Analysis of Biodiversity in the Sparkill Creek Watershed As a Course-Based Service Learning Undergraduate Experience

NEW YORK STATE WATER RESOURCES INSTITUTE Department of Biological and Environmental Engineering

230 Riley-Robb Hall, Cornell University Tel: (607) 254-7163 Ithaca, NY 14853-5701 Fax: (607) 255-4080 http://wri.cals.cornell.edu Email: [email protected]

Analysis of Microbial Biodiversity in the Sparkill Creek Watershed throughout a Course-based Service Learning Research Program for Undergraduates

Bernadette J. Connors*

Dominican College, 470 Western Highway, Orangeburg, NY 10962 [email protected]

Abstract Citizen science groups are becoming more prevalent in many communities, raising awareness of problems that impact the larger population. The types of projects that many of these groups take on can also provide undergraduates with a service learning opportunity, their participation fostering an appreciation of their actual and future contributions to society. We developed a course-based research project that gave students a chance to explore and describe the microbiome of the Sparkill Creek. In 2010, the NYS Department of Environmental Conservation added the Creek to its list of impaired waters due to the high concentration of the indicator microbe, Enterococcus spp. This research project aimed to describe the microbial species present at selected sites along the creek. Students in three courses carried out the project. During this time, they collected and filtered water samples, isolated bacteria in pure culture, performed Gram stains, API20E biochemical analyses, 16S rDNA sequence analysis, and antibiotic sensitivity testing. Success of this course-based research project is expected to help students gain an appreciation and understanding of their role in solving a problem important to the local community and how they, as future professionals, may be able to contribute to the population-at-large.

Analysis of Microbial Biodiversity in the Sparkill Creek Watershed throughout a Course-based Service Learning Research Program for Undergraduates Title

Three Summary Points of Interest This research was successful at obtaining data about which bacterial species and genera are common in the Sparkill Creek. Pure culture isolates allowed for antibiotic susceptibility testing, which revealed that there is a high percentage of bacteria present that are resistant to multiple drugs. This type of project was effective among three varied groups of students, and served as a means to show them how their expertise can inform the general public about the local environment.

Keywords Sparkill Creek watershed, Bacterial diversity, fecal coliform, course-based research experience

NYSWRI090413 2 Analysis of Microbial Biodiversity in the Sparkill Creek Watershed throughout a Course-based Service Learning Research Program for Undergraduates Title

Introduction Service learning is a well-recognized strategy to engage students in the learning process, while creating individuals that possess a mature sense of their role in the harmonious function of both local and global communities. It can be carried out at all educational levels, and among all disciplines. As reported by Campus Compact (Annual Report 2012-2013), 95% of higher education respondents (member and non-member institutions) stated that service learning was well- established on their campus with significant institutional support. Additionally, 64% of member campuses (1,100 members) required participation in a service learning project as part of their core curriculum. Independent polls of students who participated in a service learning course of some type have consistently shown favorable attitudes, some reporting that they would continue in a community service project beyond that initial experience. Successful programs have expanded student roles beyond service activities to include identifying and developing service learning projects, recruiting their peers, leading reflection activities, serving on advisory committees and making presentations both within the educational community and to the community at large (Jeandron and Robinson, 2010).

Qualitative and quantitative surveys have shown a correlation between service learning and positive learning outcomes. Blackwell (1996) reported on undergraduate and graduate students enrolled in service learning courses offered within various disciplines. Of those interviewed, 85% believed service learning should be incorporated in more classes, and the majority felt that the experience strengthened their understanding of class material. Other reports indicate that students in service learning courses achieve significantly higher mean final grades when compared to a control group (Berson and Younkin, 1998). Class discussions were found to be more stimulating, with a greater amount of student participation. In a controlled study comprised of four separate groups of undergraduate students who participated in a service learning course there were significant improvements in cognitive complexity, self-worth, and social complexity when compared to groups that did not participate (Osborne et al., 1998). Service learning has also been shown to impact ethical reasoning and affect a change in the student’s psychosocial development, specifically with regard to justice, equality and dignity (Greene, 1997).

Course-based research programs have also proven highly effective in creating an atmosphere of active learning. The goals of these research projects are to provide an opportunity by which students can apply disciplinary knowledge and discover their own potential as problem solvers, to enable the student to experience the gratification, frustration, uncertainty, enlightenment and empowerment that comes in the process of doing research, and to help students to develop the analytical skills by which a problem may be solved and the communication skills by which the researcher may present this research to the public. To incorporate course-based research in the context of a service learning project is expected to significantly strengthen the impact the education of the student as both a scientist and a member of the community.

