Microbial source tracking to identify human and ruminant sources of faecal pollution in an ephemeral Florida river



Valerie J. Harwood, Department of Integrative Biology, SCA 110, University of South Florida, 4202 E. Fowler Ave., Tampa, FL 33620, USA. E-mail: vharwood@cas.usf.edu



Levels and sources of faecal indicator bacteria (FIB) in an ephemeral Florida river were assessed under different rainfall/flow patterns to explore the effects of rainfall on water quality.

Methods and Results

Quantitative PCR for sewage markers [human-associated Bacteroides HF183 and human polyomaviruses (HPyVs)] and PCR for ruminant faecal markers were used to explore contamination sources. Escherichia coli, faecal coliform and enterococci levels consistently exceeded recreational water quality criteria, and sediment FIB levels were about 100-fold higher compared with water. HPyVs detections cooccurred with HF183, which was frequently detected near septic systems. Ruminant markers were detected only in livestock-grazing areas. Significantly greater faecal coliform and E. coli concentrations were observed under no-flow conditions and the levels of faecal coliforms in water column and sediments were negatively correlated with duration since last rain event.


Septic systems and cattle grazing in this watershed contributed to the formation of FIB reservoirs in sediments, which were persistent following prolonged rainfall.

Significance and Impact of the Study

Ephemeral water bodies that flow only under the direct influence of recent rainfall are rarely studied. FIB levels in the New River in Florida were greater during dry weather than wet weather, which contrasts with most observations and may be attributed to bacterial reservoirs formed in still pool, sediments and water-saturated soils in this subtropical environment.


Faecal indicator bacteria (FIB) such as faecal coliforms, Escherichia coli or enterococci are used as surrogates for enteric pathogens in recreational water bodies (US Environmental Protection Agency 1983, 2002, 2005). FIB are generally nonpathogenic bacteria which inhabit the gastrointestinal tract of warm and cold blooded animals and are shed in large quantities in the faeces (Harwood et al. 1999). Several epidemiological studies have identified correlations between the levels of some FIB and the incidence of gastroenteritis resulting in regulatory limits focused exclusively on FIB concentrations in the water column (Cabelli et al. 1979, 1982; Wade et al. 2003, 2006, 2008; Craun and Calderon 2006; Colford et al. 2007; Heaney et al. 2009). However, several studies have shown the extended persistence of FIB in sediments (Davies et al. 1995; Desmarais et al. 2002; Ishii et al. 2007; Korajkic et al. 2011) and submerged aquatic vegetation (Badgley et al. 2010a) that are resuspended in overlying waters upon disturbance thus confounding the relationship between FIB concentrations and public health risk (Craig et al. 2004; Jin et al. 2004; Graczyk et al. 2007; Philip et al. 2009).

FIB are shed in the faeces of various host species and thus do not provide any indication as to the source of the contamination (Scott et al. 2002; Brownell et al. 2007). Microbial source tracking (MST) methods have been developed to determine the origin of faecal contamination in impacted water bodies (Hagedorn et al. 1999; US Environmental Protection Agency 2005; Stoeckel and Harwood 2007). Library-dependent MST methods rely upon a large database of patterns derived from FIB isolated from various known faecal sources, to which patterns of isolates from unknown sources are compared (Wiggins 1996; Parveen et al. 1997; Hagedorn et al. 1999; Harwood et al. 2000). Library-independent MST methods generally, but not exclusively, target segments of genes unique to host-associated micro-organisms via either conventional PCR or quantitative PCR (qPCR) assays (Bernhard and Field 2000; US Environmental Protection Agency 2005; McQuaig et al. 2006; Korajkic et al. 2009). For example, library-independent MST methods targeting sections of the 16S rRNA gene of Bacteroidales have been developed to distinguish among several sources of faecal contamination including humans (HF183), ruminants (CF128) or canines (BacCan-UCD) (Bernhard and Field 2000; Dick et al. 2005; Kildare et al. 2007; Lamendella et al. 2007; Fremaux et al. 2009). Human polyomaviruses (HPyVs), which cause asymptomatic infections in up to 60% of adults, are found in high concentrations in municipal sewage, and a PCR and qPCR assays have been developed targeting the conserved T-antigen to detect human faecal pollution (McQuaig et al. 2006, 2009).

