Concentrations of Pathogens and Indicators in Animal Feces in the Sydney Watershed

Abstract

A fecal analysis survey was undertaken to quantify animal inputs of pathogenic and indicator microorganisms in the temperate watersheds of Sydney, Australia. The feces from a range of domestic animals and wildlife were analyzed for the indicator bacteria fecal coliforms and Clostridium perfringens spores, the pathogenic protozoa Cryptosporidium and Giardia, and the enteric viruses adenovirus, enterovirus, and reovirus. Pathogen and fecal indicator concentrations were generally higher in domestic animal feces than in wildlife feces. Future studies to quantify potential pathogen risks in drinking-water watersheds should thus focus on quantifying pathogen loads from domestic animals and livestock rather than wildlife.

The current trend in minimizing pathogen health risks to water supplies is to use a risk management-based approach to ensure delivery of high-quality water. This approach utilizes multiple barriers within the water system, including effective control of contaminant inputs through watershed management. One potential source of these pathogens in drinking-water watersheds is the feces of domestic and wildlife animal populations. Pathogens from animal feces may enter waterways by direct deposition or as a result of overland runoff containing fecal material deposited in the watershed. To construct a source material budget of pathogen inputs, it is necessary to estimate the potential impact of animal populations on surface water quality (15). Subsequent analysis can then be performed to estimate the proportion of the source material that will be inactivated through natural decay and environmental stressors, how much may be transported to the stream network, and the proportion that represents a risk of human infection.

The initial requirement for the development of a source material budget is to estimate the concentration of potential pathogens in animal feces (shedding intensity). There is limited published information on the concentration and input load of potential waterborne pathogens and fecal indicator bacteria in wildlife (native and feral animal) populations (2, 3, 8, 19). Yet it is these animals that often have the greatest access to the riparian zones and reservoir surrounds in watersheds, since, by definition, their movements are largely uncontrolled, making surface water protection difficult. By comparison, data for potential pathogen concentrations in the feces of domestic animals are more abundant, particularly for the enumeration and prevalence of the protozoan parasites Cryptosporidium and Giardia spp. (4-7, 18, 24, 25).

This study was undertaken to provide a cross-sectional estimate of the intensity of shedding of pathogenic and indicator microorganisms in animal feces present in a large, semiprotected drinking-water watershed. The study quantified potentially pathogenic protozoa (Cryptosporidium and Giardia), enteric viruses (adenovirus, enterovirus, and reovirus), and indicator organisms (fecal coliforms and Clostridium perfringens spores) in fecal samples from watershed animals. The latter are the preferred bacterial indicators for freshwater systems and were chosen for their potential application as tracing and tracking organisms, with fecal coliforms also providing a comparison to previous studies.

MATERIALS AND METHODS

Sample collection.

Domestic animal fecal samples were collected from four watersheds within the Sydney Catchment Authority (SCA) area of operations. They were the Wollondilly, Braidwood, Upper Cox's, and Wingecarribee watersheds. These watersheds have mixed land use, with significant agricultural activities interspersed with urban and rural residential areas. Sheep and cattle grazing are the dominant agricultural land uses in the watershed. High sheep stocking rates are present in the Wollondilly watershed near Goulburn (approximately 550,000 sheep) and in the Shoalhaven watershed near Braidwood (approximately 109,000 sheep). Beef cattle grazing numbers are consistent throughout the watersheds at a stocking density of approximately 25% of the sheep population. Dairy farms are also present, predominantly within the Wingecarribee watershed (approximately 10,500 adult cattle). The distribution of animal populations across different age groups in these watersheds is unknown. We therefore aimed to estimate source loads for the overall population by collecting samples from a cross section of animal ages. The exception was adult cattle (>12 months of age) and calves (<12 months of age), where a clear differentiation of the two groups could be made. Samples were collected twice from each watershed during April and May 2002 (autumn), except for the Wingecarribee, which was sampled three times.

