FINAL REPORT FOR THE EMPACT PROJECT R-82933901-0

DATA COLLECTION AND MODELING OF ENTERIC PATHOGENS, FECAL INDICATORS AND REAL-TIME ENVIRONMENTAL DATA AT MADISON, WISCONSIN RECREATIONAL BEACHES FOR TIMELY PUBLIC ACCESS TO WATER QUALITY INFORMATION

Environmental Monitoring for Public Access and Community Tracking Funded by the US Environmental Protection Agency with Matching Funds from the City of Madison, the United States Geological Survey and the Wisconsin State Laboratory of Hygiene

TABLE OF CONTENTS SUMMARY/ACCOMPLISHMENTS.......................................................................................................... 1 1.

Introduction......................................................................................................................... 2 1.1. Past Monitoring Efforts and Future Benefits ......................................................... 2 1.2. Madison Beach EMPACT Team ........................................................................... 4 Table 1-1: EMPACT Project Key Personnel ............................................ 6 Figure 1-1: Project Team Organizational Chart........................................ 8 1.3. Objectives of the Project........................................................................................ 8

2.

Data Collection, Analysis, Transfer and Management ....................................................... 9 2.1. Site Descriptions .................................................................................................... 9 2.1.1. Lake/Watershed Characteristics ............................................................... 9 2.1.2. Study Beaches........................................................................................... 9 Figure 2-1: Watershed Site Descriptions ................................................ 11

3.

Collection of Different Data Components ........................................................................ 12 3.1. Water Quality and Meteorological Data .............................................................. 12 3.1.1. Continuous Monitoring........................................................................... 12 Figure 3-1: Schematic Diagram of the U.S. Geological Survey Monitoring Station and Instrumentation.................................... 13 Table 3-1: Physical and Chemical Field Parameters .............................. 14 3.1.2. Fixed-Interval Monitoring ...................................................................... 15 3.1.3. Event-Based Monitoring......................................................................... 15 3.2. Microbiological Data ........................................................................................... 16 3.2.1. Selection of Microorganisms .................................................................. 16 3.2.2 Test Methods for Indicator and Pathogenic Microorganisms................. 17 Table 3-2: Biological Parameters, Analytical Methods and Laboratory Performing Analyses ................................................................. 19 Table 3-3: Field & Laboratory Data Collection...................................... 20 Table 3-4: Summary of Laboratory Program.......................................... 21

4.

Data Management, Analysis and Electronic Data Transfer .............................................. 21 4.1. Data Flow and Information Management ............................................................ 22 4.2. Information Management, Equipment and Security ............................................ 23 4.3. Data Analysis....................................................................................................... 24 4.3.1. Censored Data Analysis.......................................................................... 24 4.3.2. Statistical Analysis.................................................................................. 24 Table 4-1: Description of independent variables used in the discriminant function and regression.............................................................. 25

5.

Quality Assurance/Quality Control................................................................................... 26 5.1. Sample/Data Collection QA/QC.......................................................................... 27 5.1.1. Field Data Collection QA/QC................................................................. 28 5.2. Quality Control Checks ....................................................................................... 29 5.2.1. Blanks 29 5.2.2. Replicates................................................................................................ 30 5.2.3. Initial and Ongoing Precision and Recovery .......................................... 31 5.2.4. Laboratory Matrix Spike Sampling ........................................................ 31 5.2.5. Field Matrix Spike Sampling.................................................................. 31

i

5.2.6. 6.

Positive and Negative Controls............................................................... 32

Findings ............................................................................................................................ 32 6.1. Data Overview and Summary.............................................................................. 32 6.2. Statistical Analysis of Pathogen Results.............................................................. 32 Table 6-1: Summary of Olbrich Beach Sample Results ......................... 34 Table 6-2: Summary of Spring Harbor Beach Sample Results .............. 35 Table 6-3: Summary of Vilas Beach Sample Results ............................. 36 Table 6-4: Summary of All Study Beach Sample Results ...................... 37 6.3. Indicator Bacteria Results .................................................................................... 38 Figure 6-1: Olbrich Indicator Bacteria 2002-2003 ................................. 39 Figure 6-2: Spring Harbor Indicator Bacteria 2002-2003....................... 40 Figure 6-3: Vilas Indicator Bacteria 2002-2003 ..................................... 41 Table 6-5: Correlations Coefficients Between Bacterial Indicators ....... 42 Table 6-6: 2002 -2003 Average Indicator Values................................... 42 Table 6-7: Samples Exceeding Beach Closure Limits (as number and percent of all samples) ............................................................... 43 6.4. Pathogens in Beach Water as Compared to Indicator Bacteria Levels................ 43 Table 6-8: Indicator Organism Summary Statistics when all Pathogen Results are Negative or When any Pathogen Result is Positive 43 Figure 6-4: Number of pathogen positive results found in various E. coli concentration ranges .................................................................. 45 6.5. F+ Coliphage Sampling Results .......................................................................... 45 Table 6-9: F+ coliphage sampling results at three beaches in Madison, WI from June, 2002 through September, 2003.......................... 46 6.6. Beach Closings and Illness Reports..................................................................... 46 6.7. E. coli O157:H7 Sampling Results ...................................................................... 46 Table 6-10: Special sampling for E. coli O157:H7 during the summer of 2003 ............................................Error! Bookmark not defined. Table 6-11: Presumptively positive detection of E. coli O157:H7 at three beaches in Madison, WI from August, 2002 through September, 2003 ............................................Error! Bookmark not defined. Table 6-12: Presumptively positive detection of E. coli O157:H7 in waterfowl feces near Vilas Beach in Madison Wisconsin on July 31 and September 24, 2003 ....................................................... 48 Figure 6-5: Results of Grab Samples for Presumptive Presence of E. coli O157:H7 in Lake Wingra During the Summer of 2003 ............ 49 Figure 6-6: Results of Samples for Presumptive Presence of E. coli O157:H7 Collected on a Grid Throughout Lake Wingra on September 3 and 4, 2002 and Results of Samples Collected for prEsumptive Presence of E. coli O157:H7 from Surface Water and Storm Sewers that Discharge into Lake Wingra on September 12, 2003 ................................................................... 50 6.8. Predictive Value of Indicator Organisms and Environmental Parameters........... 51 6.8.1. Censored Data Analysis.......................................................................... 51 Figure 6-7: Results of censored data analysis for enterococci at Olbrich (triangles are censored values, dots are uncensored) ................. 51 Figure 6-8: Results of censored data analysis for E. coli at Spring Harbor (triangles are censored values, dots are uncensored) ................. 52 6.8.2. Phase 1 Regressions................................................................................ 52 Table 6-13: Results of the Phase 1 Regression Analyses ....................... 52

ii

Phase 2 Regressions................................................................................ 54 Table 6-14: Results of the phase 2 regression analyses .......................... 54 Discussion............................................................................................................ 55 6.8.3.

6.9. 7.

