Tfjd300-04-135171.tex

Journal of Environmental Science and Health Part A, 41:173–184, 2006Copyright C Taylor & Francis Group, LLCISSN: 1093-4529 (Print); 1532-4117 (Online)DOI: 10.1080/10934520500351884 Fecal Contamination ofAgricultural Soils Before andAfter Hurricane-AssociatedFlooding in North Carolina Michael J. Casteel,1 Mark D. Sobsey,1 and J. Paul Mueller2 1Department of Environmental Sciences and Engineering, School of Public Health,University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA2Department of Crop Sciences, North Carolina State University, Raleigh, NorthCarolina, USA Hurricane Floyd and other storms in 1999 caused widespread and extensive floodingof eastern North Carolina and environmental contamination with fecal wastes frommunicipal wastewater and livestock operations. Because wastewater contains high lev-els of pathogenic micro-organisms, principal health risks to humans from flooding areconsumption of crops grown in fecally contaminated soil and ingestion of contaminatedwater. Flood waters polluted with microbial and other contaminants also may be detri-mental to the health of livestock and plant crops. In the present study, agriculturalsoils impacted by flood waters were analyzed for bacterial and viral indicators of fe-cal contamination. Total coliforms, fecal coliforms, Escherichia coli, spores of Clostrid-ium perfringens, and both male specific (F+) and somatic coliphages were recoveredfrom soil and assayed in liquid culture media. A number of samples were positive forthe presence of fecal coliforms, E. coli, and coliphages, indicating the presence of hu-man or animal feces. Most samples were positive for total coliforms, and almost allsamples contained high levels of Cl. perfringens spores. The levels of Cl. perfringensspores were significantly (P < 0.001) higher in flooded soil (post-Hurricane Floyd) com-pared to pre-flood soil. Persistent fecal contamination of soil, as demonstrated by thehigh levels of Cl. perfringens spores, suggests the need for additional or alternativemeasures to protect crop-growing areas, including prospective microbiological monitor-ing and improved protection of watersheds from incidents capable of releasing fecalmaterial.
Key Words: Agriculture; Fecal contamination; Flooding; Hurricanes; Microbial indica-tors; Soil.
Received April 29, 2005.
Address correspondence to Michael J. Casteel, Water Quality Laboratory, San FranciscoPublic Utilities Commission, 1000 El Camino Real, Millbrae, California, 94030, USA;E-mail: [email protected] In a 6-week period during 1999, North Carolina experienced a series of severeweather events in the form of hurricanes: Hurricanes Dennis (3–7 September1999), Floyd (14–17 September 1999), and Irene (17–18 October 1999). Hur-ricanes, which often reach land on the eastern Atlantic and in Gulf Coasts ofthe United States during the months of June to November, are accompanied bylarge waves, storm surges, and elevated surface water levels. Severe floodingof large land areas often occurs, with subsequent floodwater contamination ofboth surface waters and groundwaters. During Hurricane Floyd, wind speedsexceeded 160 km/h, and precipitation in parts of eastern North Carolina was inexcess of 51 cm. River waters rose more than 12 m above flood levels in much ofthe piedmont and coastal plain regions of the state. Hurricane Dennis, with atotal rainfall of up to 25 cm, had contributed to saturated soil conditions beforeHurricane Floyd, while Hurricane Irene contributed another 10 to 15 cm of rainto flooded areas after Hurricane Floyd.
During Hurricane Floyd, surface waters and large areas of land became con- taminated with fecal and chemical wastes from compromised septic systems,municipal sewage systems, and livestock waste lagoons. Areas affected by theflooding included the Center for Environmental Farming Systems (CEFS), lo-cated at the Cherry Farm near Goldsboro, North Carolina, and bordered by theNeuse River. As a result of Hurricane Floyd, a large proportion of the CEFSwas inundated with flood water for an extended period of time. Because fecalwastes and wastewater from human and animal sources contains high levelsof pathogenic microorganisms, health hazards to humans from flooding can oc-cur by consumption of crops grown in fecally contaminated soil or ingestion offecally contaminated water. Furthermore, flood waters polluted with microbialand other contaminants may be detrimental to the health of livestock and plantcrops.