Each domestic animal fecal sample consisted of a composite (minimum of three scats) from adult cattle, calves, sheep, pigs, dogs, horses, poultry, and cats. Samples were collected aseptically in sterilized wide-mouth containers and stored at 4°C during transport and until analysis. The exception was samples for viral analysis, which were stored at −80°C prior to analysis. Following Animal Ethics approval, appropriately trained staff collected native and feral animal fecal samples. The majority of samples were collected from the Sydney University research facility Arthursleigh Farm, in the Wollondilly watershed, and from the SCA lands at Braidwood. Both of these areas are mixed-land-use agricultural areas, the latter interspersed with fragmented areas of native vegetation. These samples were collected in May and June 2002 (autumn to early winter). The native animals targeted included the Eastern gray kangaroo (Macropus giganteus), wombats (Vombatus ursinus), pademelon (Thylogale billardierii), swamp wallaby (Wallabia bicolor), brush-tail possum (Trichosurus vulpecula), platypus (Ornithorhynchus anatinus), common brown antechinus (Antechinus stuartii), wood duck (Chenonetta jubata), and a native rodent (Rattus fuscipes). Feral animals targeted included pigs (Sus scrofa), foxes (Vulpes vulpes), rabbits (Oryctolagus cuniculus), goats (Capra aegagrus hircus), deer (Odocoileus virginianus), carp (Cyprinus carpio), and cats (Felis catus). The majority of samples were processed individually; however, where the scat volume was too small for analysis, samples from the same species and subwatershed were occasionally combined to make a composite. Fecal samples from feral animals were collected directly from the rectum postmortem. Native animal samples were collected aseptically from the interior surfaces of Elliott traps or from the ground surface soon after defecation. Freshness of surface samples was ascertained by observation of defecation and the presence of a high moisture sheen.

Preparation of fecal samples for bacteriological analysis.

After samples were thoroughly mixed using a sterile tongue depressor, a 1-g aliquot of fecal material was weighed into 99 ml of sterile Ringer's solution. To assist in releasing the bacteria from the sediment and fecal material, each sample was sonicated for 30 s using a high-intensity ultrasonic processor (Vibra-Cell VCX 400; Sonics and Materials Inc., Newtown, CT) as described by Davies et al. (10). Following sonication, each sample was thoroughly shaken before subsampling into 50-ml volumes. When small volumes of fecal material were provided, a 0.5-g or 0.1-g aliquot was weighed into 99.5 ml or 99.9 ml of sterile Ringer's solution, respectively. The sonication probe was disinfected between samples by immersion in 12.5% sodium hypochlorite solution for 1 min, followed by a rinse with sterile deionized water. The probe was then immersed in sterile 26% sodium thiosulfate for 10 s, followed by a final rinse with sterile deionized water. To ensure that the cleaning process was effective, blank (sterile Ringer's solution) control samples were processed in each batch for any evidence of carryover during the sonication process.

Enumeration of bacteria.

Isolation and enumeration of fecal coliforms was carried out using standard membrane filtration methods as described in Standard Methods for the Examination of Water and Wastewater (1). Isolation and enumeration of C. perfringens spores was carried out using a standard membrane filtration method (30). In each case, serial dilutions were prepared from the sonicated solution (after it had been allowed to settle for 10 min), and 1 ml of each dilution was placed in 25 ml of sterile Ringer's solution and filtered through a 0.45-μm-pore-size cellulose ester membrane (Millipore, Sydney, Australia).

Fecal coliforms.