Public Communication ..................................................................................................... 57 7.1. Communication Methods..................................................................................... 57 7.1.1. Website Addresses.................................................................................. 58 7.1.2. Links to Other Websites for Additional Information.............................. 59 7.1.3. Other Public Communication Methods .................................................. 59 7.2. Community Outreach Activities .......................................................................... 59 7.3. Meetings, Community Events, Exhibits .............................................................. 60 7.4. Educational opportunities .................................................................................... 61 7.5. Program Changes................................................................................................. 61

REFERENCES ........................................................................................................................................... 62 ATTACHMENTS....................................................................................................................................... 67 List of Acronyms/Abbreviations.................................................................................................... 67 Matrix Spikes & Recovery Limits ................................................................................................. 69 Initial Precision & Recovery Data for Methods used in EMPACT Study..................................... 70 Ongoing Precision and Recovery Data for Male Specific Coliphage (EPA Method 1602) .......... 71 Beach Closure Sign........................................................................................................................ 74 Development of a Rapid Detection Method for Waterborne E. coli O157:H7 Poster................... 75 National Environmental Health Association Presentation............................................................. 77 FOLW Kiosk Display (2 Posters) .................................................................................................. 78 FOLW Newsletter.......................................................................................................................... 80 Microbial Contamination Remains a Mystery .................................................................. 80 Vilas Kiosk Text ............................................................................................................................ 81 Beaches and Your Health.................................................................................................. 81 How Does Water Quality Relate to Public Health?............................................. 81 Monitoring Beach Water Quality: EMPACT Beach Study ................................. 81 How You Can Help Protect Madison’s Beaches ................................................. 82 Nuestras Playas y su Salud ............................................................................................... 82 Relación entre calidad de Agua y la Salud Pública ............................................. 82 Determinando Calidad de Agua en las Playas: Estudio “EMPACT” .................. 83 Como Ud. Puede Ayudar a Protejer Nuestras Playas .......................................... 84

iii

SUMMARY/ACCOMPLISHMENTS An interdisciplinary team representing city public health officials, state and federal governments and the community have worked together to perform an extensive beach water quality study on microbial, physicochemical, and weather-related parameters in the City of Madison, Wisconsin. The goals and accomplishments were as follows: GOALS • Expand the past city recreational beach monitoring program to develop data-driven criteria for beach closures by evaluating a combination of rapid and sensitive tests for the detection of microbial indicators and pathogenic microorganisms • Develop a rapid and sensitive method for the detection of pathogenic E. coli 0157:H7 • Apply coliphage genotyping to determine the source of fecal contamination. • Apply sensitive tests for the detection of pathogens responsible for enteric illness: Giardia, Cryptosporidia, Salmonella and E. coli 0157:H7 • Determine the correlations between (1) existing environmental monitoring parameters, (2) occurrence of pathogens, and (3) meteorological, physical and water quality data collected by remote automated monitoring stations • Consider mathematical constructs for modeling pathogen occurrence • Develop innovative partnerships with community groups and agencies to facilitate dissemination of water quality data and beach closure decisions • Construct a water quality database with dynamic query capability for ready access to the public ACCOMPLISHMENTS • The study focused on developing and applying innovative environmental monitoring tools for the detection of emerging pathogens and novel microbial indicator microorganisms combined with realtime monitoring of environmental conditions, and development of an effective communication infrastructure for dissemination of data to the public. The project provided an enhanced ability to provide real time, user friendly, state-of-the-art water quality information to the public. • A large database of indicator and pathogen occurrence at inland beaches serving a major population center in the Northern Midwest was created. The collected data was loaded into Progress and Oracle databases. The continuously monitored data was made available on the world-wide-web within an hour of collection and the microbial indicator data following the completion of the analyses. The remaining microbiological analysis results were included on the web after analysis results were finalized. • Additional pages educate and explain the significance of the collected data to the public. • Changes in water quality information were disseminated via media, beach signs and a telephone hotline. • A new sensitive analytical method for detecting E. coli 0157:H7 in recreational waters was developed, modified and implemented. The usability of the method was evaluated. • Public education regarding recreational water quality issues was accomplished by poster presentations (at City-County Office building and lake side), informational materials on the web, during neighborhood fairs and through neighborhood newsletters and display of information in a kiosk at the beach. • Quantitative relationships between microbial indicator data, occurrence of pathogens, and meteorological, physical and water quality data were sought to create a predictive model for real-time assessment of the risk of pathogen occurrence and beach closure decisions. The modeling effort showed that none of the indicators was able to represent the presence or absence of pathogens. However, the bacterial indicator densities showed a strong statistically significant association with

1

some environmental parameters. The predictive model used real-time environmental data as predictors for bacterial densities when they exceeded the closure limits.

1.

INTRODUCTION

1.1.

PAST MONITORING EFFORTS AND FUTURE BENEFITS

Beach closures at both coastal and inland beaches have become more and more publicized in recent years. In order to help understand the reasons behind these beach closures, the U.S. EPA has funded several projects to study beach health issues through the Environmental Monitoring for Public Access and Community Tracking program (EMPACT). In addition, in October, 2000, the U.S. congress passed an amendment to the Federal Water Pollution Control Act called the Beaches Environmental Assessment and Coastal Health Act of 2000 to “improve the quality of recreational waters.” This act requires each state to adopt water quality criteria, test and post signs warning beachgoers of contamination of coastal recreation waters, including the Great Lakes. However, this new regulatory strategy does not apply to inland lakes. The City of Madison, Wisconsin administers thirteen public beaches on three inland lakes within the city limits. In the past 2 years, there have been 102 beach closure days at these thirteen beaches. Historically the Madison Department of Public Health has collected samples once or twice a week to perform traditional bacterial indicator tests as an indirect measure of pathogenic contamination. In the 1950’s when this testing was established the most common source of enteric viruses was untreated raw sewage. The preferred indicator organism was fecal coliform, which is easily determined. Since then fecal streptococcus, fecal Enterococcus, E. coli and enterococci have also been used. Some of the common causes for elevated bacteria counts (> 200 CFU or MPN/100 mL) at Madison beaches are: • • • • •

large populations of waterfowl recent heavy rainfall flushing the storm sewer system lack of water movement in the beach area very occasionally, a broken sewer main will pollute the beaches with sanitary sewage possible in-situ incubation of organisms in moist sand and among aquatic weeds.

To minimize recreational water users’ exposure to pathogens, indicator test results are considered in beach closure decisions based in part upon USEPA monitoring guidelines. These guidelines were developed based on epidemiological research that indicated an association between the presence of fecal coliforms, E. coli and the enterococci group with swimming-associated gastrointestinal disease (Cabelli, 1981, 1983 and Dufour, 1984). However, a number of concerns exist with reliance upon bacterial indicators: • • • •

coliform bacteria and certain enterococci bacteria may have non-fecal, environmental sources such as soil, warm, moist beach sand, piles of grass clippings, accumulated organic material in storm sewers, vegetation and certain industrial wastes (Cabelli, 1978; Hendricks, 1978; Whitman, 2003) bacterial indicators exhibit shorter survival times in natural waters than enteric pathogens such as parasitic protozoa and enteric viruses (Feachem, et al., 1983; Reynolds, 2003); rapid fluctuations in bacterial indicator densities due to rainfall and other environmental influences may make weekly testing efforts ineffective; 24+ hour turnaround times required for confirmed bacterial testing results may lead to situations where responses to pathogenic contamination events are ineffective;

2

• •

existing microbial indicator tests do not always accurately reflect the presence or absence of virulent pathogens, and; it is not possible to discriminate between human and animal sources of fecal contaminants using traditional indicator tests.