Tributaries can act as a conveyance medium of faecal contamination via collection of land runoff and serve as a vehicle between impacted fresh water bodies and the coast. New River, a second-order stream with intermittent flow, is located within the watershed of the Hillsborough River which discharges into the Tampa Bay estuary. Previously, the Florida Department of Environmental Protection (FDEP) performed an assessment that confirmed faecal coliform impairment in New River. Based upon the results, the Total Maximum Daily Load Report (TMDL) for New River was issued in 2004 (Florida Department of Environmental Protection 2004). FDEP then initiated an MST study as a part of the Basin Management Action Plan (BMAP) implementation for tributary waters in the Hillsborough River basin in 2007 (Florida Department of Environmental Protection 2008); however, conditions during this study were characterized by very low rainfall, and only one sediment sample was collected during flowing conditions.

The purpose of this study was to further assess the sources of faecal pollution in the New River watershed through comprehensive sampling of water and sediments from multiple locations during flowing conditions, with a focus on high-flow periods. Quantification of FIB by membrane filtration was combined with library-independent MST identification of human- and ruminant-associated markers by conventional PCR. Quantitative PCR methods for measuring human-associated viral (HPyVs) and bacterial markers (HF183) were also employed.

Materials and methods

Sampling schedule and study area

Water and sediment samples were collected from eight sites on the New River in Pasco County, FL over a period of 11 months (October 2010 to September 2011) on eight separate dates (sample events). The frequency and timing of the sample events were determined based upon rainfall and flow conditions. Measurable flow was present during all sampling events except in October and November 2010, and March 2011. With the exception of the October 2010 sampling, all samples were collected following a rainfall event of at least 0·50 inches during the previous 7 days.

The majority of New River is contained in a single channel with the exception of its headwaters, where the current separates to form two distinct branches (Fig. 1). The larger eastern branch originates in wetlands and pasturelands surrounded by wildlife areas. The western branch originates in wetlands situated among relatively new housing subdivisions. The river flows only in response to rainfall and remains confined within its narrow stream banks during normal wet season conditions. Flow is not persistent and decreases rapidly following cessation of rainfall. Flow may be nonexistent during long dry spells, which generally occur in the winter and spring. Following periods of prolonged rainfall or significant storm events, the river overtops its banks and inundates low-lying areas associated with the riverine floodplain.

Figure 1.

The geographical position of the New River tributary as it flows north to south. Inset shows detail of sample sites. Symbols next to sites designate microbial source tracking markers that were detected at least once at the site during the study. (x) HPyVs; (■) HF183 and (●) RumBac.

The Review of Department of Agriculture/Natural Resource Conservation Service (Soil Survey of Pasco County, 1982) soil mapping data indicate that the majority of the New River basin contains poorly to moderately drained soils that exhibit a high groundwater table with slow to very slow infiltration rates. However, there are several areas, most notably within a residential area south of the branch confluence, which contain well-drained soils with high infiltration rates. The entire New River Basin contains <10% impervious surface and the majority of porous land within the basin consists of improved pastures utilized for livestock grazing, and natural areas that support populations of wildlife. The area south of SR 54 (Fig. 1) consists of low-density residential parcels occupied by conventional and mobile homes. Domestic sewage is treated in onsite sewage treatment and disposal systems (OSTDS). Livestock, including cattle, horses and goats, are commonly present at low density.

Sampling sites

Three sites were routinely sampled (NR2, NR3 and DNR3) and five additional sites were sampled on a flexible schedule over the 11-month period (Table 1). NR2 and NR3 were the most-sampled sites during the previous 2008 BMAP implementation study. NR3 is located directly upstream of the main residential area, whereas NR2 is positioned within the residential area itself, downstream from a mobile home community with large capacity OSTDS. Site DNR3 was added because of the large number of homes south of NR2 with OSTDS located in the proximity to the tributary. Box culvert construction at DNR3 during the 2011 study resulted in replacement of contaminated sediments with clean sand. The impact of sediment replacement on FIB concentrations was assessed at this site. A variable site, DNR1, was added during the fourth sampling event to examine the conditions in a rural residential area containing OSTDS downstream of main sampling area and surrounded by large tract of active grazing land.