Little or no information is available on the microbiological quality of agri- cultural and other soils before and after flooding with fecally contaminatedwater. Such data would allow farmers, scientists, and state officials to assessand better control potential risks to human and animal health when floodingoccurs as a result of severe weather conditions. The purpose of this study was todetermine the sanitary microbiological quality of agricultural soil impacted byhurricane-associated flooding based on assays of microbial indicators of fecalcontamination in soil samples collected before and after Hurricane Floyd.
Study Area, Soil Sample Collection, and Soil Processing The study was conducted at CEFS located in Goldsboro, North Carolina (approximate latitude/longitude = 35◦23 30 /−78◦1 41 ). The CEFS, which is Hurricane Impact on Agricultural Soils in North Carolina Table 1: Description of soil samples† and soil moisture content (% ± standarddeviation (SD)).
content (%) ± SD
†Soils analyzed in this study were loamy sands, belonging to the Wickham series (thermic typicHapludults) or the Tarboro series (thermic Typic Udipsamments).
††Mean values are representative of all samples; soil moisture contents did not differ signifi-cantly between series.
# Wickham series comprised 66% (n = 19) and 64% (n = 18) of the total number of samples for 3/1/99 and 10/25/99, respectively. Tarboro series comprised 34% (n = 10), 36% (n = 10), of the total number of samples for 3/1/99 and 10/25/99, respectively.
§These soils had not been impacted by floods during the hurricane season of 1998.
§§Samples had been collected in the aftermath of three major hurricanes: Hurricane Dennis(September 3–7, 1999), Hurricane Floyd (September 14–17, 1999), and Hurricane Irene(October 17–18, 1999).
bordered by the Neuse River to the south, is a partnership between the N.C.
Department of Agriculture and Consumer Services, North Carolina State Uni-versity, North Carolina Agricultural and Technical State University, and otherstate and federal agencies. The CEFS is principally used for research, demon-strations, and education, and accommodates a diverse mix of field and hor-ticultural crops. The soils analyzed in this study had been used to cultivateimportant North Carolina agricultural commodities such as corn, soybeans,cotton, peanuts and wheat.
Soil sampling was conducted on 1 March 1999 and on 25 October 1999 (Table 1). Samples were collected from the surface (≤15 cm depth) into sterilebags at the same points on each date using a global positioning system andgeo-referenced points, transported to the laboratory in coolers, and stored at−70◦C until analyzed. Soil moisture contents were determined by drying rep-resentative portions of each sample in a muffle oven at 105◦C for 3 h followedby cooling in a desiccator. For microbiological analyses, 10 g of each sample wastransferred into a milk dilution bottle, followed by addition of 95 mL of 3% beefextract/0.01% Tween-80 (pH 7), mechanical shaking for 30 min at 100 RPM,and serial 10-fold dilution in 0.2 M phosphate buffered saline (PBS; pH 7.2).
Soil Analyses for Total Coliforms, Fecal Coliforms, and E. coli For analysis by the American Public Health Association multiple fermen- tation tube method,[1] 1 mL aliquots of the diluted soil samples were incubatedin 10 mL of lauryl tryptose broth (LTB) in 16 × 125 mm test tubes, each con-taining an inverted Durham tube, at 37◦C for 48 h. Tubes were examined forgrowth and gas (presence of both = presumptive positive for (total) coliforms)at 24 and then 48 hours. Aliquots (0.05 mL) from all positive LTB cultureswere then dispensed into 8 mL of EC media containing 4-methylumbelliferyl-β-D-glucuronide (0.01%) in 13 × 100 mm borosilicate glass test tubes, each containing an inverted Durham tube, capped loosely, and incubated at 44.5◦Covernight. Positive and negative controls were included, consisting of a controlfor fecal coliforms (Klebsiella pneumoniae), E. coli (E. coli B), and a negativecontrol (media only). Tubes were examined for the presence of growth and gasand were checked for blue fluorescence under a handheld, long-wavelengthultraviolet lamp. The presence of both growth and gas were considered confir-matory for fecal coliforms, while the presence of growth, gas, and fluorescencewere considered confirmatory for the presence of E. coli. Presence-absence datawere converted to Most Probable Number (MPN) values, with the correspond-ing upper and lower 95% confidence limits, using standard tables.[1] Resultsare reported as log10 total coliforms, fecal coliforms, or E. coli/100 g dry soil.