Fecal coliforms were isolated and enumerated by placing the membrane on an absorbent pad presoaked in mFC broth (without the addition of 1% rosolic acid in 0.2 N sodium hydroxide; Difco, Basingstoke, United Kingdom). Bacteria were resuscitated at 35°C for 2 to 4 h, followed by incubation for 18 to 22 h at 44.5°C. All blue colonies greater than 1 mm in diameter were counted and recorded as presumptive fecal coliforms. A maximum of six typical colonies from each sample were confirmed by inoculation into lauryl tryptose broth containing a Durham tube and tryptone water (Oxoid, Sydney, Australia). After incubation at 44.5°C for 24 h, the tubes were examined for growth and for gas and indole production. Colonies giving a positive lauryl tryptose broth result and a positive indole result were confirmed as fecal coliforms.

Clostridium perfringens.

For the isolation and enumeration of C. perfringens spores, the sample was heat shocked to 75°C for 10 min to kill vegetative cells and immediately cooled on ice. One milliliter of each serial dilution was filtered and the membrane placed on Perfringens Agar (Oxoid) supplemented with 4-methylumbelliferyl phosphate (Sigma, Sydney, Australia). Perfringens Agar plates were incubated in an anaerobic jar with an anaerobic gas-generating kit (Anaerogen; Oxoid) at 35°C for 18 to 24 h. Colonies were examined for fluorescence under long-wave UV light (366 nm), and all black or gray colonies that exhibited some degree of fluorescence or no fluorescence (atypical C. perfringens) were subcultured onto brain heart infusion agar (BHIA; Oxoid). The BHIA was incubated anaerobically at 35°C for 18 to 24 h; then colonies from the BHIA were inoculated both into tubes of lactose gelatin and into tubes of nitrate motility medium that had been preboiled for 10 min to remove oxygen. The tubes were incubated aerobically at 35°C for as long as 48 h for lactose gelatin tubes and 24 h for nitrate motility medium tubes. Colonies that produced gas and hydrolyzed gelatin in lactose gelatin medium, were nonmotile, and reduced nitrate in nitrate motility medium were confirmed as C. perfringens.

Comparison of bacterial enumeration with and without sonication.

Throughout the study, a sample from a batch was randomly selected and processed in parallel with the other samples but without sonication. Instead, samples were thoroughly mixed and shaken before serial dilutions were prepared and processed. This was done to assess whether the use of sonication damaged either the vegetative fecal coliform cells or the C. perfringens spores sufficiently to reduce recoveries. Laboratory quality control samples, including fecal material spiked with Escherichia coli and C. perfringens, as well as sterile dilution water, were processed for every batch.

The differences in recovery efficiency for fecal coliforms and C. perfringens from sonicated versus nonsonicated samples were analyzed using the Wilcoxon signed-rank test (21). P values of <0.05 were considered significant.

Recovery and staining method for Cryptosporidium and Giardia.

A 0.5-g aliquot of the fecal sample was weighed into a sterile 50-ml centrifuge tube. Diethyl ether was used in dispersal and defatting of calf fecal samples as described by Davies et al. (9). To quantify recovery efficiency, each individual sample was seeded with ColorSeed (Biotechnology Frontiers Ltd. [BTF], Sydney, Australia) by vortex mixing a ColorSeed vial for 30 s and adding the contents to the fecal sample. The ColorSeed vial was rinsed twice with approximately 4 ml elution buffer (34), and each time the vial was vortex mixed and the contents added to the fecal aliquot. The seeded fecal suspension was refrigerated overnight at 4°C. The dispersal of the fecal material was initiated by addition of 20 ml 0.002 M sodium pyrophosphate and vortex mixing for 2 min. After incubation at room temperature for 30 min, the fecal suspension was centrifuged at 3,000 × g for 10 min at 4°C. Using vacuum aspiration, all but approximately 7 ml of the supernatant was removed and the pellet resuspended by vortex mixing. The pellet was redispersed by sieving through a metal mesh (area, approximately 2.5 cm²; pore size, approximately 1.5 mm) into a sterile 50-ml centrifuge tube. The original sample tube was rinsed with sterile deionized water, and these rinsings were passed through the sieve. The combined sieved material was stirred with a sterile wooden stick and the stick rinsed into the tube. The volume was made up to 50 ml with sterile deionized water and centrifuged at 3,000 × g for 10 min at 4°C. The supernatant was vacuum aspirated to leave approximately 5 ml in the tube. The tube was vortex mixed to resuspend the pellet, and the material was transferred to an L10 tube (Dynal, Oslo, Norway). The 50-ml tube was rinsed with 3 ml sterile deionized water, vortex mixed, and the rinse transferred to the L10 tube. Rinsing was repeated with a further 2 ml of sterile deionized water.

