Microbial community assembly and metabolic function in top layers of slow sand filters for drinking water production

Slow sand filters (SSFs) are widely applied to treat potable water; the removal of contaminants (e.g., particles, organic matter, and microorganism) occurs primarily in the top layer. However, the development of the microbial community and its metabolic function is still poorly understood. In the present study, we analyzed the microbial quantity and community of the influents sampled from the effluent of the last step (rapid sand filtration) and of the top layers of SSFs (Schmutzdecke, 0–2 cm, 4–6 cm, 8–10 cm) sampled near terminal head loss when the Schmutzdecke (SCM) was most developed in two full-scale drinking water treatment plants (DWTPs). The two DWTPs use the same artificially recharged groundwater source. The biomass in the filter, quantified by flow cytometric intact cell counts (ICC) and adenosine triphosphate (ATP), decreased rapidly along the depth till 8–10 cm (>1 log TCC; >75% ATP); the decrease was most pronounced from the SCM to the surface sand layer (0–2 cm), after which the biomass stabilized quickly at lower depths (2–10 cm). Remarkably, beta diversity showed that SSFs layers of the same depth in two DWTPs with distinctive filter age and plant location clustered together, which indicated their insignificant effects in shaping microbial communities in SSFs. The alpha diversity indices followed the trend of the biomass, suggesting more active and diverse communities in SCM layer. PICRUSt-based function prediction revealed significant over-representation of metabolism and degradation of complex organic matters (e.g., butanoate, propanoate, xenobiotic, D-Alanine, chloroalkene, and bisphenol) in SCM layer, the functional importance of which was confirmed by the co-occurrence patterns of the dominant taxa and metabolic functions. Using an island biogeography model, we found that microbial communities in SSFs were strongly assembled by selection (68 OTUs, 50.0% sequences), rather than by simple accumulation of the microbial communities in the influents (120 OTUs, 44.8% sequences). Our findings enhance the understanding of microbial community assembly and of metabolic function in the top layers of SSFs, and constitute a valuable contribution to optimizing the design and operation of biofilters in full-scale DWTPs.