Soil Analyses for Cl. perfringens Spores Portions of the diluted soil were heated at 65◦C for 20 minutes for the destruction of vegetative cells. Heated samples were assayed for the pres-ence of Cl. perfringens spores using a modification of the previously describediron-milk method (IMM).[2] In the modified method, 1 gram of ferrous sulfate(FeSO4 · 7H2O) was dissolved in 50 mL of sterile water, and sterile water wasadded to the contents of one can (ca. 350 mL) of evaporated milk and adjustedto a final volume of 500 mL. Then, both materials were added to 450 mL ofsterile water for a final volume of 1 L of IMM. Ten 10 mL portions of the IMMdispensed in 16 × 125 mm test tubes were inoculated with 1 mL of the heatedsamples, capped, and then incubated at 41.5◦C for up to 3 days. Samples werescored positive or negative for stormy fermentation. The presence of Cl. perfrin-gens in samples positive for stormy fermentation was confirmed by anaerobicincubation of a small volume of the culture on mCp agar (Accumedia, Balti-more, MD) followed by exposure of yellow colonies to concentrated ammoniumhydroxide fumes for 20 seconds. Red or dark pink colonies (exhibiting acidphosphatase cleavage of phenolphthalein diphosphate) were counted as pre-sumptive Cl. perfringens. Presence-absence data were converted to MPN values,with the corresponding upper and lower 95% confidence limits, using standardtables.[1] Results are reported as log10 MPN Cl. perfringens spores/100 g drysoil.
Soil Analyses for Male-Specific and Somatic Coliphages Coliphages were detected and enumerated using a quantitative modifica- tion of an enrichment and presence/absence technique.[3] Assay of F+ coliphagesused host E. coli Famp (ATCC #700891) while assay of somatic phages used hostE. coli CN13 (ATCC #700609). Tryptic soy broth (TSB) media with antibioticsspecific for each host were prepared as specified.[3] Bacterial hosts were propa-gated in TSB for 12 hours, and 0.05 mL aliquots of the 12-hour cultures wereinoculated into 50 mL of fresh TSB followed by incubation for 6 hours to produce Hurricane Impact on Agricultural Soils in North Carolina log-phase growth. Agar plates containing 0.75% tryptic soy agar medium (e.g.,regular strength TSB plus 0.75% bacteriological grade agar) and host bacteriawere used as spot plates. Agar was autoclaved, allowed to cool, the 6-hour hostculture of E. coli Famp or E. coli CN13 was added to a final concentration of3% (V/V), and the mixtures were swirled gently and poured into plastic petridishes.
For coliphage analysis, 10 g and 1 g portions of each soil sample were weighed, in triplicate, into separate 50 mL polypropylene centrifuge tubes. Toeach tube was added 45 mL of a medium containing 3% TSB, 0.38% glycine,0.3% Tween-80, 0.1% glucose, 0.03% calcium chloride and 0.015% magnesiumsulfate. For somatic phage enrichment, the tubes were also supplemented withnalidixic acid (0.01%) and log phase E. coli CN13 (4%). For male-specific col-iphage enrichment, the tubes were supplemented with streptomycin sulfate(0.015%), ampicillin (0.015%), and log phase E. coli Famp (4%). Tubes were in-verted several times to mix the soil with the medium without introducing airbubbles, and then incubated overnight at 37◦C with shaking at 100 RPM. TenµL (0.01 mL) aliquots of an enriched sample were pipetted onto spot platesfor both bacterial hosts. The spots were allowed to dry, the plates were incu-bated at 37◦C overnight, and lysis zone formation was scored as positive forthe presence of either somatic or F+ coliphages according to host. Data for spotplate-confirmed positive and negative enrichment cultures were converted toMPN values using the Thomas equation.[4] Results are reported as log10 MPNsomatic or male-specific coliphages/100 g dry soil.