The Research Project Microbial contamination of water sources is an important issue that faces urbanized areas of the United States. Of particular concern is the presence of Escherichia coli, Enterococcus spp., Salmonella spp., Giardia lamblia, and Cryptosporidium parvum in drinking water reservoirs. A powerful indicator of sewage contamination are the fecal coliforms, and these have been adopted as the standard organism by which water is deemed safe for consumption. The EPA has set strict no tolerance limits on the total coliforms allowed in drinking water, and mandates that an operating system is out of compliance if more than 5% of its monthly water samples contain these organisms. From a human health standpoint the presence of these microbes poses a risk to individuals using these water sources for both consumption and recreation. Many of these bacteria may be resistant to multiple antibiotics, which poses further issues to human wellness. In terms of environmental health, their presence affects not only the normal microflora of the body of water, but the plants and animals that reside in the area as well.

The standard indicator organism used in testing water for the presence of fecal contamination is Enterococcus spp., although many other detrimental bacteria, fungi, protozoa, and viruses may be present that affect the overall quality

(Water Quality Standards, 2015). Enterococcus is a Gram positive, facultative anaerobe that is found in both human and nonhuman-animal alimentary canals. There are approximately 28 species within this genus that have been well- characterized through their growth characteristics, physiology, biochemistry, and molecular makeup (Angeletti et al., 2001, Brtkova et al., 2010, Bhardwaj et al., 2013, Manero et al., 1999). They are hardy bacteria that can out compete for resources very successfully and, as such, disturb the balance of natural microflora in the environment. Numerous studies have demonstrated a strong correlation between the various species of this indicator organism and their animal hosts (Devriese et al., 2002). For this reason, species diversity analysis may lead to a better understanding of the source of this contamination, particularly when overlayed on a map of the sewage lines, septic tanks, and businesses in the region.

The use of Enterococcus as the indicator species does not preclude an examination of the microbiome of the same environment in the hope of elucidating the source of a particular contamination (Guang et al., 2004, Water Quality Standards, 2015). Their use as indicators of human fecal pollution and contamination has been challenged, namely because enterococci are likewise found in animal feces, on plants, and in the soil (Devriese et al., 1987). Therefore, actions must be taken to identify host-specific species of other microbes in order to assist in the distinction between human fecal contamination and additional environmental sources of enterococci (Boehm & Sassoubre 2015).

As shown in the map of the project area (Figure 1), the Sparkill Creek Watershed spans parts of Rockland County, NY and Bergen County, NJ. The stream’s headwaters are located in the Palisades at the northern end of the watershed, eventually turning back toward the north and emptying into the Hudson River at Piermont Marsh. Sparkill Creek water quality is monitored by Riverkeeper in partnership with the Sparkill Creek Watershed Alliance (SCWA). In 2010 NYS Department of Environmental Conservation added Sparkill Creek to its List of Impaired Waters (“303(d) List”) due to elevated levels of fecal indicator bacteria. Exploratory sampling conducted near the mouth of the creek in 2011-2014 showed that the creek suffers from extremely high levels of fecal contamination (Figure 1). Homes and businesses in the area exist aside these waters, and seepage of this contamination into the groundwater means that area neighborhoods are also likely affected by this contamination.

With completion of this project we expect students will not only gain technical skills but also gain an understanding of the role that they can play in addressing community problems upon graduation. It is also our expectation that the enthusiasm of our community partners and activists will be bolstered by these results. Their input, along with an understanding the type and nature theof bacterial contaminants, is critical to remediating the Sparkill Creek watershed.