Current recreational water quality criteria focus on indicator bacteria levels. The state of Wisconsin and the City of Madison use E. coli concentrations to determine when beach advisories are necessary. These and other indicator bacteria are thought to have value as predictors of fecal contamination, pathogen presence, and therefore, risk of swimming associated illness. For this reason, much attention has been directed towards monitoring and predicting E. coli levels in recreational water. A previous study of Lake Erie beaches indicated that important factors in predicting E. coli concentrations included rainfall, wastewater discharges, a resuspension index, and wave heights (Francy and Darner, 1998). A subsequent study on five Lake Erie beaches and one inland beach found that regression models used to predict E. coli concentrations were different for each individual beach. Some of the variables that were important in these regressions included rainfall, wave height, turbidity, current direction, number of birds present, streamflow of tributaries, wind, the number of dry days, and the concentration of E. coli from the previous day (Francy et al., 2002). It seems likely that each individual beach may have a unique set of parameters that drive bacteria concentrations. In fact, in Madison, evidence of this variation at individual beaches has been observed over the course of many years. Epidemiological studies suggest a positive relationship between high concentrations of E. coli and enterococci in ambient waters and incidents of gastrointestinal illnesses associated with recreational activities. E. coli and enterococci are currently considered the best indicators of illness Dufour, 1984, USEPA, 1986. Considering that microbial indicators are relied upon heavily for estimating the safety of recreational waters, an important issue to focus on is the correlation between microbial indicators and pathogen presence. However, efforts to generate quantitative relationships between the indicator organisms and enteric pathogens have largely been unsuccessful. There is little data on this subject for recreational waters, but for surface waters in general, recent research suggests that most microbial indicators are not useful as reliable predictors of pathogen presence (Horman et al., 2004, Lemarchand and Lebaron, 2003, LeChevallier, 2002). Results from some studies have suggested that bacteriophages have some utility as an indicator of enteric viruses (Havelaar, et al., 1993, Lewis, 1995) but other studies have not found useful correlations (Horman et al., 2004, Lewis, 1995). The objective of this report is to present results of pathogen, fecal indicator, and environmental variable monitoring data collected during the 2002 and 2003 swimming seasons, how these parameters are related, and how monitoring results were disseminated to the public in an efficient and useful manner. Samples were collected from 3 beaches, each located on a different inland lake in Madison. Four different fecal indicators, four different pathogens, and a suite of environmental variables that can be measured and reported real-time, were analyzed. The four organisms chosen to represent pathogen presence were Cryptosporidium, Giardia, Salmonella, and E. coli O157:H7. Three historically used fecal indicator variables, fecal coliform, E. coli, and enterococci, and one nontraditional indicator, F+ coliphage, were used as potential indicators of pathogen presence. Meteorological, physical, hydrologic and water quality data were collected by automated monitoring stations. Much of the data were disseminated over the world-wide-web for timely public access. The constructed database will be available for use in the continuing beach monitoring program and for future directed studies. Other vehicles for dissemination of data included local media outlets, telephone hotline and beach signs. After the data collection period, data was analyzed to determine relations between microbial indicators and pathogens, between environmental, indicators and pathogens, and between environmental variables and microbial indicators. Quantitative relationships between microbial indicator data, occurrence of pathogens, and meteorological, physical and water quality data were sought to create a predictive model for real-time assessment of the risk of pathogen occurrence and beach closure decisions. Results of this study provide a regression model to help 3

predict the probability of a water quality criteria exceedance based on E. coli concentrations at Madison beaches. A uniquely important benefit derived from the research has been the cooperation and partnerships that were established between the many units of government involved. The federal government, in the form of the United States Geological Survey (USGS), has played an active role in the technical aspects of the project by providing innovative sampling of weather events, real-time environmental data collection, data analysis, and model development. The State of Wisconsin was represented by the technical efforts and analytical capabilities provided by the Wisconsin State Laboratory of Hygiene (WSLH). Local units of government were included via the MDPH’s leadership and testing roles, and the involvement of the City of Madison Information Systems. Neighborhood associations and citizen action groups, as well as the university community, including the University of Wisconsin Health Services, the University Extension, and the community, including the Friends of Lake Wingra (FOLW) participated in the communication and outreach activities. The team approach serves as an excellent model for other organizations to use in addressing the complex issue of swimming water quality. An immediate as well as long term benefit of the study is the creation or a water quality information dissemination system that is not only accessible, but actually is accessed by large and diverse segments of the community. This dissemination system ranges from a high tech interactive WEB page to the more traditional methods. The construction of a system for collection and communication of swimming water quality information to the users of the Madison lakes is a welcomed long-term benefit of the project. The access to understandable data for making good decisions related to recreational use of Madison lakes is a lasting benefit and outcome of the study.

1.2.

MADISON BEACH EMPACT TEAM

One of the objectives for the EMPACT project was development of cooperative partnerships between government agencies and community groups to facilitate dissemination of water quality data and beach closure decisions. The interdisciplinary team of city public health officials, technical personnel for the state and city public health laboratories, scientists representing federal agencies and neighborhood association groups met several times during planning and implementation phases of the project sharing the collective expertise. Table 1-1 indicates the EMPACT project personnel. Figure 1-1 shows the organizational chart. Madison Department of Public Health (MDPH) had overall responsibility for all phases of the EMPACT project. MDPH performed field tasks, and was responsible for project management. The U.S. Geological Survey was responsible for the continuous monitoring of physical/chemical parameters in the field, data analysis and development of regression models. The WSLH developed methods and performed all pathogen testing. All project partners discussed and formulated the fibal study design together. The community partners performed various outreach and educational activities. Madison Department of Public Health The Madison Department of Public Health is a full-service Health Department. The laboratory is part of the Environmental Health Division. The Environmental Technical Services Section includes a chemistry laboratory, a microbiology laboratory, and field services unit.

4

Wisconsin State Laboratory of Hygiene The Wisconsin State Laboratory of Hygiene was established in 1903 to provide public and environmental health education, research and testing services and is only one of two public health testing laboratories in the United States directly affiliated with a state University. The WSLH is a full service laboratory performing microbiological and chemical, including radiochemical testing services. Senior staff hold tenured academic positions within the University of Wisconsin – Madison from departments such as Preventive Medicine and Civil and Environmental Engineering, and maintain an active role in the campus teaching mission. The pathogen testing for the EMPACT project is performed in newly constructed state-of-the-art facility. U.S. Geological Survey Created by an act of Congress in 1879, the USGS has evolved over the ensuing 120 years, matching its talent and knowledge to the progress of science and technology. Today, the USGS stands as the sole science agency for the Department of the Interior. The members from the Wisconsin District of the Water Resources Division of the USGS are located in Middleton, Wisconsin. City of Madison Information Systems Department The City of Madison maintains an information management system, including a secure, reliable, maintainable, and supportable computer resource network. The IS Department staff develop, operate, and maintain the system. Together with USGS, the IS Department constructed a database for this project. The IS Department developed world wide web (WEB) based communication tools, and disseminated relevant information to the general public and scientific community via WEB pages. Role of Stakeholders/Community Partners:

“Promoting a healthy Lake Wingra through an active watershed community.” The primary role of the stakeholders/community partners has been to assist in the communications aspects of the project. The researchers have actively recruited partners throughout the project. The two community partners successfully committed were the University of Wisconsin-Extension (UWEx), and the Friends of Lake Wingra (FOLW). UWEx Educators Network provided access to an established network of educators from kindergarten through college to help get lake water quality information out to a large community segment. The University of Wisconsin Environmental Health Office provided testing data from the two swimming areas the University operates on Lake Mendota. The FOLW provided a direct conduit to the neighborhood associations and involved citizens interested in Madison lake water quality issues. The association of FOLW with Edgewood College, Edgewood Campus School and Edgewood High School are also fostered additional community involvement opportunities. The involvement of these two groups has enablef outreach into the community and enhanced the understanding and appreciation of the lakes and their associated watersheds.