The average log10 MPN levels of microbes per 100 g dry soil and their corresponding 95% confidence intervals were calculated. The data were thenexamined for statistically significant differences (α = 0.05) between microbiallevels in soil by date using one-way analysis of variance (ANOVA) followed bythe Tukey–Kramer multiple comparisons test (InStat v3.0, Graphpad Software,San Diego, CA).
Fifty-seven soil samples were analyzed in the present study, with 29 samplescollected during 3/1/99 and 28 samples collected during 10/25/99 (Table 1).
Soil samples collected 3/1/99 had not been impacted by flooding of the pre-ceding hurricane season, and samples collected 10/25/99 had been collectedapproximately 7 days following the extended flooding at the CEFS by Hur-ricanes Dennis, Floyd, and Irene. All soil types were loamy sand, and theirofficial series are reported along with Table 1. As shown in Table 1, the av-erage soil moisture content (% by weight) for soil samples collected during 3/1/99 and 10/25/99 was 14% (range, 5–25%), and 12% (range, 5–17%),respectively.
Results for total coliforms, fecal coliforms, and E. coli are shown in Table 2.
Low levels of fecal coliforms were infrequently detected in samples collectedon 3/1/99 and 10/25/99. As shown in Table 2, detectable levels of E. coli werepresent in 7% and 4% of soil samples collected 3/1/99 and 10/25/99, respectively.
Average levels of E. coli in soil samples ranged from 3.5 to 4.0 log10 MPN/100g in all samples collected in 1999. Because fecal indicator bacteria occur inonly very low concentrations (typically less than 2 bacterial cells per gram) inmost uncontaminated soils,[5] it is evident that some of the soils analyzed inthis study had become contaminated with human or animal feces. While it ispossible that the E. coli-containing samples collected on 10/25/99 had becomefecally contaminated as a result of flooding following Hurricane Floyd, the levelsof E. coli in these post-flood samples were not significantly higher than thesamples collected on 3/1/99, months before the flooding occurred. Neither fecalmaterial nor biosolids in the form of fertilizer were used at the sampling sites inthe present study. Hence, these soils may have become fecally contaminated bysome other mechanism, such as by wildlife intrusion or by runoff from nearbydomestic or agricultural animal sources.
In contrast to the low levels of fecal coliforms and E. coli, high levels of total coliforms in soil were present. Such data are expected, because the totalcoliform group includes bacteria that occur naturally in soil. Average levels oftotal coliforms were lowest (4.0 log10 MPN/100 g) in samples collected on 3/1/99,and samples collected on 10/25/99 contained 4.1 log10 MPN total coliforms/100 g.
There was not a significant difference between total coliform levels in samplesfrom 3/1/99 and 10/25/99.