To facilitate the detection of low concentrations of (oo)cysts and to maximize specificity, immunomagnetic separation was used, employing the Dynal GC Combo kit according to the manufacturer's instructions, with the following modifications. Following the transfer of material from the L10 tube to a 1.5-ml tube, the supernatant from the 1.5-ml tube was used to rinse the L10 tube and then transferred back to the 1.5-ml tube. In addition, the liquid containing the dissociated cysts and oocysts was not neutralized with sodium hydroxide if samples were stained on the same day.

Dissociation was performed according to the Dynal GC Combo kit instructions, with the following modifications. With the microcentrifuge tubes in the Dynal magnetic apparatus (MPC-M) and the magnetic strip in place, the tubes were rested horizontally for 10 s to allow the magnetic beads to attach to the MPC-M. The supernatant was discarded carefully, so as not to disturb the bead matrix. The centrifuge tubes were uncapped, and 300 μl of 0.1 M HCl was added to each. After recapping, each tube was vortex mixed for 10 s and incubated at room temperature for 10 min in a vertical position. The tubes were again vortex mixed for 10 s, and each tube was tapped at the base to dislodge any remaining pellet. The magnetic strip was replaced in the MPC-M, and the tubes were repositioned in the MPC-M. The tubes were allowed to rest horizontally for 10 s. With care taken not to disturb the bead matrix, the acid wash was transferred to a 13-mm membrane (pore size, 0.8 μm; Millipore, Sydney, Australia) positioned on a vacuum apparatus with the vacuum on.

To stain the Cryptosporidium oocysts and Giardia cysts, 3 drops of cold methanol was applied to the surface of each membrane and allowed to stand for 1 min. The excess fluid was aspirated under a vacuum, and the membrane surface was washed with 1 ml of methanol. The membrane was allowed to dry for 5 s, overlaid with 80 μl of 4′,6′-diamidino-2-phenylindole (DAPI; 0.8 μg.ml⁻¹), and allowed to stand for 2 min. The entire membrane surface was washed with 200 μl of wash buffer (BTF), which was simultaneously aspirated under a vacuum. Eighty microliters of Easystain (BTF) was applied to the membrane and allowed to incubate for 15 min. The fluid was then aspirated under a vacuum. The membrane surface was washed with 200 μl wash buffer (BTF) while vacuum aspirating. The membrane was transferred to a microscope slide with forceps and overlaid with mounting medium (BTF), and a coverslip was applied.

The slide-mounted membranes were examined with an Axioskop epifluorescence microscope (Zeiss, Germany) using filter set 09 (blue light excitation) for Easystain (BTF), filter set 02 (UV light excitation) for DAPI staining, and filter set 15 (green light excitation) for ColorSeed (BTF). The membrane was initially scanned at a magnification of ×200 for appropriate objects stained with Easystain, which were further examined at a higher magnification when necessary to ensure correct identification. The identification criteria described in U.S. EPA method 1623 (33) were used for Easystain-labeled and DAPI-stained objects.

Each raw result was adjusted for the recovery efficiency of the individual sample matrix by using the following calculation:

where NI is the number of ColorSeed inoculated into the sample and NR is the number of ColorSeed recovered from the sample.

Virus analysis.