Table 1. Description of sample sites and potential sources of faecal contamination based on land use
SiteDistances between sampling sites (miles)GPS coordinatesSampling frequencyPotential sources
NR6 (East upstream branch)NR6-NR528°13′53·863″NVariableWildlife
NR5 (East upstream branch)NR5-NR428°13′19·140″NVariableCattle, Wildlife
NR7 (West upstream branch)NR7-NR428°13′17·769″NVariableSeptic, Cattle, Wildlife
NR4 (Confluence of upstream branches)NR4-NR328°13′16·436″NVariableSeptic, Cattle, Wildlife
NR3NR3-NR228°13′11·029″NConsistentSeptic, Cattle, Wildlife
NR2NR2-DNR328°12′51·582″NConsistentSeptic, Wildlife
DNR3DNR3-DNR128°12′28·156″NConsistentSeptic, Wildlife
DNR1DNR-Hillsborough river28°10′45·621″NConsistentSeptic, Cattle

Additional sites were added in August and September north of NR2 to better assess contributions from cattle grazing. Here, the two branches of New River reach a confluence at a marsh (NR4). On the west branch (NR5), the sample was taken on the upstream side of a culvert, where pollution from the cattle grazing area was expected. A sample was also collected north of the cattle grazing area on this branch (NR6), where there was no apparent source of faecal pollution. The sample on the east branch (NR7) was taken where evidence of faecal pollution from cattle and feral pigs was seen. In July, composite soil samples (described in 'Materials and methods') were collected from the newly added sites. Composites 1 and 2 were taken at NR5 (west branch), north and south of the culvert, respectively, and Composite 3 was taken at NR7 (east branch). All composites were taken 10 feet from the edge of the stream, 10 feet apart. A total of six soil samples (~10 g each) were collected to form each composite.

Sample collection

Surface-level grab samples of water (1 l) and the top layer of sediment (~25 g and ~3 cm depth) were collected in sterile containers, placed on ice for transport and processed within 8 h of collection. Physical parameters were measured including water temperature, salinity, turbidity, pH and dissolved oxygen (mg l−1). Rainfall one, 3 and 7 days prior to each sampling date was also documented for each sampling event using Weather Underground data (http://www.wunderground.com). Furthermore, flow rate in m3 s−1 was measured at NR2, NR3 and DNR3. Flow was only measured during sample collection in January, April and July, using a Marsh McBierney Flo-Mate Model 2000 portable flow meter.

Enumeration of faecal indicator bacteria

Water and sediment samples were concentrated by membrane filtration on nitrocellulose membranes (0·45-μm pore size, 47 mm diameter) to enumerate faecal coliforms, E. coli, and enterococci. Sediment samples were first diluted 1 : 10 with sterile buffered water (0·0425 g l−1 KH2PO4 and 0·4055 g l−1 MgCl2, pH 7·2) and then handshaken for 2 min to release bacteria attached to particles (Boehm et al. 2009). Faecal coliforms were enumerated on mFC agar after 24 h incubation at 44·5°C (1995); E. coli were enumerated on mTEC agar after a 2 h incubation at 35°C followed by 24 h incubation at 44·5°C (US Environmental Protection Agency 2002a); enterococci were enumerated on mEI agar at 41°C after 24 h incubation (US Environmental Protection Agency 2002b); colonies on plates were counted and concentrations were reported as CFU 100 ml−1 or CFU 100 g−1 (wet weight) for water and sediment samples, respectively.