5

Table 1-1: EMPACT Project Key Personnel Position USEPA Project Officer

Contact Madalene Stevens Project Officer USEPA 1200 Pennsylvania Ave, NW, 8722R Washington, DC Phone: 202-564-2246 Fax: 202-565-2443 e-mail: [email protected] Tommye Schneider Principal Investigator Madison Department of Public Health. 210 Martin Luther King, Jr, Blvd, Room 507 Madison, WI 53709 Phone: 608-294-5306 Fax: 608-266-4858 e-mail: [email protected] Kirsti Sorsa, PhD Project Manager Madison Department of Public Health. 210 Martin Luther King, Jr, Blvd, Room 507 Madison, WI 53709 Phone: 608-294-5356 Fax: 608-266-9730 e-mail: [email protected] David Faust Co-investigator City of Madison Information Services. 210 Martin Luther King, Jr, Blvd, Room GA-4 Madison, WI 53709 Phone: 608-267-4909 Fax: 608-261-9289 e-mail: [email protected] Jon Standridge Co-Investigator Wisconsin State Laboratory of Hygiene 2601 Agriculture Dr. PO Box 7996 Madison, WI 53707-7996 Phone: 608-224-6209 Fax: 608-224-6213 e-mail: [email protected]

MDPH Principal Investigator

MDPH Project Manager

City of Madison Major Co-Investigator

WSLH Major Co-Investigator

6

Position USGS Major Co-Investigator

Contact Steven Corsi Co-Investigator United States Geological Survey 8505 Research Way Middleton, Wisconsin 53562-3581 Phone: 608-821-3835 Fax: 608-821-3817 e-mail: [email protected] John Walker Co-Investigator United States Geological Survey 8505 Research Way Middleton, Wisconsin 53562-3581 Phone: 608-821-3853 Fax: 608-821-3817 e-mail: [email protected] Robert Waschbusch Co-Investigator United States Geological Survey 8505 Research Way Middleton, Wisconsin 53562-3581 Phone: 608-821-3868 Fax: 608-821-3817 e-mail: [email protected] James Lorman, PhD Professor of Biology Edgewood College 1000 Edgewood College Drive Madison, WI 53711 608-663-6921 (phone) 608-663-2339 (fax) e-mail: [email protected] Rick Johnson University of Wisconsin, Madison 608-262-2986 (phone) e-mail: [email protected]

USGS Major Co-Investigator

USGS Major Co-Investigator

Friends of Lake Wingra Liaison

University of Wisconsin Environmental Health Liaison

7

Figure 1-1: Project Team Organizational Chart Madison Department of Public Health (MDPH)

Sample Collection Event Samples

Fixed-Interval Samples

Sample Analysis E.coli O157:H7, coliphage, Salmonella,Cryptosporidium, Giardia, Serotyping

Public Communication

E.coli, fecal coliforms, enterococci

WSLH

MDPH

MDPH

USGS

Analysis Results Real-Time Beach Data USGS

USGS

Madison IT Division

Community Partner Groups

Study Progress and Findings

MDPH

Madison IT Division

WSLH Madison Public Web Page

1.3.

Madison Public Web Page

PUBLIC

OBJECTIVES OF THE PROJECT

Beach Monitoring in Madison, Wisconsin – The objective of the comprehensive beach water quality monitoring program at three Madison lakes was to improve tools to identify early indicators of health risks from pathogenic organisms and help define beach conditions through data based decision-making. The goal was to collect relevant, high quality environmental data and deliver it to the public as rapidly as possible. The study focused on the application of innovative environmental monitoring tools to determine changes in water quality which may result in adverse health effects to those who rely on these resources for recreation. The partners performed an extensive study on beach water quality for physical, microbial and weather-related parameters. Automated, remote-monitoring instruments collected real-time rainfall, wind speed and direction, air and water temperature, solar radiation, turbidity, dissolved oxygen, and chlorophyll content data. Water samples were collected and analyzed for a battery of fecal indicators (fecal coliform, E. coli, enterococci and F+ coliphages) and pathogens (E. coli 0157:H7, Salmonella, Cryptosporidium and Giardia). A sensitive gene probe technology was applied to discriminate between human and animal sources of fecal pollution. Additionally, the project facilitates dissemination of water quality data and beach closure decisions for the public. Specific objectives of the study included the following: • • •

Expansion of the current city recreational beach monitoring program to develop data-driven criteria for beach closures by evaluating a combination of rapid and sensitive tests for the detection of microbial indicators and pathogenic microorganisms Development of a rapid and sensitive method for the detection of pathogenic E. coli 0157:H7 Application of coliphage genotyping to determine the source of fecal contamination.

8

• • • • •

Application of sensitive tests for the detection of pathogens responsible for enteric illness: Giardia, Cryptosporidia, Salmonella and E. coli 0157:H7 Determination of the correlations between (1) existing environmental monitoring parameters, (2) occurrence of pathogens, and (3) meteorological, physical and water quality data collected by remote automated monitoring stations Consideration of mathematical constructs for modeling pathogen occurrence Development of innovative partnerships with community groups and agencies to facilitate dissemination of water quality data and beach closure decisions Construction of a water quality database with dynamic query capability for ready access to the public

2.

DATA COLLECTION, ANALYSIS, TRANSFER AND MANAGEMENT

2.1.

SITE DESCRIPTIONS

The City of Madison, Wisconsin contains three recreational lakes with over 20 miles of shoreline Figure 2-1 shows the map of the sites. The lakes receive contaminanats from non-point sources only. The lakes are heavily used for recreational activities including sail boating, power boating, wind surfing, water skiing, swimming, scuba diving, canoeing, kayaking, fishing and jet skiing. The citizens of Madison treasure the recreational value of their lake resources.