Fecal samples were diluted by suspending 2 g of homogenized feces in 20 ml of phosphate-buffered saline. A 10-ml aliquot of the fecal suspension was extracted with an equal volume of carbon tetrachloride, to remove lipids. Six milliliters of the extracted fecal sample was inoculated onto cell culture after being passed through a sterile 0.45-μm syringe filter to remove bacteria and fungi. Confluent cell cultures (LLC-MK₂, MRC-5, and HEp-2) were prepared in cell culture tubes (Corning, Crown Scientific, Sydney, Australia). These cell lines are sensitive for the detection of reoviruses, many adenoviruses, and enteroviruses including polioviruses, coxsackieviruses, and echoviruses. One milliliter was inoculated into two tubes of each cell culture, which was equivalent to 0.6 g of the original sample. Due to the cellular toxicity of samples from herbivorous animals, a 1:10 dilution of sample was also inoculated, equivalent to 0.06 g of the original sample. Inoculated cell cultures were incubated at 37°C for 90 min to allow viral adsorption to the cells. Following this incubation, the inoculum was removed from the cell culture and replaced with fresh medium to minimize sample toxicity, and cultures were incubated at 37°C. The cultures were examined microscopically at least twice a week for a period of 28 days, or until a viral cytopathic effect (CPE) was evident. Enterovirus, reovirus, and adenovirus identification was achieved by observing the specific appearance of CPE on the various cell lines. All viral isolates were confirmed by subpassaging.

RESULTS

Fecal coliforms.

All pooled fecal samples from domestic animals contained high levels of fecal coliforms (Table 1). The highest median and maximum concentrations from the domestic animal samples were 1.1 × 10⁸ CFU · g⁻¹ and 9.5 × 10⁸ CFU · g⁻¹, respectively, from poultry fecal samples. Feral and native animals also had high median concentrations, with the majority ranging between 8.1 × 10³ and 9.6 × 10⁶ CFU · g⁻¹. The highest concentrations were recorded for samples from a feral pig (4.9 × 10⁹ CFU · g⁻¹), a feral goat (3.4 × 10⁹ CFU · g⁻¹), and a kangaroo (1.1 × 10⁹ CFU · g⁻¹). The exceptions were antechinus, platypus, and carp, which had median concentrations of <100 CFU · g⁻¹.

TABLE 1.

Fecal coliform and C. perfringens spore concentrations in animal fecal samples

Animal	Fecal coliforms			C. perfringens spores
	No. of positive samples/total samples	Concn (CFU · g−1 [wet wt])		No. of positive samples/total samples	Concn (CFU · g−1 [wet wt])
		Median	Range		Median	Range
Domestic animals
Calf	9/9	3.0 × 106	5.8 × 105-8.3 × 108	2/9	<100	<100-6.6 × 105
Adult cattle	9/9	1.8 × 105	1.3 × 103-8.5 × 106	2/9	<100	<100-500
Sheep	9/9	6.6 × 105	1.0 × 105-1.9 × 108	2/9	<100	<100-7.2 × 106
Horse	9/9	3.8 × 104	4.0 × 103-3.3 × 106	3/9	<100	<100-100
Pig	9/9	7.1 × 106	5.8 × 105-4.1 × 108	9/9	2.9 × 105	200-1.8 × 107
Poultry	9/9	1.1 × 108	1.6 × 106-9.5 × 108	8/9	4.6 × 103	<100-4.3 × 106
Cat	8/8	2.3 × 106	3.3 × 104-4.1 × 107	8/8	3.3 × 106	7.9 × 105-6.9 × 107
Dog	9/9	3.1 × 107	8.4 × 106-1.2 × 108	8/9	3.6 × 105	<100-5.7 × 107
Native wildlife
Antechinus	3/12	<100	<100-5.0 × 103	0/12	<100	<100-<1.0 × 103
Brushtail possum	2/2	1.6 × 105	1.4 × 105-1.8 × 108	0/2	<100	<100-<100
Wood duck	8/9	8.1 × 103	<100-6.9 × 104	4/9	<100	<100-8.0 × 104
Kangaroo	11/11	2.1 × 106	7.5 × 103-1.1 × 109	1/11	<100	<100-600
Platypus	4/11	<100	<100-4.6 × 104	0/11	<100	<100-<100
Wallabya	10/10	6.5 × 105	600-2.9 × 107	0/10	<100	<100-<100
Rat	5/5	2.1 × 103	1.2 × 103-8.0 × 104	0/5	<100	<100-<100
Wombat	7/7	4.0 × 103	100-5.7 × 104	2/7	<100	<100-100
Feral wildlife
Carp	0/1	<100		0/1	<100
Deer	1/1	2.2 × 106		0/1	<100
Fox	1/1	9.6 × 106		1/1	4.3 × 103
Goat	5/5	1.4 × 106	4.6 × 104-3.4 × 109	5/5	<100	<100-<100
Rabbit	3/4	5.0 × 105	<200-1.0 × 106	0/4	<100	<100-<100
Feral cat	2/2	6.9 × 106	2.8 × 106-1.1 × 107	2/2	3.4 × 106	2.3 × 106-4.4 × 106
Feral pig	5/5	4.1 × 104	1.0 × 103-4.9 × 109	0/5	<100	<100-<100