DNA extraction and PCR for microbial source tracking markers

For MST assays, the pH of water samples was lowered to 3·5 using 20% HCl. This procedure gives viral particles a net positive charge, promoting adsorption to the nitrocellulose membranes (McQuaig et al. 2006, 2009). Five hundred millilitre of water was filtered through 0·45-μm pore-sized nitrocellulose filters. Filters were stored at −20°C until processed. DNA from filters was extracted using the MoBio PowerSoil® DNA Isolation Kit following the manufacturer's instructions. Conventional PCR assays were carried out utilizing primer sets targeting the 16S rRNA gene of ruminant- and human-associated Bacteroides, using previously published protocols (Harwood et al. 2009). In addition, a previously developed PCR assay specific to the T-antigen of HPyVs was also performed (McQuaig et al. 2009). PCR products were visualized via gel electrophoresis. Quantitative PCR (qPCR) assays were carried out for the HF183 target as well as HPyVs (Bernhard and Field 2000; McQuaig et al. 2009). All samples, including method and extraction blanks, were assessed in duplicate. Results were expressed as gene copies 100 ml−1 of a water sample. The mean reaction efficiency for the HPyVs assay was 90·8%, and mean R2 of the standard curve was 0·989. The average reaction efficiency for the HF183 assay was 94·6%, and R2 of the standard curve was 0·991. A fresh standard curve was included with each plate of 96 reactions. Synthetic plasmids (Integrated DNA Technologies, Inc., San Jose, CA) with a target sequence for HPyVs or HF183 were used as standards (Harwood 2011; Staley et al. 2012), and the threshold of detection for each target was 10 gene copies 100−1 ml of water sample.

Data analysis

All bacterial concentrations were log10 transformed prior to statistical analyses. Differences in FIB concentrations were assessed by anova, with Tukey's post hoc tests where significant differences were determined. Two-tailed Pearson correlation analysis was used to identify significant relationships between the FIB concentrations, the presence of host-specific markers, total rain amount measured within 7 days prior to sampling, duration since last rain event (a day with total rainfall of 0·1 inches or more), and physical–chemical factors. In addition, a chi-square test for independence was performed to explore the association between the sampling sites and host-specific markers as determined by conventional and quantitative PCR. Finally, the coefficient of variation (CoV) was determined for water and sediment samples for each indicator organism. All results were considered significant at the α-level of 0·05.

Faecal coliform loading estimates

An estimate of faecal coliforms loads from failed septic systems was made using the following equation: L = 37·85×N×Q×C×F, where L is the daily load of faecal coliforms (CFU per day); N is the total number of septic tanks in the watershed (septic tanks); Q is the discharge rate for each septic tank; C is the concentration of faecal coliforms for the septic tank discharge; and F is the septic tank failure rate (US Environmental Protection Agency 2001). A commonly cited value for per capita wastewater production rate is 75 gallons per day per person and the average household size for Pasco County is about 2·66 people/household The commonly cited concentration (C) for septic tank discharge is 1 × 106 CFU 100 ml−1 for faecal coliforms (US Environmental Protection Agency 2001). A failure rate (F) of 5% was assumed based on information obtained from the Pasco County Department of Health on the number of septic tank repairs in the area.


Faecal indicator bacteria concentrations

FIB levels in water or sediments samples were not significantly different among the sites (Fig. 2); however, in the water column, there were significant differences by date for all FIB when data from sites NR3, NR2 and DNR3 were pooled (Fig. 3a). In sediments, the significant differences by date were observed only for enterococci (Fig. 3b). Data for all sites on each date are included in Fig. S1. One-time sample maxima criteria for recreational water (faecal coliforms 2·60 log10 100 ml−1, E. coli 2·37 log10 100 ml−1, enterococci 1·80 log10 100 ml−1) were exceeded in 62·5, 37·5 and 100% of sampling events for faecal coliforms, E. coli, and enterococci, respectively (Fig. 3a). The CoV was calculated over all sites and dates for each FIB for water and sediments. CoVs in water were 28% for both faecal coliforms and E. coli, and 24% for enterococci, while in sediments CoVs were 30% (faecal coliforms and E. coli) and 23% (enterococci) in sediments.

Figure 2.