2.1.1. Lake/Watershed Characteristics Lake Mendota covers 9,842 acres with a maximum depth of 83 feet, has five City beaches, and seven non-City swimming areas. Land use of the Lake Mendota watershed includes 62.6% agricultural influence, 16.5% grassland, forest, and open spaces, 12.6% open water and wetlands, and 8.3% urban influence. The University of Wisconsin-Madison borders a portion of the south shore of the lake, the City of Madison borders the remainder of the south shore as well as the east shore, the City of Middleton borders the West shore, and a mix of suburban, agriculture, and park lands border the north shore. Lake Monona covers 3,274 acres with a maximum depth of 74 feet, has 7 City beaches and 1 non-City swimming area. Land use of the Lake Monona watershed includes 54.7% agricultural influence, 17.5% grassland, forest, and open spaces, 13.3% open water and wetlands, and 14.4% urban influence. The City of Madison surrounds the entire lake. Lake Wingra covers 345 acres with a maximum depth of 21 feet and has one City beach. Land use of the Lake Wingra watershed includes 0.5% agricultural influence, 31.8% grassland, forest, and open spaces, 14.1% open water and wetlands, and 53.6% urban influence. The Vilas beach is near the outlet from the lake, adjoins the County Zoo and is by far the most heavily used beach operated by the City of Madison. The lake is located completely within the City of Madison with the entire south side of the lake bordered by the University of Wisconsin Arboretum and the north side with mainly influence from residential and commercial areas and Madison City parks.

2.1.2. Study Beaches Three public swimming beaches were selected for monitoring, one on each of the three lakes representing differing lake morphology and beach user levels. The three beaches selected and their locations were: Vilas Beach: Lake Wingra; high user levels; Latitude: 43°03’ 29”, Longitude: 89° 24’ 45”. Numerous storm sewer effluents surrounding the beach area have potential to significantly impact beach water quality.

9

Olbrich Beach: Lake Monona; medium user levels; Latitude: 43°05’ 16”, Longitude: 89° 19’ 50’”. The mouth of Starkweather Creek is just north of the beach, and the runoff from this creek has the potential to significantly impact beach water quality. Land use in the rapidly urbanizing Starkweather Creek watershed currently includes 28.8% agricultural influence, 29.0% grassland, forest, and open spaces, 4.9% open water and wetlands and 37.3% urban influence. Spring Harbor Beach: Lake Mendota; low user levels; Latitude: 43°04’ 58”, Longitude: 89° 28’ 09”. Spring Harbor storm sewer enters Lake Mendota just to the east of the beach and has the potential to significantly impact beach water quality. The Spring Harbor storm sewer drainage area is completely developed and includes 1.6% agricultural influence, 33.9% grassland, forest, and open spaces, 0.3% open water and wetlands, and 64.2% urban land use. The mouth of Pheasant Branch, an 18.3 mi2 watershed, is located approximately 1.7 miles northeast of the beach and has potential to impact the beach water quality as well. Land use in the rapidly urbanizing Pheasant Branch watershed currently includes 56.4% agricultural influence, 23.1% grassland, forest, and open spaces, 2.5% open water and wetlands, and 18.1% urban influence.

10

Figure 2-1: Watershed Site Descriptions

11

3.

COLLECTION OF DIFFERENT DATA COMPONENTS

3.1.

WATER QUALITY AND METEOROLOGICAL DATA

Three types of monitoring were performed at each beach: continuous, fixed-interval, and event-based monitoring. Continuously monitored data included meteorological parameters, water quality parameters, and physical parameters. Samples were collected for microorganisms using both a fixed-interval and an event-based schedule. Tables 3-2 through 3-4 summarize the field and laboratory program. During the 2 swimming seasons of the study (from approximately Memorial Day to Labor Day 2002 and 2003), numerous water quality, meteorological and physical parameters were continuously monitored by the USGS, National Weather Service and NOAA. City of Madison beach personnel recorded daily swimmer and waterfowl counts. Madison Department of Public Health personnel collected fixed-interval samples at each beach 5 days per week for indicator organism and pathogen analysis. Custom-designed automatic water quality samplers collected indicator organism and pathogen samples from twelve events at each beach. Events were defined as rainfall runoff periods, high turbidity periods, periods of high user counts or high wind/wave periods.

3.1.1. Continuous Monitoring Continuously monitored parameters included water temperature, air temperature, precipitation, turbidity, specific conductance, dissolved oxygen, wind speed and direction, wave height, direct, diffuse and ultraviolet b solar radiation. Specific equipment used to collect data include YSI 6000 series multiparameter water quality sensors, a number of Campbell Scientific instruments (CR10X dataloggers, anemometers, temperature probes, storage modules, relay drivers, and telephone modems), tipping bucket rain gages manufactured by Rainomatic, and commercially available refrigerators and high-volume peristaltic pumps. Table 3-1 describes the purpose of some of the measurements A small building was installed at each site and outfitted with the following equipment: • • • • • • • • • •

datalogger--served as the station controller and data recorder from the various monitors, modem and telephone–for remote communications, tipping bucket raingage–for precipitation measurement, air temperature probe–for air temperature measurement a YSI 6600 multi-parameter water quality meter–for continuous measurement of water temperature, turbidity, specific conductance and dissolved oxygen Pressure transducers for measurement of wave height and water level a modified ISCO 3700 refrigerated automatic water-quality sampler equipped with: 4, 10 liter plastic sample collection bottles, Teflon lined sample collection tubing, a valve system for improved rinsing of the sample tubing

A detailed diagram of a monitoring station is shown in Figure 3-1. In addition to the data collected at each beach, wind speed and direction data were collected at the USGS office in Middleton, Wi. west of the beach locations and by the National Weather Service at the Dane County Regional Airport in Madison Wi. east of the beach locations. The National Oceanic and Atmospheric Administration (NOAA) Integrated Surface Irradiance Study (ISIS) provided solar radiation data for the Madison area.

12

The electronic datalogger at each station was programmed to record water quality measurements every 10 minutes and trigger water sample collection when required. Data was recorded using internal memory of the datalogger, retrieved via modem each hour and made available through the USGS World Wide Web site within 10 minutes of retrieval. The web presentation included graphs of each individual parameter for the previous 31 days. Figure 3-1: Schematic Diagram of the U.S. Geological Survey Monitoring Station and Instrumentation

Datalogger - controls station functions, triggers samples and stores stream level, water temperature, rain gauge and sampler data

Tipping bucket rain gauge Tef lon-lined inlet and out let lines f or f lowt hrough s ample c ollec t ion c ham ber

R elay driv er to t rigger sampler and to control c ellular phone

Modem for remote access to data

Valve system

Storage Module data backup system

R ef rigerat ed Aut omatic W at er-Qualit y Sampler 12 Volt Battery 4 10-lit er bot t les

Wav e height sensor

To Lake

Multiparameter water quality meter

USGS automated sampling system To Lake

13

Table 3-1: Physical and Chemical Field Parameters Parameter Cumulative Rainfall

Turbidity

Specific Conductance

Dissolved oxygen (DO)

