In order to identify the origin of the fecal contamination observed in French estuaries, two libraryindependent microbial source tracking (MST) methods were selected: (i) Bacteroidales host-specific 16S rRNA gene markers and (ii) F-specific RNA bacteriophage genotyping. The specificity of the Bacteroidales markers was evaluated on human and animal (bovine, pig, sheep, and bird) feces. Two human-specific markers (HF183 and HF134), one ruminant-specific marker (CF193), and one pig-specific marker (PF163) showed a high level of specificity (>90%). However, the data suggest that the proposed ruminant-specific CF128 marker would be better described as an animal marker, as it was observed in all bovine and sheep feces and 96% of pig feces. F RNA bacteriophages were detected in only 21% of individual fecal samples tested, in 60% of pig slurries, but in all sewage samples. Most detected F RNA bacteriophages were from genotypes II and III in sewage samples and from genotypes I and IV in bovine, pig, and bird feces and from pig slurries. Both MST methods were applied to 28 water samples collected from three watersheds at different times. Classification of water samples as subject to human, animal, or mixed fecal contamination was more frequent when using Bacteroidales markers (82.1% of water samples) than by bacteriophage genotyping (50%). The ability to classify a water sample increased with increasing Escherichia coli or enterococcus concentration. For the samples that could be classified by bacteriophage genotyping, 78% agreed with the classification obtained from Bacteroidales markers.

The microbiological quality of coastal or river waters can be affected by faecal pollution from human or animal sources. An efficient MST (Microbial Source Tracking) toolbox consisting of several host-specific markers would therefore be valuable for identifying the origin of the faecal pollution in the environment and thus for effective resource management and remediation. In this multidisciplinary study, after having tested some MST markers on faecal samples, we compared a selection of 17 parameters corresponding to chemical (steroid ratios, caffeine, and synthetic compounds), bacterial (host-specific Bacteroidales, Lactobacillus amylovorus and Bifidobacterium adolescentis) and viral (genotypes I-IV of F-specific bacteriophages, FRNAPH) markers on environmental water samples (n = 33; wastewater, runoff and river waters) with variable Escherichia coli concentrations. Eleven microbial and chemical parameters were finally chosen for our MST toolbox, based on their specificity for particular pollution sources represented by our samples and their detection in river waters impacted by human or animal pollution; these were: the human-specific chemical compounds caffeine, TCEP (tri(2-chloroethyl)phosphate) and benzophenone; the ratios of sitostanol/coprostanol and coprostanol/(coprostanol+24-ethylcopstanol); real-time PCR (Polymerase Chain Reaction) human-specific (HF183 and B. adolescentis), pig-specific (Pig-2-Bac and L. amylovorus) and ruminant-specific (Rum-2-Bac) markers; and human FRNAPH genogroup II.

Fecal contaminations of inland and coastal waters induce risks to human health and economic losses. To improve water management, specific markers have been developed to differentiate between sources of contamination. This study investigates the relative decay of fecal indicator bacteria (FIB, Escherichia coli and enterococci) and six human-associated markers (two bacterial markers: Bacteroidales HF183 (HF183) and Bifidobacterium adolescentis (BifAd); one viral marker: genogroup II F-specific RNA bacteriophages (FRNAPH II); three chemical markers: caffeine and two fecal stanol ratios) in freshwater and seawater microcosms seeded with human wastewater. These experiments were performed in darkness, at 20 °C and under aerobic conditions. The modeling of the decay curves allows us (i) to compare FIB and markers and (ii) to classify markers according to their persistence in seawater (FRNAPH II < HF183, stanol ratios < BifAd, caffeine) and in freshwater (HF183, stanol ratios < FRNAPH II < BifAd < caffeine). Although those results depend on the experimental conditions, this study represents a necessary step to develop and validate an interdisciplinary toolbox for the investigation of the sources of fecal contaminations.

An inter-laboratory study of the accuracy of microbial source tracking (MST) methods was conducted using challenge fecal and sewage samples that were spiked into artificial freshwater and provided as unknowns (blind test samples) to the laboratories. The results of the Source Identification Protocol Project (SIPP) are presented in a series of papers that cover 41 MST methods. This contribution details the results of the virus and bacteriophage methods targeting human fecal or sewage contamination. Human viruses used as source identifiers included adenoviruses (HAdV), enteroviruses (EV), norovirus Groups I and II (NoVI and NoVII), and polyomaviruses (HPyVs). Bacteriophages were also employed, including somatic coliphages and F-specific RNA bacteriophages (FRNAPH) as general indicators of fecal contamination. Bacteriophage methods targeting human fecal sources included genotyping of FRNAPH isolates and plaque formation on bacterial hosts Enterococcus faecium MB-55, Bacteroides HB-73 and Bacteroides GB-124. The use of small sample volumes (≤50 ml) resulted in relatively insensitive theoretical limits of detection (10-50 gene copies or plaques × 50 ml(-1)) which, coupled with low virus concentrations in samples, resulted in high false-negative rates, low sensitivity, and low negative predictive values. On the other hand, the specificity of the human virus methods was generally close to 100% and positive predictive values were ∼40-70% with the exception of NoVs, which were not detected. The bacteriophage methods were generally much less specific toward human sewage than virus methods, although FRNAPH II genotyping was relatively successful, with 18% sensitivity and 85% specificity. While the specificity of the human virus methods engenders great confidence in a positive result, better concentration methods and larger sample volumes must be utilized for greater accuracy of negative results, i.e. the prediction that a human contamination source is absent.

Enzyme assays for 4-methylumbelliferyl-fi-D-galactopyranosidase and 4-methylumbelliferyl- ,-D-glucuronidase activities were used for rapid detection (25 min) of fecal water pollution and to determine the impact of sewage discharge in coastal waters. Two coastal areas were investigated: (i) an estuary characterized by a high degree of contamination downstream of a discharge from a sewage treatment plant and a low degree of water renewal and (ii) a ijord with a low degree of pollution and a high degree of water renewal. Statistical analysis showed that a global correlation curve could be used to estimate concentrations of culturable fecal coliform bacteria in the two coastal areas, although environmental factors important for cell physiology (e.g., salinity) varied at different sampling locations. The sensitivity limit for detection of 4-methylumbelliferyl-Ip-Dglucuronidase activity corresponded to bacterial concentrations on the order of 10 to 100 CFU/100 ml. The 4-methylumbelliferyl- ,-D-galactopyranosidase assay was less sensitive because of a higher rate of substrate autohydrolysis. The detection limit corresponded to bacterial concentrations on the order of 100 to 1,000 fecal coliforms per 100 ml.