Marsico Tackamac Greenbush… Dry Days Spruce… Inertia… Blauvelt… Orangebur… Orangetow… Tappan 303 Tappan… NY Arm at… Sparkill… Moturis Sparkill Rt.… Piermont…

010002000 Enterococcus (MPN/100 ml upper test limit 2196)

Figure 1. The levels of the indicator organism, Enterococcus spp., were determined using the IDEXX Enterolet system, and EPA-approved method to estimate the number of these bacteria in a water sample. Numbers are given for both wet and dry testing periods. Sites of collection are located on the adjacent map. All sites are found to be above the EPA Recreational Water Quality Criteria maximum of 35 cfu/100 ml for direct recreational contact. Blue lines represent the modeled stream (Sparkill Creek Citizen Science Pathogen Indicator Project (NEIWPCC). Lawerence Vail, submitted December 2014)

Results & Discussion

From a scientific perspective, the amount of data that can be collected using course-based research experiences (CRE) is significant, with repetition of experiments a built-in factor when carried out among multiple groups. Students in three different courses participated in this project, in two different, yet similar, ways. One group was responsible for collecting information on Enterococcus isolates (Table 1), while the other groups worked with random bacterial isolates. The information collected included species identification, with a concurrent analysis of biochemical traits and antibiotic susceptibilities. As can be seen in Table 1, isolates of Enterococcus, grown on Slanetz and Bartley selective media (verified to be Gram positive, catalase positive bacteria) were subjected to 16S rDNA sequencing. Students then compared the sequences using simple BLASTN searches after quality processing. Enterococcus faecalis and Enterococcus faecium were the most prevalent species cultured from the water samples (22 of the 49 isolates). These two are also the most prevalent species cultured from human sources, accounting for more than 90% of clinical isolates (de Perio et al., 2006). E. durans and mundtii account for 11/49 isolates analyzed in this study, with 8 of the 49 isolates being identified as E. hirae. Characteristics, including risk of infection, antibiotic resistance, and typical habitat are given in Table 2. Devriese et al (2002) did note that E. hirae and durans show 16S rDNA sequence similarity close to 98.8% of the time, and are often

This report was prepared for the New York State Water Resources Institute (WRI) and the Hudson River Estuary program of the New York State Department of Environmental Conservation, with support from the NYS Environmental Protection Fund Analysis of Microbial Biodiversity in the Sparkill Creek Watershed throughout a Course-based Service Learning Research Program for Undergraduates Title difficult to differentiate with conventional biochemical tools. Since Enterococcus can be found in the GI tract of multiple warm-blooded animals, as reported in numerous sources, it may be considered a poor tool to use in the determination of contaminating sources of fecal pollution.

To more thoroughly understand the population of microbes present at selected sites, two other groups of students cultured random bacterial isolates following water filtration and growth on undefined media (TSA). In these studies, Gram staining, growth on selective and differential media, API20E testing, and 16S rDNA sequencing were performed. Table 3 shows the common species found amongst the six sites, as well as the natural habitat of these bacteria. Gram negative isolates were also analyzed using the API20E test strips, and identity revealed using APIweb. There was significant discrepancy between 16S rDNA sequence identity and that revealed by APIweb. Table 4 indicates the discrepancy as percent mismatch between the two sources.

All three groups of students determined the antibiotic susceptibility and resistance of the isolates that they were working with. Figure 2 shows the analysis of antibiotic resistance among the non-enterococcal isolates. Using thirteen different antibiotics, students categorized the ability of the isolates to resist the activity of multiple drugs. As seen, in all six sites tested, the majority of isolates were resistant to between 5 and 9 of the antibiotics used. This information, we predict, will promote support and activism from the local community. Since there are businesses and private homes near the banks of the Sparkill Creek, multiple drug resistance poses a more serious health hazard.

College and university campuses nationwide are beginning to evaluate the effectiveness of CREs. These are enticing especially to small schools with prohibitive research funding. This project represented an innovative approach to CREs in that it combined the CRE with service-learning, with the aim of providing a local citizen science group with data that could be used toward understanding the contamination found in the Sparkill Creek watershed. Anecdotally, it can be said that students felt a greater ownership of the project, knowing that their work would be disseminated to the larger community.

Figure 2. Antibiotic Susceptibilities of Isolated Microbes. Bacterial isolates incubated in TSB were used to create lawns on Müller-Hinton agar plates. Each of the 13 antibiotic discs were evenly dispensed onto the plates. Antibiotics used in this test included ampicillin, chloramphenicol, erythromycin, gentamicin, kanamycin, linezoid, minocycline, neomycin, novobiocin, penicillin, streptomycin, tetracycline, and vancomycin. This figure represents the percent of bacteria resistant to the various number of antibiotics per site. Careful analysis of this data shows that most isolates were resistant to multiple drugs.

