these low-abundance bacteria on the membrane surface. Furthermore, the characterization of microbial interactions using a random matrix theory (RMT)-based network approach obtained a network consisting of 120 nodes and 228 edges. Specifically, network analysis showed the presence of an intense competition among bacterial species in the fouling-related communities, suggesting that negative interactions have an important effect on determining the microbial community structure. More importantly, the LEfSe algorithm and network analysis showed that most of the core species of the bio-cake, such as Burkholderiaceae, Bacillus and Rhodothermaceae, merely amounted to a very low relative abundance (<1%), suggesting their unrecognized and over-proportional ecological role in triggering the initial biofilm formation and subsequent biofilm maturation during MBR operation. Overall, this work should improve our understanding of the bacterial community structure on the fouled membranes in MBRs.