Chlorophyll

Purpose of Use Cumulative rainfall represents a running total of rainfall from an arbitrary starting time. To determine the precipitation for a given period (for instance a thunderstorm), subtract the cumulative rainfall value at the beginning of the period from the cumulative rainfall value at the end of the period. Rainfall for this project was measured with a tipping bucket raingage which tips a bucket and sent an electrical pulse every 0.01 inch of rainfall. This electrical pulse was registered by a datalogger and added to the current cumulative rainfall value. Turbidity is an optical property measuring the clarity of water. The greater the amount of total suspended solids (TSS) in the water, the murkier it appears and the higher the measured turbidity. The clarity of a natural body of water determines its condition and productivity. Suspended matter, such as soil particles, organic matter and microscopic organisms cause turbidity. It is measured in nephelometric turbidity units (NTU). Crystal clear water has a turbidity of less than 1 while the turbidity increases with cloudiness of the water. Specific conductance estimates the electrical conductivity of the water if it were at a temperature of 25˚C. This is a measure of the total dissolved salts or the total amount of dissolved ions in the water. Dissolved salts can originate from the soil and rocks as well as agricultural and urban runoff. Examples include dissolution of carbonate minerals from limestone, nitrogen and phosphorus from fertilizers, dissolved herbicides or chloride from roadsalt. The conductivity sensor consists of two metal electrodes that are 1.0 cm apart. The electrodes are exposed to water and a constant voltage is applied across the electrodes. An electrical current flows through the water due to this voltage and is proportional to the concentration of dissolved ions in the water. The more ions present, the more conductive the water is, and the higher the specific conductance value. Distilled or deionized water has very few dissolved ions and so the specific conductance is at or near 0.0 µS/cm. Specific conductivity is expressed as microSiemens per centimeter (µS/cm). Dissolved oxygen consists of microscopic bubbles of oxygen gas in the water. Like terrestrial animals, fish and other aquatic organisms need oxygen to live. Oxygen is also needed for many chemical reactions that are important for lake functions. Oxygen is produced during photosynthesis throughout the daylight hours and consumed 24 hours a day plant and animal respiration and decomposition of organic matter. Oxygen can also enter a lake through the air, groundwater and surface water. DO measurements are expressed in mg/L. Oxygen concentrations are much higher in air, which is about 21% oxygen, than in water, which is a tiny fraction of 1 percent oxygen. Chlorophyll is the green coloration (a pigment) that all green plants, including algae, produce. Chlorophyll captures the energy of sunlight and uses it to synthesize carbohydrates from CO2 and water during photosynthesis. It is a useful measure of algal biomass (living mass) in the lake water.

14

3.1.2. Fixed-Interval Monitoring The City of Madison Department of Public Health (MDPH) collected all fixed-interval samples using the department’s standard operating protocols for beach sampling. At a point near the center of the swimming area of the beach, MDPH personnel waded into the beach to a depth of approximately 2 feet. Sterilized 150 mL glass bottles were used for the indicator bacteria analyses. Polyethylene bottles of 4 and 20 L capacity were used for the pathogen analyses. Wide-mouth 1 L polyethylene bottles were used to fill the large containers. A sample bottle was inverted, submerged to a depth of about 1 foot and then turned upright and allowed to fill. The bottle was then capped and placed in a cooler with ice packs and transported to the MDPH laboratory for analysis. Water samples for fecal indicators (fecal coliform, E. coli, enterococci) were collected 5 times per week (weekdays) and 3 weekends during each swimming season. E. coli O157:H7 and coliphage samples were collected 3 times per week. and Cryptosporidium, Giardia and Salmonella once per week. Indicator bacteria samples were collected on 129, 135, and 133 days form Spring Harbor, Olbrich Park and Vilas Park beaches, respectively during the two swimming seasons. Cryptosporidium, Giardia, and Salmonella samples were collected on 24, 22-23, and 24-25 days form Spring Harbor, Olbrich Park and Vilas Park beaches and E. coli O157:H7 and coliphage samples were collected on 67 days from each of the three beaches. Field personnel also measured water temperature and made field observations such as presence of weeds and algae, waterfowl, bird or other animal excrements, swimmer and waterfowl activity, and noted diaper/fecal accidents. The information was included on the chain-of-custody forms that accompanied the samples. Additionally, City of Madison lifeguards monitored bather loads and bird populations.

3.1.3. Event-Based Monitoring The USGS-Water Resources Division performed all event sampling, which consisted of intensive sampling during periods when selected environmental parameters experienced high short-term variability. Refrigerated automatic samplers were used to collect water samples during event periods. The automatic samplers used for this project were ISCO 3700 samplers that were modified by the USGS specifically for microorganism sample collection. The samplers used a peristaltic pump to withdraw water from the lake through Teflon-lined tubing into a valve system. The valve system flushed the sample tubing for 40 seconds (approximately 2.5 liters) before depositing water into refrigerated sample bottles. After subsample collection the sample tube was purged. Event sample collection was initiated based on rainfall, turbidity, wind speed/wave height or anticipated bather count. Time based sub-samples were then collected and composited over the event duration. Sampling in this manner resulted in mean concentrations for the sampling periods. Water samples collected with automatic samplers were split into collapsible 1-L plastic cubitainers for pathogen analysis and sterile 250 ml glass bottles for indicator analysis using a Teflon-lined stainless steel churn splitter. Detailed descriptions of these splitters and their use can be found in Ward and Harr (1990). All indicator organisms and pathogens included in the fixed interval sampling were collected and analyzed for each sampling event. In addition, continuously monitored parameters were available on a real-time basis throughout the sampling period. Water samples were processed and delivered to the WSLH for pathogen testing and the MDPH for indicator testing (Table 3-2). A total of 12 event periods were sampled at each of the EMPACT study beaches throughout the two swimming seasons.

15

3.2.

MICROBIOLOGICAL DATA

3.2.1. Selection of Microorganisms Several criteria were used to select the pathogens and indicator microorganisms for inclusion in the study. Pathogenic microorganisms were selected based on the severity of the disease they can cause, their relevance as human pathogens in Wisconsin recreational waters, and the availability of robust detection methods. Indicator microorganisms were selected based on their current and historical use in the analytic laboratories of the city’s Department of Public Health (MDPH) and the state’s public health testing laboratory (WSLH). Coliphages were selected based on their potential utility for indexing the presence of pathogenic bacteria and viruses. Established, field-validated assay methods were used when available to test for Salmonella, Cryptosporidia, Giardia, coliphage, E. coli, fecal coliforms, and enterococci. An innovative and sensitive method for the detection of E. coli O157:H7 was developed specifically for this study. Fecal Coliform The fecal coliform group of bacteria has been routinely used since the 1960’s as an indicator of fecal contamination of recreational waters. The general concept is that detection of this group of bacteria in the source water indicates fecal contamination. However, the fecal coliform group includes not only Escherichia coli emanating from fecal origins, but also includes some thermo-tolerant Klebsiellae that may come from non fecal sources. Fecal coliform testing was included in this project to provide a link to the historic data available from previous testing at the Madison beaches. E. coli Beginning in the mid-1980’s, new technologies were developed allowing water microbiologists to directly enumerate E. coli from water samples. These new technologies detect E. coli originating from a fecal source without the confounding interferences associated with the fecal coliform test. Early work by Cabelli 1981 and Dufour 1984 showed that detection of elevated levels of E. coli in fact, provided a better association with disease occurrence in swimmers than did fecal coliform. E. coli was included as a reliable fecal indicator that is widely used for recreational water evaluations. Enterococci Enterococci are small round gram positive bacteria found in the intestinal tracks of warm-blooded animals, including humans. The work of Cabelli and Dufour mentioned above suggests that swimming associated gastroenteritis is best predicted by increased levels of enterococci. Coliphages Coliphages are viruses that infect E. coli and other coliform bacteria. These fecal indicators are thought to indicate the presence of a human virus because they are present in human/animal feces and sewage (Havelaar et al., 1993, Sobsey et al., 1996). These bacterial viruses are used as fecal indicators since they are similar in size and morphology to the enteric viruses capable of infecting humans. Some of these bacterial viruses (the male-specific coliphages, also called F+ coliphages) also exhibit survival times which are consistent with those documented for viral pathogens such as poliovirus, hepatitis A virus, rotaviruses and the Calicivirus family (Allwood et al., 2003). The male-specific coliphages are those which infect only male strains of coliform bacteria which express physical appendages used in bacterial sexual conjugation. Since these appendages (sex pili) are expressed only by male strains of bacteria, and only under the elevated temperatures consistent with the mammalian gastrointestinal tract, it is theorized 16