Isolate Species Isolate Species Designation Identification Designation Identification A1 E. faecalis B1 E. durans A2 E. mundtii B2 E. durans A3 E. mundtii B3 E. hirae A4 E. durans B4 E. faecium A5 E. faecalis B5 E. hirae A6 E. faecalis B6 E. durans A7 E. faecium B7 E. faecium A8 E. mundtii B8 E. faecium A9 E. faecalis B9 E. faecium A10 E. mundtii B10 E. hirae A11 E. mundtii B11 E. hirae A12 E. faecalis B12 E. hirae Isolate Species Isolate Species Designation Identification Designation Identification

C1 Indeterminable D1 E. faecalis

C2 E. faecalis D2 Indeterminable

C3 E. faecalis D3 E. hirae

C4 E. faecalis D4 E. hirae

C5 E. mundtii D5 Indeterminable

D6 E. hirae C6 Indeterminable C7 Indeterminable D7 E. faecalis C8 E. hirae D8 E. faecalis C9 E. faecium D9 E. faecalis C10 Indeterminable D10 E. faecalis C11 E. faecium D11 E. faecalis C13 E. faecium D12 Indeterminable Table 1: Species identification of Enterococcus isolates. The Tackamack site is indicated by “A”, Blauvelt site by “B”, Rt 303/340 site by “C”, and Tappan site by “D”.

Enterococcus sp. Nosocomial Drug Illness type Typical habitat infection Resistance Enterococcus No Yes Enterococcal infection, rare GI Tract animal durans predominantly, feces &soil,calm

water, aquatic plants Enterococcus Yes Yes, Vancomycin Resistant GI Tract, feces faecium Enterococcus &soil,calm Bacteremia, UTI, tissue water, aquatic infections plants Enterococcus Yes Yes Bacteremia, UTI, tissue GI Tract, feces faecalis infections &soil,calm water, aquatic plants Enterococcus No Yes Enterococcal infection, rare GI Tract animal hirae predominantly,

feces &soil,calm water, aquatic plants Enterococcus No Yes Enterococcal infection, rare GI Tract, feces mundtii &soil,calm water, aquatic plants

Table 2. Characteristics of eneterococcal species found among four sites in the Sparkill Creek.

Piermont Rt 303/ Rt Tackamack Sites Blauvelt Rd Spruce St Clausland Rd Bridge 340 Park Enterococcus Citrobacter Shigella Enterobacter gallinarum murliniae sonnei aerogenes Citrobacter Hafnia Fecal koseri paralvei Entercococcus faecalis Hafnia paralvei Clinical Enterobacter Enterococcus Enterococcus Enterococcus Specimens ludwigii casseliflavus casseliflavus casseliflavus Staphylococcus Staphylococcus capitis epidermidis Staphylococcus Skin epidermidis Staphylococcus warneri Bacillus Bacillus Citrobacter Bacillus Bacillus Bacillus pumilus anthracis murliniae pumilus anthracis anthracis Proteus Proteus Proteus Proteus Lysinibacillus Bacillus Widespread mirabilis mirabilis mirabilis mirabilis pakistanensis cereus in nature Staphylococcus Roseovarius Pseudomonas Lysinibacillus Pseudomonas sciuri halotolerans argentinensis sphaericus protegens Serratia Proteus proteamaculans mirabilis Table 3. Common Species and Genera. Water samples were filtered through Büchner funnels containing nitoccellulose membrane filters. These filters were placed on TSA plates and grown overnight at 37˚C. A total of 80 isolates were then randomly selected and sent out for DNA sequencing using the 27F 16S rDNA primer. This table shows the common species found amongst the six sites as well as the natural habitat of these bacteria.

Site Spruce Clausland 303/340 Piermont Blauvelt Tack Park

% mismatch 92.3% 100% 50.0% 66.7% 76.9% 100% with API20E

Table 4. Similarity between API20E Biochemistry Test and DNA Sequencing. In order to identify the isolates collected, both DNA sequencing and API20E test strips were used. Since the API20E biochemistry test yielded results which differed from the DNA sequencing test, percent mismatch between the two analyses amongst the six sites were calculated. This comparison was done for only those isolates identified as G- bacteria.