^aIncludes swamp wallaby and wallaroo samples.

Clostridium perfringens.

Concentrations of C. perfringens spores in pooled animal fecal samples from domestic animals were highly variable (Table 1), with the majority of horse, calf, sheep, and adult cattle samples being either equal to or less than the detection limit of 100 CFU · g⁻¹. The domestic and feral cat samples had the highest median concentrations of C. perfringens (3.3 × 10⁶ CFU · g⁻¹ and 3.4 × 10⁶ CFU · g⁻¹, respectively), followed by the domestic pig (2.9 × 10⁶ CFU · g⁻¹). Concentrations of C. perfringens spores in native and feral animal feces were low, frequently below the detection limit of 100 CFU · g⁻¹. The exceptions were the feral cat, wood duck, and fox samples and one sample of kangaroo feces.

Fecal coliform and C. perfringens recovery with and without sonication.

The recovery of fecal coliforms was significantly greater when sonication was used (P = 0.02; n = 12) than when no sonication was used. There were only two occasions, involving one dog and one calf fecal sample, when recoveries using sonication were less than those obtained with no sonication. There was no significant difference in C. perfringens recoveries (P = 0.08; n = 12) between treatment with and without sonication.

Cryptosporidium and Giardia.

Cryptosporidium oocysts were detected in at least one composite sample of the feces from each of the domestic animal species, with the exception of poultry, where no oocysts were detected (Table 2). Median concentrations, however, were zero for most domestic species, except pigs (367 oocysts · g⁻¹), sheep (17 oocysts · g⁻¹), and calves (9 oocysts · g⁻¹) (Table 2). Among native animals, kangaroos (4/11 samples), and possum (2/2 samples) were positive for Cryptosporidium oocysts. Median values for the native and feral animals were also zero with the exception of the possum feces (54 oocysts · g⁻¹), deer feces (6 oocysts · g⁻¹), and rabbit feces (38 oocysts · g⁻¹). Fewer samples were collected from feral animals than from domestic and native animals due to the difficulty of obtaining samples.

TABLE 2.