Mean faecal indicator bacteria (FIB) concentrations in the water column (a) and sediments (b) for consistently sampled sites NR2, NR3, DNR3 and DNR1. FIB regulatory limits for recreational waters are depicted in panel a (Log10 CFU ml−1). Error bars represent standard deviation. (image Fecal coliforms; (imageEscherichia coli; (image enterococci; (image HPyVs; (image HF183; (image RumBac; (image FC-reg.liimits; (image E.c-reg.limits and (image Ent-reg.limits. HPyVs, human polyomaviruses.

Figure 3.

Mean faecal indicator bacteria (FIB) concentrations in the water column (a) and sediments (b) over the course of the study at consistently sampled sites NR2, NR3 and DNR3. FIB regulatory limits for recreational waters are depicted in panel a (Log10 CFU ml−1). The levels of culturable indicator organisms in columns that share lower case letter designation (including no letter) are not significantly different. Error bars represent standard deviation. F-HF183, H-HPyVs, R-ruminant bacteroides. (image Fecal coliforms; (imageEscherichia coli; (image) enterococci; (image FC limits; (image EC limits; (image ENT limits and (image mean flow.

Average FIB concentrations from composite soil samples collected on the banks of the upstream river branches were 4·3 and 2·4 log10 CFU 100 g−1 for composites 1 and 2 (north and south of the western branch corridor), and 2·75 log10 CFU 100 g−1 for composite 3 (eastern corridor). At DNR3, where box culverts were installed between the sampling events conducted in March and April, sediment layers in the immediate location were replaced. Significantly lower concentrations of each FIB were observed in sediments following the replacement (i.e. comparing FIB concentrations in October through March vs those in April through September). Faecal coliforms, E. coli and enterococci concentrations decreased by 86, 74 and 89%, respectively. No analogous reduction in FIB levels was observed in the water column. FIB levels at the other sites did not show an analogous drop in water or sediments.

Significant positive correlations were observed among all FIB in the water column (r2 = 0·73, 0·35 and 0·41, between faecal coliforms and E. coli, faecal coliforms and enterococci, and E. coli and enterococci, respectively). Similarly, FIB levels in the sediments were significantly correlated with one another (r2 = 0·69, 0·60 and 0·61 between faecal coliforms and E. coli, faecal coliforms and enterococci, and E. coli and enterococci, respectively).

In water samples, significant positive correlations were observed between E. coli and ionic strength (r2 = 0·42), E. coli and turbidity (r2 = 0·35), and between faecal coliforms and ionic strength (r2 = 0·36). Furthermore, significant negative correlation was seen between faecal coliforms and the duration since the last rain event (r2 = 0·57) in the water column as well as sediments (r2 = 0·40). Enterococci concentrations were not correlated with the physical/chemical variables.

Flow measurements were obtained at NR2, NR3 and DNR3 during three sample events in 2011 – January, April and July. The flow ranged between 0·07 and 0·13 cubic metres per second in January and July, and between 0·28 and 1·03 cubic metres per second in April (Table 2). Linear regression was performed to test for correlations with FIB concentrations and flow conditions. Significant negative correlation (r2 = 0·52) was observed between the flow rate and the concentration of faecal coliforms in the water column when data from the three sites were pooled. Data from a 2007 study (Florida Department of Environmental Protection 2008) that included these sites were combined with data from the current study. Observations from the pooled data set were sorted according to the presence or absence of flow in the river. Significantly greater faecal coliform and E. coli concentrations were observed under no-flow (stagnant) conditions compare with flowing conditions. Enterococci concentrations, however, did not differ based on flow conditions.

Table 2. Flow rates measured during the 2011 sample events at selected sites in cubic metres per second (m3 s−1)
Flow (m3 s−1)

Microbial source tracking and quantitative PCR

Markers associated with human sewage contamination (human-associated Bacteroides HF183 or HPyVs) were detected at all sites, except NR5 and NR6, at least once during the study period by both conventional and quantitative PCR. The conventional and qPCR methods for HF 183 and HPyVs were always in agreement (i.e. the presence or absence of the target). The HF183 marker was detected more frequently than HPyVs, which were detected in only two samples – NR2 and DNR1 in April – and these detections cooccurred with detection of HF183. The ruminant Bacteroides marker was detected at all sites except the site north of the cattle grazing area (NR6), the confluence of the stream branches at the marsh (NR4) and the southernmost site (DNR1) (Table 3).