Policy Implications

As can be seen on the modelled map of the creek (Figure 1), a portion of the Rockland County area will be most affected by the findings of this research study. In addition, the New Jersey area (Bergen County) which is part of the Sparkill Creek watershed will be affected. The purpose, from a community perspective, was to promote remediation of the studied waterways that will impact the local community. Once a greater understanding of the source of contamination is determined, both through microbiome testing and determination of antibiotic resistance, we can work with community partners to propose methods by which a more sustainable management system could be put in place. Given that the primary objective of this research was educational, there is little to say on how it will impact policy on policy decisions. Since there was outreach to a local community group, however, we expect that the involvement of our organization will spur further collaborations.

Methods Isolation of bacteria in pure culture Sites along the Sparkill Creek were chosen and sterilized Nalgene bottles (1L) were used for the collection. Water was collected after rinsing the bottles three times with the creekwater, being sure to dispose of the washwater downstream of the collection site. Once in the laboratory, students filtered 100mL of water through sterilized 0.45uM filters (Millipore) using an applied vacuum. Filters were removed with forceps and placed “collection-side” down onto Slanetz and Bartley media (Fisher Scientific), which is selective for the growth of enterococci. This procedure was repeated with the filters being plated on tryptic soy agar (TSA). Plates were incubated at 37C for 3 days, followed by isolation of pure cultures by way of a four quadrant streak onto tryptic soy agar (Carolina Biologicals).

Gram staining Bacterial smears were created and heat-fixed using colonies obtained from the streak plates. A standard Gram stain procedure was then carried out. Students observed their isolates, along with the control stains (Escherichia coli and Staphylococcus epiderimidis), using the oil immersion light microscopes. Gram reaction and cell morphologies were noted.

API20E testing API20E strips (bioMerieux) were used to examine the biochemical diversity among the isolates. Single colonies were aseptically placed in 3mL of sterile 0.7% saline and then vortexed for 30 seconds. To inoculate the strips, 100uL of the cell suspension was measured into each cupule, with the exception of CIT, VP, and GEL, which were filled completely. The cupules containing tests for LDC, ODC, ADH, H2S, and URE were also covered with sterile mineral oil to create an anaerobic environment. Trays were incubated at 37C for 24 hours before observation. Positive and negative results were indicated by a color change. APIweb was used to reveal the species identity.

Antibiotic sensitivity testing A single large colony was used to inoculate 1mL of sterilized tryptic soy broth for each of the isolates, which were incubated at 37C for 24 hours. Spread plates (Mueller-Hinton) were created using the overnight cultures by spreading the cell suspension using a sterile swab. Two plates that served as controls (E. coli and the other with S. epidermidis) were also included. Antibiotic disks (BD BBL Antimicrobial Sensitivity DisksTM) were placed onto the agar surfaces and plates were incubated at 37C for 24 hours. Zones of inhibition (mm) were measured, and these diameters were compared to the values that dictate sensitivity or resistance. Antibiotics tested included: tetracycline (30 µg), vancomycin (30 µg), ampicillin (10 µg), erythromycin (15 µg), chloramphenicol (30 µg), minocycline (30 µg), linezolid (30 µg), kanamycin (30 µg), penicillin G (10 µg), novobiocin (30 µg), neomycin (30 µg), and streptomycin (30 µg).

DNA Sequence analyses 1. Individual colonies of Enterococcus were sent to Genewiz, Inc (Plainfield, NJ) for 16S rDNA sequencing. Both forward and reverse sequences were delivered in a .fastq format. Using the Blue Line of DNA Subway, students uploaded their sequences, trimmed them using the Trimmer function, and judged the identity of any remaining unknown bases (“N”) by looking at the chromatograms. These sequences were then entered into a BLASTN search. Default search parameters were utilized, although organism type was entered (Enterococcus), and both models and uncultured sample sequences were excluded. Possible identities were then determined by examining the maximum score, total score, query cover, e value, and identity of both forward and reverse sequences. To obtain a phylogenetic tree, DNA Subway was utilized once more. The best set of data was chosen out of the forward and reverse sequences. These data were entered in DNA Subway in FASTA format. “MUSCLE” was selected and this allowed for the phylogenetic tree formation. To observe the tree, PHYLIP NJ was selected and it brought you to the next page that showed the finished tree. 2. Individual isolates of unknown bacteria were grown overnight in tryptic soy broth. Genomic DNA was isolated using the Qiagen PureYield DNA Isolation kit. PCR amplification of the 16S rDNA region using primers 27F (AGA GTT TGA TCM TGG CTC AG and 1492R: CGG TTA CCT TGT TAC GAC TT) and 1518R (AAGGAAGGTGATCCANCCRCA) were used to amplify the region for sequencing, and primer 27F was then used for sequencing. Unpurified PCR product was sent to BioBasic for sequencing. The same methodology of analysis was performed for these sequences as for the Enterococcus, without creation of phylogenetic trees.