Despite the wide acceptance and use of fecal coliforms and E. coli as micro- bial indicators of fecal contamination, there is growing evidence that thesebacteria are poor predictors of other types of enteric microbes, such as en-teric viruses.[6]Enteric bacteriophages (viruses that infect enteric bacteria) havebeen proposed as alternative indicators of human enteric viruses.[7] Specifi-cally, male-specific (bacterial viruses that infect through a bacterium’s F-pili,referred to as F+) coliphages with ribonucleic acid (RNA) genomes (F+RNAcoliphages) may be better indicators than fecal coliforms for monitoring thevirological quality of environmental media, because they more closely resem-ble human enteric viruses. F+ coliphages are present in wastewater, and aresufficiently persistent in the environment (and resistant to water treatmentprocesses) to be acceptable enteric virus indicators.[8,9] Somatic coliphages, orbacterial viruses that infect a bacterium through its outer cell layer, also arepotentially useful indicators of human enteric viruses.[10] Similar to the low levels of E. coli, relatively low levels of coliphages were de- tected in soils analyzed in the present study (Table 3). Somatic coliphages werenot detected in samples from 3/1/99, but average levels of 1.1 log10 MPN/100 g Hurricane Impact on Agricultural Soils in North Carolina were detected in 11% of samples from 10/25/99. Male-specific coliphages weredetected in 14% of samples collected 3/1/99 but were not detected (i.e., were<0.47 log10 MPN/100 g) in samples collected on 10/25/99. The levels and occur-rence of coliphages in soils before and after flooding were lower compared toE. coli or fecal coliforms. This may be due to the lower initial levels of coliphagescompared to these bacteria in feces or wastewaters, or the possible rapid migra-tion of coliphages through the soil after deposition on land following flooding.
Previous investigations have demonstrated that viruses entering the subsur-face environment and that are adsorbed to soil particles are not irreversiblybound and may be eluted by percolating fluids under certain conditions.[11]Male-specific coliphages have been shown to adsorb poorly to soil particles,[12,13]and under saturated soil conditions, some coliphages can move laterally over adistance of about 900 meters in a relatively short time period, at a rate of about350 meters per day.[14] While fecal coliforms and E. coli indicated the presence of fecal contami- nation in some samples tested, these bacteria may also be poor predictors ofother types of fecal microbes, such as the (oo)cysts of the protozoan parasitesCryptosporidium parvum and Giardia lamblia. Anaerobic spore forming en-teric bacteria, such as Cl. perfringens, have been proposed as an alternative in-dicator of protozoan (oo)cysts and other environmentally persistent pathogenssuch as helminth ova.[7] In contrast to the low prevalence and levels of E. coliand coliphages detected in soil in the present study, 86–100% of soil sampleswere positive for Cl. perfringens spores, with some individual samples contain-ing tens of millions of spores. As shown in Table 3, all samples analyzed from3/1/99 contained Cl. perfringens spores, and 2 samples had levels of spores abovethe detection limit of >7.4 log10 MPN/100 g. Average levels of Cl. perfringensspores in soils were 4.7 and 6.6 log10 MPN/100 g in samples from 3/1/99 and10/25/99, respectively. Levels of spores in post-flood samples (10/25/99) weresignificantly (P < 0.001) higher than spore levels in pre-flood samples (3/1/99).
Cl. perfringens is consistently associated with human fecal wastes,[15] and the levels of its spores may be used for the comparison of fecal contamina-tion in the environment over extended periods of time. Previous investigatorshave shown that enumeration of Cl. perfringens spores is especially useful inmonitoring the deposition and movement of sewage,[6] and is a more reliableindicator of fecal contamination than fecal coliforms when non-point sourcesof pollution, such as flooding or farmland runoff are considered.[16,17] More re-cently, Fujioka[21] monitored coastal waters at marine beaches known to receivewater from stream or storm drains for contamination by non-point and pointsource fecal pollution and concluded that Cl. perfringens was a useful indica-tor of fecal contamination for such purposes. The higher levels of Cl. perfrin-gens spores in post-flood soil samples analyzed in this study may have beendue to deposition from fecally contaminated flood water, as elevated levels ofCl. perfringens spores also were detected in the vicinity of the CEFS in the Neuse River after Hurricane Floyd.[19] This hypothesis is further supported byprevious findings, in which densities of fecal indicators, bacteria, and virusesin sediments exceeded those in the overlying water by as much as three ordersof magnitude,[20] and the levels of Cl. perfringens spores in this study (rang-ing from 4.7 to 6.6 log10 MPN spores/100 g) are similar to the level of 4.8 log10MPN/100 g reported in tidally influenced soil in Florida.[18] Our findings supportthe use of Cl. perfringens spores as fecal indicators, based on their associationwith non-point source pollution events such as flooding and their persistencein the environment.