Table 3. Distribution and detection frequency of microbial source tracking markers by PCR and range of concentrations by qPCR for host-associated markers
Binary PCRQuantitative PCR
Number of positive samples (%)Gene copies 100 ml−1
Site (Sampling events)HF183HPyVsRuminant Assoc. BacteroidesHuman Assoc. BacteroidesHPyVs
  1. HPyVs, human polyomaviruses.

NR3 (8)2 (29%)02 (29%)<10–128<10
NR2 (8)2 (29%)1 (14%)1 (14%)<10–160<10–135
DNR3 (8)5 (71%)01 (14%)<10–2360<10
DNR1 (3)1 (33%)1 (33%)0<10–132<10–167
NR7 (2)1 (50%)02 (100%)<10–193<10
NR6 (2)000<10<10
NR5 (2)001 (50%)<10<10
NR4 (2)1 (50%)00<10–452<10

Contingency analysis identified a significant association between the frequency of detection of HF183 determined by qPCR and individual sampling locations. The greatest frequency of detection and the largest quantities of HF183 were measured at DNR3 (Table 3). Furthermore, the detected quantities of the HF183 correlated with the levels of E. coli (r2 = 0·42).


The US Clean Water Act requires state agencies to establish a TMDL for each pollutant causing exceedance of water quality standards applicable to the designated use of the water body (US Environmental Protection Agency 2001). The TMDL for total and faecal coliforms in the New River watershed was established in 2004. During the data collection period, the levels of faecal coliforms exceeded water quality standards on 42·8% of occasions and the exceedances occurred under all flow conditions (Florida Department of Environmental Protection 2004). In the present study, the one-time sample maximum criteria for recreational waters were exceeded in 62·5, 37·5 and 100% of sampling events for faecal coliforms, E. coli, and enterococci respectively.

Sediments have been previously implicated as reservoirs for FIB in various studies (LaLiberte and Grimes 1982; Solo-Gabriele et al. 2000; Desmarais et al. 2002; Korajkic et al. 2009; Badgley et al. 2011). Here, FIB concentrations in 100 g of wet sediment for faecal coliforms, E. coli and enterococci were on average 88, 138 and 36 times higher than those found in 100 ml of water at all sites. Elevated levels of FIB in soil samples adjacent to the river, which remained very high after prolonged rainfall (>72 h), suggest that soils are also serving as a reservoir of FIB in this system. FIB can be deposited on the land surface and subsequently be conveyed to the stream during and after rainfall events (Fujioka et al. 1998; Anderson et al. 2005); in this case, cattle and feral pigs are implicated by MST markers and the abundance of faeces observed. Others have observed positive correlations between rainfall and FIB levels (Fujioka et al. 1998; Guber et al. 2006; Hill et al. 2006; Korajkic et al. 2009; Staley et al. 2012). Negative correlation between the levels of faecal coliforms in water column and sediments and duration because the last rain event has been documented during this study. It appears that the periodically inundated areas adjacent to subtropical rivers with intermittent flow create a favourable environment for persistence and possible growth of the FIB, which are then carried to the stream in run-off.

Compilation of historical data from the 2007 BMAP study (Florida Department of Environmental Protection 2008) and the data collected in this study showed that significantly lower concentrations of faecal coliforms and E. coli, but not enterococci, were found under flowing conditions compared to nonflowing conditions (Fig. 4). During nonflowing conditions, the river stream is reduced to nonconvective pools of water with abundant microbial life. However, upon the return of the water flow, the bacteria are distributed down the stream and their concentrations decline probably due to the large volume of water added to the waterway.

Figure 4.

Mean faecal indicator bacteria concentrations (log10 CFU 100 ml−1) at all sites under nonflowing vs flowing conditions. Error bars represent standard deviation. The numbers of samples (n) are shown at the bottom of the graph. The levels of culturable indicator organisms in columns that share a letter designation (including no letter) are not significantly different. (image Fecal coliforms; (imageEscherichia coli; (image enterococci; (image FC limits; (image EC limits and (image ENT limits.