Cryptosporidium and Giardia concentrations in animal fecal samples

Animal	Cryptosporidium				Giardia
	No. of positive samples/total samples	Concn (oocysts · g−1 [wet wt])		ColorSeed recovery (%)	No. of positive samples/total samples	Concn (cysts · g−1 [wet wt])		ColorSeed recovery (%)
		Median	Range			Median	Range
Domestic animals
Calf	4/7	9	0-183	44-85	7/9	133	0-533	3-40
Adult cattle	2/9	0	0-10	42-87	7/9	68	0-293	37-80
Sheep	6/9	17	0->6,897	22-86	6/9	26	0-504	13-73
Horse	1/9	0	0-5	20-82	2/9	0	0-8	45-81
Pig	7/9	367	0-6,000	1-49	5/9	11	0->1.6 × 104	2-61
Poultry	0/7	0	0-0	1-32	2/9	0	0-67	2-70
Cat	3/7	0	0-17	59-93	1/7	0	0->7,143	8-58
Dog	2/8	0	0->5,000	5-91	5/8	835	0->6,061	6-77
Native wildlife
Antechinus	0/3	0	0-0	8-11	0/8	0	0-0	3-51
Brushtail possum	2/2	54	20-89	41-65	0/2	0	0-0	40-41
Wood duck	0/9	0	0-0	27-68	4/9	0	0-339	38-77
Kangaroo	4/11	0	0-1,257	14-81	0/11	0	0-0	29-78
Platypus	0/6	0	0-0	49-85	0/6	0	0-0	54-100
Wallabya	0/10	0	0-0	32-87	0/10	0	0-0	17-81
Rat	0/4	0	0-0	29-69	0/4	0	0-0	30-70
Wombat	0/5	0	0-0	16-61	0/5	0	0-0	25-73
Feral wildlife
Carp	0/1	0		56	0/1	0		33
Deer	1/1	6		31	0/1	0		47
Fox					1/1	1.5 × 104		4
Goat	0/3	0	0-0	16-27	0/3	0	0-0	34-58
Rabbit	1/2	38	0-77	13-33	0/2	0	0-0	19-63
Feral cat	0/1	0		51	0/1	0		61
Feral pig	0/5	0	0-0	1-79	0/5	0	0-0	1-47

^aIncludes swamp wallaby and wallaroo samples.

Giardia cysts were present in at least one sample of the feces from each of the domestic animal species (Table 2). The highest maximum concentrations were recorded for samples from domestic cats (>7,143 cysts · g⁻¹), dogs (>6,061 cysts · g⁻¹), pigs (>16,667 cysts · g⁻¹), and sheep (504 cysts · g⁻¹). In contrast to the Cryptosporidium oocyst isolation, median values in the domestic animals were greater than zero for the majority of animal species (Table 2). Giardia cysts were less common in native and feral animal feces than Cryptosporidium oocysts, with Giardia cysts detected only in the wood duck (4/9) and fox (1/1) samples.

Viruses.

Infectious enteric viruses were detected predominantly in the domestic animal fecal samples (Table 3). Viruses were detected in 4/5 calf feces tested, with the majority of samples being positive for reovirus. One calf feces sample tested positive for enterovirus, which was subsequently confirmed as a strain of bovine enterovirus. Similarly, 4/6 adult cattle fecal samples were positive for virus; however, in this case only reovirus was isolated. Only one fecal sample from pigs was positive for virus (reovirus). No viruses were isolated from domestic cats, dogs, horses, poultry, or sheep. Reovirus was detected in one antechinus fecal specimen (Table 3). No viable virus was detected in any of the other native or feral animal fecal samples: carp (n = 1), deer (n = 1), possum (n = 2), platypus (n = 4), rat (n = 4), wombat (n = 7), duck (n = 9), wallaby/wallaroo (n = 10), kangaroo (n = 11), feral cat (n = 1), fox (n = 1), feral rabbit (n = 2), feral goat (n = 5), or feral pig (n = 5).

TABLE 3.

Summary of virus data for animal fecal samples

Species	Total no. of samples	No. of samples positive for:
		Adenovirus	Reovirus	Enterovirus
Domestic animals
Pig	4	0	1	0
Calf	5	0	3	1
Adult cattle	6	0	4	0
Native wildlife, antechinus	4	0	1	0

DISCUSSION

Concentrations of C. perfringens spores, Cryptosporidium, Giardia, and enteric viruses were higher in the feces of domestic animals than in the feces of wildlife animals in the three watersheds investigated. This indicates that the emphasis of future studies and risk management strategies for these watersheds should be placed on domestic animal management. These results are congruent with those of a Canadian study by Heitman et al. (20) that found that concentrations of Cryptosporidium oocysts and Giardia cysts in wildlife feces were significantly lower than those in domestic livestock or human effluent.