While the persistence of Cl. perfringens spores in soil at the CEFS, and thesignificantly higher spore levels following major flooding suggests that soil canbecome contaminated as a result of flooding with water containing fecal wastes,other mechanisms of fecal contamination may not be discounted. However, thedata presented in this study shows that Cl. perfringens is useful for comparingrelative levels of fecal contamination between environmental samples that varytemporally. The low levels of E. coli and coliphages in soil suggest that these mi-crobes are less suitable for comparing the relative levels of fecal contaminationof soil, either as a result of flooding or by some other means.
The location of the CEFS and other agricultural operations in the flood- plains of major waterways in North Carolina and elsewhere highlights the im-portance of determining the sanitary quality of soil and water impacted by flood-ing as an important goal. Hurricanes occur with regularity in North Carolinaand in other coastal regions, and during the next two decades, it is predictedthese storms will cause damage 5–10 times worse than previously experiencedin the Gulf and Atlantic Coast states (Prof. William M. Gray, Colorado StateUniversity, Dept. of Atmospheric Sciences, personal communication). This pre-diction is based on the increasing density of human activity in areas knownto be at risk for flooding and high winds. Hence, there is a clear need for mea-sures to protect agricultural and other areas at risk for flooding. Such measuresshould include prospective and retrospective microbiological monitoring of soiland improved protection of watersheds and crop-growing areas from incidentscapable of releasing fecal material.
Using the methods employed in the present study, it is possible to recover and quantitate the levels of fecal contamination in agricultural soils by enu-meration of fecal indicator micro-organisms. With relatively early or advancenotices in the form of weather advisories that are now available,samples couldbe collected and assayed immediately before and after major storm events us-ing the methods described in this study. Such data would allow farmers, scien-tists, environmental managers and regulatory authorities to better assess and Hurricane Impact on Agricultural Soils in North Carolina respond with control measures to minimize potential risks to human, animal,and plant health when flooding occurs as a result of severe weather conditions.
Moreover, these methods can be applied to determining the sanitary quality ofagricultural soils when they become contaminated by other mechanisms, suchas runoff from livestock operations or land application of municipal wastewaterand biosolids.
This work was performed in collaboration with Melissa Bell, Frank Louws,Nancy Creamer, and others in the Departments of Plant Pathology and SoilScience, North Carolina State University, Raleigh, North Carolina. We thankOtto D. Simmons, III, Christine Stauber, Sharon Long, and Carlton Andersonfor their assistance with some of the microbial assays. This research was sup-ported in parts by funds from North Carolina State University, the University ofNorth Carolina Water Resources Research Institute, the United States Depart-ment of Agriculture and the United States Environmental Protection Agency.
American Public Health Association. Standard Methods for the Examination of Wa- ter and Wastewater, 18th Edition; APHA, Washington, DC, 1992.
St. John, W.D.; Matches, J.R.; Wekell, M.M. Use of iron-milk medium for enumera- tion of Clostridium perfringens. J. Assoc. Anal. Chem. 1982, 65(5), 1129–1133.
United States Environmental Protection Agency (USEPA). Method 1601: Male Spe- cific (F+) and Somatic Coliphage in Water by Two-Step Enrichment. Office of Water,Washington, D.C. Report No. EPA 821-R-01-030, 2001.
Thomas, H.A., Jr. Bacterial densities from fermentation tube tests. J. Amer. Water Works Assoc. 1942, 34, 572.
Geldreich, E.E.; Huff, C.B.; Bordner, R.H.; Kabler, P.W.; Clark, H.F. The faecal coli- aerogenes flora of soils from various geographical areas. J. Appl. Bacteriol. 1962, 25,
87–93.