that these coliphages are more likely to represent enteric viruses which may be present in natural waters contaminated with sewage. Salmonella Salmonella is the most frequently isolated causative bacterial agent of gastroenteritis in Wisconsin (Wisconsin State Laboratory of Hygiene verbal comment). It is capable of causing vomiting and diarrhea illness in warm-blooded animals including humans. The organism is a zoonotic pathogen that is often implicated in human infections caused by contact with an infected animal often an avian species. All of the Madison beaches are frequented by a variety of waterfowl. These waterfowl are likely carriers of Salmonella, and thus this pathogen was included as part of the study. While ingestion of water is a possible source of infection, it only occurs when the water is grossly contaminated since the infective dose (the number of bacteria you would need to swallow to get sick) is 10,000 or greater. Low levels of this organism in beach water would be unlikely to cause disease. Cryptosporidia/Giardia Both Cryptosporidia and Giardia are single celled parasites that have been implicated in numerous waterborne disease outbreaks including those resulting from recreational exposures. They may cause diarrhea in warm-blooded animals including humans. Both Craun (1990) and Rose e.t al. (1997) published excellent review articles describing these incidents. The zoonotic nature of the organisms coupled with their low infective dose (the number of cells you would need to swallow to become sick) make them prime candidates for causing disease from recreational exposure to water. Their presence at any level in swimming water is considered a potential risk to swimmers. E. coli O157:H7 Among the pathogenic strains of E. coli, the enteropathogenic and enterohemorrhagic strains are among the most virulent, particularly in children under the age of four. More than 100 serotypes are associated with disease, however the “O” antigen serotype (specifically O157:H7) is the most frequently observed (Murray, et al., 2003). Although most frequently isolated from calves, this serotype has been found in lambs, goats, piglets, cats, dogs, deer, rabbits and humans (Griffin, 1995). This microorganism has been identified in pastures and contaminated ground water and has been recently associated with several high visibility outbreaks of waterborne infectious disease where several dozen deaths occurred (Johnson, et al., 2003, Keene, et al., 1994, Swerdlow, et al., 1992). Ingestion of as little as ten bacterial cells may result in clinical disease in humans, particularly among the immune-suppressed. Hence, in order for detection methods to be robust, they must be capable of detecting low levels of contamination. Unfortunately, current methods require several days of processing for confirmatory results, and routine E. coli testing in water using commercial methods (enzyme-based assays such as ColilertTM) cannot detect it at any concentration. Existing methods for detection of E. coli O157:H7 in water are non-standardized, have poor sensitivity or require enrichment steps, resulting in assay times of more than 72 hours. Hence, the Wisconsin StateLaboratory of Hygiene developed an innovative strategy for detection, based on sample enrichment, concentration by immunomagnetic separation, and immunofluorescent detection using flow cytometry. The time required to obtain presumtive test results is approximately 8 hours. The analysis results can be confirmed in approximately 24 hours.

3.2.2 Test Methods for Indicator and Pathogenic Microorganisms Fecal Coliform

17

The tests available for detecting this group of enteric bacteria are based on the ability of the target organisms to ferment the carbohydrate lactose at the elevated incubation temperature of 44.5°C. The tests were performed at the MDPH laboratory with the membrane filter fecal coliform count (MFFCC) procedure outlined in Standard Methods for the Examination of Water and Wastewater, (APHA, 1998) and detailed in the attached appendices. E. coli E. coli counts were enumerated at the MDPH laboratory using the Colilert Quanti-trayTM (IDEXX) detection system as described in Standard Methods for the Examination of Water and Wastewater. (APHA, 1998) The Colilert test was chosen because of its demonstrated superiority for detecting stressed organisms at the WSLH. Enterococci Enterococci counts were obtained using a proprietary defined substrate enzyme based test system called EnterolertTM. This method is described in detail by Fricker and Fricker (1996) where they conclude that EnterolertTM should be generally accepted as the best technique for enumerating enterococci from water. They found the new method compared favorably to the accepted membrane filter method with no statistical difference found using the paired t-test statistic and the coefficient of variation was r=0.927. Authors of a 1998 study also concluded that the EnterolertTM test was equivalent to the membrane filter technology with a coefficient of variation r=.91 (Abbott et. al., 1998). Salmonella Salmonella testing was performed at the WSLH using the concentration, enrichment, and selective growth techniques followed by serological testing methods described in the 20th edition of Standard Methods for the Examination of Water and Wastewater (APHA, 1998). The protocols allow for some latitude in methods. The researchers used diatomaceous earth enhanced membrane filtration for concentration followed by enrichment in selenite broth. Three different volumes of water were passed through three different filters and inoculated into three different enrichments to provide some idea as to the relative concentrations of the organisms. Enrichments were plated on xylose lysine desoxycholate agar. Suspicious colonies were picked to urea and triple sugar iron agar slants for confirmation. Confirming colonies were further verified as Salmonella by agglutination using polyvalent “O” antisera. Cryptosporidia/Giardia During the first 2 months in 2002 (June 10, 2002 through August 4, 2002), Cryptosporidia/Giardia were enumerated at the WSLH using USEPA performance based method 1623 which included the use of flow cytometry for the separation of cysts and oocysts (USEPA,1999). WSLH replaced the flow cytometry portion of the test with immunomagnetic separation (IMS) for samples starting August 5, 2002 until the end of the study to eliminate algal interference. The WSLH has fine tuned the method for Wisconsin waters incorporating filtration with track etched polycarbonate membranes followed by centrifuge concentration, immunomagnetic separation, and microscopic immunofluorescence detection with confirmation by DAPI staining and differential interference contrast (Archer et al., 1995). Coliphages