The CoV revealed that fluctuation in enterococci levels in water and sediments was considerably smaller than that of faecal coliforms and E. coli, and, as previously mentioned, enterococci concentrations did not tend to vary with rainfall or flow conditions. Several enterococci strains, including a particular genotype of Enterococcus casseliflavus, can form naturalized resident populations in the Tampa Bay watershed (Badgley et al. 2010b). These findings call into question the validity of utilizing enterococci in water quality monitoring in subtropical, freshwater environments.

A MST study completed in 2008 as part of BMAP, utilizing conventional PCR and MST indicated likely sources of FIB at several New River sampling locations associated with on-site sewage treatment and disposal systems (OSTDS) (Florida Department of Environmental Protection 2008). Similarly, the detection of human sewage associated markers with conventional and quantitative PCR in the present study indicated that large capacity septic tanks (1500–2000 gals) with multiple sewage connections in mobile home and RV parks, many of them located within <100 feet from the stream, represent an important source of faecal pollution of New River.

The greatest quantities of the HF183 marker were detected at DNR3 site. Strong sewage odours were detected there on several sampling occasions. The HF183 marker was also detected upstream of the general sampling area in the east branch of New River (NR7) and the marsh confluence (NR4), where the great majority of the land is used for cattle grazing or serves as natural land that supports populations of wildlife. The HF183 Bacteroides marker has been previously classified as 98% human specific. Occasionally, this marker can be found in dogs, but there is no evidence of its presence in cattle or hogs (Ahmed et al. 2009). The drain field of a church OSTDS adjacent to the agricultural area was the suspected source of the human-associated pollution.

Human polyomaviruses were detected only at two sampling sites, NR2 and DNR1, and the detections cooccurred with human-associated Bacteroides. These results reinforce the conclusions in the 2008 BMAP implementation regarding OSTDS near NR2. Several OSTDS in the immediate area of this location that had been replaced or repaired over the years have been identified.

The ruminant-associated Bacteroides marker was detected only in the livestock grazing or wildlife areas. Over 15 000 acres within the New River basin contain areas devoted to grazing or which could be utilized by wildlife. There is evidence of a sizable population of feral hogs in the vicinity of sampling sites NR4, 5, 6 and 7, where cattle grazing also occurs. Therefore, agricultural and wildlife sources cannot be discounted as contributors to the faecal bacteria impairment in New River. While it is frequently impractical to attempt to ameliorate contributions from wildlife, reductions in contributions from agricultural related sources can be achieved through the implementation of the Best Management Practices (BMPs).

Based on the results of the MST analysis, FIB loading from OSTDS portrays a significant factor contributing to the impairment in New River. Even one failing OSTDS located in proximity to the stream may be a large and chronic source of bacterial contamination. An estimate of faecal coliform loading from contributing basin areas containing OSTDS adjacent to the stream was performed for the three continuously sampled sites. LIDAR topographic data available from Southwest Florida Water Management District (SWFWMD) were overlaid on 2009 aerial photography (SWFWMD 2004). Drainage basin data developed in the 2011 update to the Hillsborough River Watershed Management Plan were then added to the aerial photography to delineate the drainage basin boundaries (PBJS 2010). The calculated loading estimates were 4·2 × 1010, 2·0 × 1010 and 1·43 × 1010 CFU day−1 at sites NR3, NR2 and DNR3.

In conclusion, failing OSTDS and discharges from improperly functioning septic drain fields located in secondary basins are most likely the major contributors to the microbial loading in New River. However, numerous ponds, wetlands and drainage swales tend to mitigate the impact and attenuate FIB concentrations before they reach the river. Therefore, to understand the dynamics of water quality in given water body, sampling under both flowing and nonflowing conditions should be performed. MST analysis and qPCR enumeration of human-associated markers proved to be invaluable for the identification of point sources and origins of the faecal contamination. Finally, the stream sediments and the soil in the periodically inundated areas may serve as reservoir for FIB and must be, therefore, considered during remediation procedures.


The Pasco County (FL) Stormwater Management Division provided funding for this manuscript.