Abstract

Seasonal changes in removal of natural organic matter (NOM) by drinking water biofilters are often attributed to temperature differences. Bench-scale sand biofilters treating NOM isolated from a surface water source were operated in parallel at 5, 20, and 35°C to isolate the effect of temperature from other water quality and operational parameters, which also vary seasonally. The biofilter operated at 5°C achieved significantly lower removal of NOM and the NOM fraction that reacts with disinfectants (disinfection byproduct precursors) compared to the filters operated at 20 and 35°C, which had similar performance levels. Viable biomass, measured as lipid phosphate, was significantly higher at the top and bottom of the filter operated at 20°C. Phospholipid fatty acid (PLFA) profiles indicated an increasing gradient in markers for Gram-negative bacteria and microeukaryotes as biofilter operation temperature decreased, replacing general fatty acids and markers for Gram-positive bacteria and sulfate-reducing bacteria, which were observed in greatest abundance in the filter operated at 35°C. Principal components analysis differentiated the microbial PLFA profiles based on biofilter operation temperature and filter depth. These results were corroborated by identifications of the dominant microbial colonies isolated on R2A agar. Introduction Filtration with attached biomass can be an effective drinking water treatment process for the control of biodegradable organic matter (BOM). However, biofilter performance is impacted by factors such as water quality (e.g., BOM concentration and characteristics, pH, turbidity), temperature, backwash chemistry, and design parameters. Optimization of this treatment step requires continuing refinement of our knowledge about the relative importance and impact of these factors on biofilter performance. Recent experiments have assessed biofilter performance under different conditions such as varying contact time (1-5), biomass concentration in the filter bed (6, 7), preozonation dose (8, 9), filter media (10, 11), and backwash chemistry (10-13). The removal of total organic carbon (TOC), biodegradable dissolved organic carbon (BDOC), assimilable organic carbon (AOC), natural organic matter (NOM) constituents that react with chlorine or other disinfectants to form chlorinated or oxidized products [disinfection byproduct (DBP) precursors], ozonation byproducts, and turbidity have been used to quantify biofilter performance Several studies have demonstrated that filter designs that support a greater biomass provide better removal of BOM in drinking water biofilters Many drinking water sources experience seasonal temperature variations of 20-30°C. This variability in source water temperature may impact removal of BOM and particulates by biofilters, as well as impact the microbial community structure present in biofilters. Previous research assessing the impact of temperature on biofiltration has been limited to long-term studies on full-and pilot-scale biofilters, where the filters were run at ambient temperature, and seasonal fluctuations were monitored. These studies showed that TOC or dissolved organic carbon (DOC), AOC, and chlorine demand removal efficiencies decline during the winter months A problem with directly relating temperature to biofilter performance in these studies is that conditions besides temperature, such as BOM concentration and nature, particulates, and microbial community structure, fluctuate seasonally due to normal seasonal cycles in surface water bodies (24). These water quality changes often require adjustments in treatment plant operation, such as coagulant and ozone doses, and pH. During these pilot-and full-scale studies of biofiltration, such operational adjustments were required and the stability of the BOM characteristics was not addressed; thus, the impact of temperature could not be isolated. This study was designed to examine the impact of temperature on biofilter performance and community structure under controlled conditions, such that the effects of seasonal water quality and microbial variability are eliminated. Materials and Methods Filter Operation and Sampling. Biofilter sand media were acclimated in the upflow mode with settled Ohio River water for 1 month prior to packing 40 cm of the acclimated media into 2.54 cm × 60 cm glass chromatography columns (Ace Glass, Louisville, KY). Sand was the chosen biofilter medium for these experiments because it is the most commonly used filter medium in full-scale rapid media filters. While high microbial densities can be achievable with granular activated carbon (GAC), it is difficult to separate BOM removal by adsorption from microbial utilization. During the 45-73 days of filter operation, solutions of NOM, isolated by nanofiltration (NF) from Manatee Lake (FL) water (MLW), were used as the feedwater. The NOM was isolated in one batch in October 1997 with a Filmtec NF90 membrane filter (Hiniker, Mankato, MN) using a modification of the method of Serkiz and Perdue (25). DOC analysis and fractionation into hydrophobic and hydrophilic fractions by using XAD-8 resin were performed immediately following concentration of the NOM and 2 months later, during the study. The DOC †