Abstract
Large seasonal variations in microbial drinking water quality can occur in distribution networks, but are often not taken into account when evaluating results from short- term water sampling campaigns. Temporal dynamics in bacterial community characteristics were investigated during a two- year drinking water monitoring campaign in a full- scale distribution system operating without detectable disinfectant residual. A total of 368 water samples were collected on a biweekly basis at the water treatment plant (WTP) effluent and at one fixed location in the drinking water distribution network (NET). The samples were analysed for heterotrophic plate counts (HPC), Aeromonas plate counts, adenosine- tri- phosphate (ATP) concentrations, and flow cytometric (FCM) total and intact cell counts (TCC, ICC), water temperature, pH, conductivity, total organic carbon (TOC) and assimilable organic carbon (AOC). Multivariate analysis of the large dataset was performed to explore correlative trends between microbial and environmental parameters. The WTP effluent displayed considerable seasonal variations in TCC (from 90 x 10(3) cells mL(-1) in winter time up to 455 x 10(3) cells mL(-1) in summer time) and in bacterial ATP concentrations (< 1-3.6 ng L-1), which were congruent with water temperature variations. These fluctuations were not detected with HPC and Aeromonas counts. The water in the network was predominantly influenced by the characteristics of the WTP effluent. The increase in ICC between the WTP effluent and the network sampling location was small (34 x 10(3) cells mL(-1) on average) compared to seasonal fluctuations in ICC in the WTP effluent. Interestingly, the extent of bacterial growth in the NET was inversely correlated to AOC concentrations in the WTP effluent (Pearson's correlation factor r = - 0.35), and positively correlated with water temperature (r = 0.49). Collecting a large dataset at high frequency over a two year period enabled the characterization of previously undocumented seasonal dynamics in the distribution network. Moreover, high- resolution FCM data enabled prediction of bacterial cell concentrations at specific water temperatures and time of year. The study highlights the need to systematically