Payment, P.; Richardson, L.; Edwardes, M.; Franco, E.; Siemiatycki, J. A prospective epidemiological study of drinking water related to gastrointestinal illness. Wat. Sci.
Tech. 1991, 24, 27–28.
Ashbolt, N.J.; Grabow, W.O.K.; Snozzi, M. Indicators of microbial water quality. In Water Quality: Guidelines, Standards and Health. Fewtrell, L.; Bartram, J., Eds.; Inter-national Water Association Publishing, London, UK, 2001; 289–316.
Havelaar, A.H.; Pot-Hogeboom, W.M.; Pot, R. F-specific RNA bacteriophages in sewage: Methodology and occurrence. Wat. Sci. Tech. 1984, 17, 645–655.
Havelaar, A.H.; Pot-Hogeboom, W.M.; Furuse, K.; Pot, R.; Hormann, M.P. F-specific RNA bacteriophages and sensitive host strains in faeces and wastewater of human and
animal origin. J. Appl. Bacteriol. 1990, 69, 30–37.
Payment, P.; Franco, E. Clostridium perfringens and somatic coliphages as indica- tors of the efficiency of drinking water treatment for viruses and protozoan cysts. Appl.
Environ. Microbiol. 1993, 59(8), 2418–2424.
Landry, E.F.; Vaughn, J.M.; Thomas, M.Z.; Beckwith, C.A. Adsorption of en- teroviruses to soil cores and their subsequent elution by artificial rainwater. Appl. En-
viron. Microbiol. 1979, 38(4), 680–687.
Goyal, S.M.; Gerba, C.P. Comparative adsorption of human enteroviruses, simian rotaviruses, and selected bacteriophages to soils. Appl. Environ. Microbiol. 1979, 38,
241–247.
Meschke, J.S.; Sobsey, M.D. Comparative adsorption of Norwalk virus, poliovirus 1 and F+RNA coliphage MS2 to soils suspended in treated wastewater. Water Sci. Tech.
1998, 38(12), 187–189.
Melnick, J.L.; Gerba, C.P. Viruses in water and soil. Public Health Rev. 1981, 9(3–4),
Emerson, D.J.; Cabelli, V.J. Extraction of Clostridium perfringens Spores from Bot- tom Sediment Samples. Appl. Environ. Microbiol. 1982, 44(5), 1144–1149.
Fujioka, R.S.; Shizumura, L.K. Clostridium perfringens, a reliable indicator of stream water quality. J. Water Pollut. Control Fed. 1985, 986–992.
Sorenson, D.L.; Eberl, S.G.; Dicksa, R.A. Clostridium perfringens as a point source indicator in non-point polluted streams. Water Res. 1989, 23(2), 191–197.
Desmarais, T.R.; Solo-Gabriele, H.M.; Palmer, C.J. Influence of soil on fecal indica- tor organisms in a tidally influenced subtropical environment. Appl. Environ. Microbiol.
2002, 68(3), 1165–1172.
Bales, J.D.; Oblinger, C.J.; Sallenger, Jr., A.H. Two Months of Flooding in Eastern North Carolina, September-October 1999: Hydrologic, Water Quality, and Geologic Ef-fects of Hurricanes Dennis, Floyd, and Irene. U.S. Geological Survey, Water ResourcesInvestigations Report 00–4093, 2000.
LaBelle, R.L.; Gerba, C.P.; Goyal, S.M.; Melnick, J.L.; Cech, I.; Bogdan, G.F. Re- lationships between environmental factors, bacterial indicators, and the occurrence of
enteric viruses in estuarine sediments. Appl. Environ. Microbiol. 1980, 39(3), 588–596.
Fujioka, R.S. Monitoring coastal marine waters for spore-forming bacteria of faecal and soil origin to determine point from non-point source pollution. Water Sci. Tech. 2001,
44(7), 181–188.

Source: http://www.microbial-intelligence-group.us/images/Casteel_et_al_2006.pdf