Point prevalence studies, which examine pathogen prevalence in a given population at a particular point in time, are relatively common (12, 13, 26, 29, 35); fewer studies have quantified pathogen excretion rates (27, 31). Although prevalence studies are useful for disease management, these types of studies are insufficient for risk assessment. Both the prevalence and the concentration of pathogens in feces are required to establish pathogen source loads in watersheds and hence to assess the risk from fecal contamination of source waters. Estimation of source loads is further complicated because it is difficult to compare the data from different studies due to differences in methodologies, pooling of fecal samples, seasonality, animal age, herd immunity, and temporal variation. For example, Power et al. (28) showed that the proportion of Cryptosporidium infections in Eastern gray kangaroos that could not be detected by normal clinical methods contributed a substantial load of oocysts to the watershed. For this reason we used a sensitive detection method capable of quantifying low concentrations of Cryptosporidium oocysts in animal feces and included both wildlife and domestic animals, focusing on those species that were most abundant in the watershed and that excreted the greatest volume of manure.

Coliforms and C. perfringens spores are typically used as indicators of fecal pollution of receiving waters. In the present study, C. perfringens was isolated mostly from domestic animal feces and was rarely detected in wildlife animal feces. This suggests that C. perfringens spores could be a useful indicator of fecal inputs from agricultural and urban development in watersheds. This was not the case for fecal coliforms, which were isolated from the majority of the fecal samples examined whether from domestic or wildlife animals and at high concentrations in most cases. The levels of fecal coliforms in feces in this study were substantially higher than pathogen levels. Mathematical models that rely on fecal coliforms as surrogates for the estimation of pathogen loads (16, 32) could therefore result in overestimation of risk.

Like the bacterial indicators and protozoa, viruses were isolated more frequently from the domestic animal population than from either the native or the feral population, albeit only from calves, cattle, and pigs. Of the enteric viruses tested, reoviruses were isolated most often. The isolation of reovirus from the animal populations was not surprising, given that these viruses are ubiquitous; the only three known serotypes (17) infect a range of vertebrates including cattle, sheep, swine, and other mammals.

A comparison of the effect of sonication treatment versus no sonication treatment on the recoveries of fecal coliforms and C. perfringens showed that the sonication method gave equivalent or higher counts for the majority of samples tested and would therefore be recommended as the method of choice for the isolation of these microorganisms from feces. There was no evidence that sonication resulted in substantial damage to either vegetative bacteria or spores. These findings are in agreement with those of previous studies on the effectiveness of sonication for enumerating bacteria from sediments (11, 14).

The sampling period for this project covered the late fall and early winter of 2002, during a period of prolonged drought in the watershed. None of the samplings were impacted by rainfall of any significance, making the results primarily a dry-weather, cross-sectional estimate of the intensity of microbial shedding from animals within the watershed. Since a study by Power et al. (28) identified a strong seasonal effect on the levels of Cryptosporidium in kangaroo feces in this watershed, the intention of future studies will be to investigate the seasonal and other temporal effects on pathogen input loads, particularly from domestic animals.

Estimation of pathogen source loads is an essential first step in the development of a predictive mathematical model of pathogen origin, fate, and transport in watersheds. The application of molecular genotyping techniques (22, 23) and infectivity assessments to a wider range of pathogens isolated from fecal samples may also refine the human health risk associated with pathogen loads in watersheds, at least for the Cryptosporidium and Giardia isolates. Notwithstanding the isolation of potential human pathogens in this study, there have been no reported waterborne disease outbreaks in the population served by drinking water from this watershed.