Performance of two quantitative PCR methods for microbial source tracking of human sewage and implications for microbial risk assessment in recreational waters

Abstract

Before new, rapid quantitative PCR (qPCR) methods for assessment of recreational water quality and microbial source tracking (MST) can be useful in a regulatory context, an understanding of the ability of the method to detect a DNA target (marker) when the contaminant source has been diluted in environmental waters is needed. This study determined the limits of detection and quantification of the human-associated Bacteroides sp. (HF183) and human polyomavirus (HPyV) qPCR methods for sewage diluted in buffer and in five ambient, Florida water types (estuarine, marine, tannic, lake, and river). HF183 was quantifiable in sewage diluted up to 10(-6) in 500-ml ambient-water samples, but HPyVs were not quantifiable in dilutions of >10(-4). Specificity, which was assessed using fecal composites from dogs, birds, and cattle, was 100% for HPyVs and 81% for HF183. Quantitative microbial risk assessment (QMRA) estimated the possible norovirus levels in sewage and the human health risk at various sewage dilutions. When juxtaposed with the MST marker detection limits, the QMRA analysis revealed that HF183 was detectable when the modeled risk of gastrointestinal (GI) illness was at or below the benchmark of 10 illnesses per 1,000 exposures, but the HPyV method was generally not sensitive enough to detect potential health risks at the 0.01 threshold for frequency of illness. The tradeoff between sensitivity and specificity in the MST methods indicates that HF183 data should be interpreted judiciously, preferably in conjunction with a more host-specific marker, and that better methods of concentrating HPyVs from environmental waters are needed if this method is to be useful in a watershed management or monitoring context.

Fig 1

Sampling sites used for determination of P_LOD. Sites include Green Swamp (GS) (tannic water), Hillsborough River (HR) (river water), Lake Carroll (LC6) (lake water), Bahia Beach (BB) (estuarine water), and Fort DeSoto (FD) (marine water). The most dilute P_LOD values observed during the study are shown for HPyVs and HF183. The map was generated by use of the Tampa Bay Water Atlas website (http://www.tampabay.wateratlas.usf.edu/).

Fig 2

Relationship of the P_LOD for HF183 and HPyVs with gastrointestinal illness rates estimated for norovirus in sewage. Median probability of GI illness from norovirus (●) is plotted for accidental ingestion of water contaminated with each sewage dilution corresponding to a limit of detection. Error bars show the 25th and 75th percentile ranges for illness probability. Numbers beside each marker designate the water body corresponding to that LOD: 1, Bahia Beach; 2, Fort DeSoto; 3, Green Swamp; 4, Lake Carroll; 5, Hillsborough River. For source waters with a range of LOD, the maximum and minimum P_LOD are included.