Outreach Comments The project goals and outcomes were shared with the Sparkill Creek Watershed Alliance, a community group focused on restoring the Sparkill Creek watershed to a healthy state.

Student Training This project was carried out with three groups of students. The courses within which this was carried out were: BI337: Evolution (12 students) (Enterococcus identification, with sequencing) BI229: Molecular Microbiology (22 students) (Unknown identification without sequencing) BI441: Research Seminar (2 students) (Unknown identification with sequencing)

References

Angeletti S, Lorino G, Gherardi G, Battistoni F, De Cesaris M, Dicuonzo G (2001) Routine Molecular Identification of Enterococci by Gene- Specific PCR and 16S Ribosomal DNA Sequencing. J Clin Microbiol Vol 39, No 2 pp 794–797

Berson JS and Younkin, WF (1998) Doing Well By Doing Good: A Study Of The Effects Of A Service-Learning Experience On Student Success. Paper presented at the American Society of Higher Education, Miami, FL.

Bhardwaj S, Bharme K, Dhawale J, Patil M, Divase S. (2013) Enterococcus faecium and Enterococcus faecalis, the nosocomial pathogens with special reference to multi-drug resistance and phenotypic characterization. International Journal of Pharmaceutical Science and practice. 2(1): 1-10.

Blackwell AP. (1996) Students’ Perceptions of Service Learning Participation in the College of Health and Human Services at the University of Southern Mississippi. Unpublished Dissertation, The University of Mississippi.

Boehm, Alexandria B., and Lauren M. Sassoubre. "Enterococci as Indicators of Environmental Fecal Contamination." National Center for Biotechnology Information. N.p., 5 Feb. 2015. Web. 3 Nov. 2015.

Brtkova A, Filipova M, Drahovska H, Bujdakova H (2010) Characterization of enterococci of animal and environmental origin using phenotypic methods and comparison with PCR based methods. Veterinarni Medicina, 55 (3): 97–105.

Campus Compact Annual Report 2012-2013 de Perio MA, Yarnold PR, Warren J (2006) Risk factors and outcomes associated with non-Enterococcus faecalis, non- Enterococcus faecium enterococcal bacteremia. Infect Control Hosp Epidemiol. 27(1):28-33.

Devriese LA, Van De Kerckhove A, Kilpper-Balz IR and Schleifer, KH (1987) Characterization and Identification of Enterococcus Species Isolated from the Intestines of Animals. Int J Systematic Bacteriology. Vol 37, No. 3, p 257-259.

Devriese LA , Vancanneyt M , Descheemaeker, P , Baele M, Van Landuyt HW, Gordts B , Butaye P , Swings J. and Haesebrouck F. (2002), Differentiation and identification of Enterococcus durans, E. hirae and E. villorum. Journal of Applied Microbiology, 92: 821–827.

Greene D. (1997) The use of service-learning in client environments to enhance ethical reasoning in students. American Journal of Occupational Therapy, 51(10), 844-852.

Guang J, Englande, AJ, Bradford, H; Huei-Wang J (2004) Comparison of E.Coli, Enterococci, and Fecal Coliform as Indicators for Brackish Water Quality Assessment Water Environment Research, Vol. 76, No 3, pp 245-255

Jeandron C and Robinson G (2010) A Climate for Service Learning Success. AACC.

Kristich CJ, Rice LB, Arias CA. Enterococcal Infection—Treatment and Antibiotic Resistance. 2014 Feb 6. In: Gilmore MS, Clewell DB, Ike Y, et al., editors. Enterococci: From Commensals to Leading Causes of Drug Resistant Infection

Manero A and Blanch AR (1999) Identification of Enterococcus spp. with a Biochemical Key. Appl Environ Microbiol Vol 65, No10 pp4425-4430.

Osborne SP (1998) Naming the Beast: Defining and Classifying Service Innovations in Social Policy. Human Relations 51:1133-1154.

Water quality standards for coastal and great lakes recreation waters. (2015, May 11). Retrieved 6 November 2015, from https://federalregister.gov/a/04-25303