18

The researchers used USEPA Methods 1602 for male specific coliphage detection (USEPA, 2001). Additional genetic serotyping analysis using the hybridization assay described by Furuse and further modified by Hsu (1995), was also applied to distinguish between human and animal fecal contamination; Male-specific RNA coliphages are divided into four primary serogroups. Previous literature has indicated that two serotypes are primarily associated with animal feces (I and IV), and the other two are primarily associated with human feces, swine feces and sewage (II and III, with type III phages found exclusively in human fecal sources) (Havelaar et. al., 1990, Cole et. al., 2003). However, a comprehensive study on the waste from all possible sources to the beaches monitored for this study is not currently available and was not conducted in conjunction with this study. Therefore, conclusions based on F+ coliphage data for the study rely on previously published results from separate timeframes and geographical regions. E. coli O157:H7 A sensitive and rapid method for detecting E. coli O157:H7 was developed and implemented specifically for this project. Various concentration techniques including diatomaceous earth and 0.4 micron membrane filtration were evaluated prior to choosing filtration through 142 mm diameter, 0.4 micron polycarbonate, track-etched filters. Exposed filters are placed in 25 mL modified buffered peptone enrichment media and incubated at 37°C for six hours, allowing a three to four log increase in numbers. The organisms are concentrated from the enrichment using immunomagnetic separation with Dynabeads® anti-E. coli O157. Recovered organisms are stained with a fluorescein-labeled antibody highly specific for E. coli O157:H7 and screened using flow cytometry. Samples exhibiting a significant number of fluorescing organisms in the flow cytometer are considered presumptive positive and are plated onto Rainbow® Agar O157 and CHROMagar® O157 plates and incubated for 18 hours. Typical colonies are confirmed by serological identification using O157 antibody in slide agglutination test. Multiple spikes into complex lake water samples have confirmed the sensitivity of the assay down to 10 organisms/500 mL. During the first year of data collection operator judgment was used to decide if the number of fluorescing organisms was sufficient to warrant classifying the sample as presumptively positive. During the second year the events were enumerated by the cytometer and a specific cut off point was used to determine presumptive positives. The fluorescence signals generated by the beach samples were compared to the signals from positive and negative control samples. The Either method is somewhat arbitrary and thus any analysis using the presumptive O157 positivity data should be handled carefully. Table 3-2: Biological Parameters, Analytical Methods and Laboratory Performing Analyses Parameter Biological Constituents Coliphages Cryptosporidia E. coli E. coli 0157:H7 Enterococci Fecal Coliforms Giardia Salmonella

Method EPA 1601/Serotyping EPA1623 Colilert Quanti-tray Enrichment, flow cytometry/immunofluorescence, method under development Enterolert Membrane Filter EPA 1623 Filtration enrichment

19

Laboratory WSLH WSLH MDPH WSLH MDPH MDPH WSLH WSLH

Table 3-3: Field & Laboratory Data Collection Agency USGS

Frequency Continuous Monitoring every 10 min

Parameters Air and water temperature Cumulative rainfall Turbidity Wind speed and direction Wave Height (ft) Average Water Level Specific conductance Chlorophyll Dissolved oxygen

MDPH Sample Collector

Fixed interval sampling 5 - 6 times/week

Lifeguards

Fixed interval sampling 3 times daily

Bather counts Bird counts (ducks, gulls, geese) Bird droppings Raking.discing Visual observations for: color turbidity, algae, weeds, waves, general condition of the beach

Bird counts Droppings Kids with diapers Laboratories MDPH

Fixed interval sampling 5-6 times/week 5-6 times/week 5-6 times/week

E. coli Fecal coliform Enterococci

WSLH

3 times/week 3 times/week 1 time/week 1 time/week 1 time/week

E. coli O157:H7 Coliphage Cryptosporidium Giardia Salmonella

Event sampling Total 51 Total 50 Total 51

E. coli Fecal coliform Enterococci

Total 36 Total 36

E. coli O157:H7 Coliphage

MDPH

WSLH

20

Agency

Frequency

Parameters

Total 34

Cryptosporidium

Total 34 Total 36

Giardia Salmonella

Table 3-4: Summary of Laboratory Program

3

E. coli 0157:H7

# Samples Analyzed 2002 2003 Total Total Proposed Lab SH OB VS SH OB VS WSLH 35 35 35 33 33 34 205 216

3 1 1 1 5 5 5

Coliphages (1601/1602) Coliphages (nucleic acid) Salmonella Cryptosporidia/Giardia Fecal Coliforms E. Coli Enterococci

WSLH WSLH WSLH WSLH MDPH MDPH MDPH

35 7 12 12 64 65 65

35 4 12 12 69 70 70

35 4 12 12 67 68 68

32 5 12 12 63 64 64

32 2 10 11 64 65 65

32 3 13 12 143 143 143 65+74

201 25 71 71 470 475 475

216 72 72 72 216 216 216

E. coli 0157:H7 Coliphages (1601/1602) Coliphages (nucleic acid) Salmonella Cryptosporidia/Giardia Fecal Coliforms E. Coli Enterococci

WSLH WSLH WSLH WSLH WSLH MDPH MDPH MDPH

3 3 1 3 3 3 3 3

4 4 0 4 4 4 4 4

3 3 2 3 3 3 3 3

9 9 0 9 9 9 9 9

8 8 2 8 7 8 8 8

9 9 1 9 8 8 9 9

36 36 6 36 34 35 36 36

36 36 36 36 36 36 36 36

E. coli 0157:H7 Coliphages (1601/1602) Coliphages (nucleic acid) Salmonella Cryptosporidia/Giardia Fecal Coliforms E. Coli Enterococci Total

WSLH WSLH WSLH WSLH WSLH MDPH MDPH MDPH

3 3 0 3 6 11 11 11 264

3 2 0 2 6 17 17 17 301

3 2 0 2 5 14 14 14 280

5 6 2 5 8 15 15 15 325

6 7 0 7 6 17 17 17 328

7 27 4 24 1 3 7 26 7 38 17 91 17 91 17 91 568 2066

32 32 0 32 18 32 32 32 1794

Proposed Frequency per week FI

Number of events sampled 6 Events 6 6 6 6 6 6 6 QA

na na na na na na na na

Microorganism

QA samples include blanks and duplicates 2003 (MDPH 2 field blanks)

4.

DATA MANAGEMENT, ANALYSIS AND ELECTRONIC DATA TRANSFER

This section describes the data generation and flow, including collection, analysis, storage and retrieval, delivery, final repository and maintenance.

21

4.1.

DATA FLOW AND INFORMATION MANAGEMENT

Data management was a comprehensive team effort. The City of Madison, the USGS, and the Wisconsin State Laboratory of Hygiene worked together on a very regular basis to ensure the proper flow of data. The City of Madison is the final reporsitory of all data generated during the project. Continuously monitored water quality, meteorological, and physical data collected by USGS scientists from the monitoring stations was stored by on-site dataloggers. Each beach monitoring station recorded continuously monitored data using the internal memory of a Campbell Scientific CR10X datalogger. The data was backed up immediately with integral storage modules, downloaded on an hourly basis by telephone modem connection, and uploaded directly to the USGS National Water Information System (NWIS) database. The NWIS was the primary repository for the physical and water quality data collected during this project. In addition, a USGS Oracle database was developed for storage and query of the microbiological data. Continuous data reached project web pages within 10 minutes from the time of retrieval. The past 31 days of monitoring data was presented in graphical format as provisional on the USGS real-time web page. The city of Madison beach information web site contained a hyperlink to the USGS graphical presentation of data At the end of the monitoring study, the USGS used calibration data from the continuous sensors for final data corrections, computations and reporting of final records. The MDPH generated microbiological data was entered into an Access database and transferred through custom input forms into the Madison Inforamtion System database. After verification of data integrity, the data were made available on the public web page and transferred to the USGS for population of the Oracle database for use in the research component of the project. The most recent water quality, meteorological, and physical data were extracted from the USGS data to include on the beach information site for each selected beach. Final data were formulated into spreadsheets, and made available for downloading to interested parties. The WSLH entered microbiological data into their local database system (Concurrent DMS) manually where it was transferred to the USGS system using automated file transfer protocols. The USGS used an automated system to upload WSLH data into the NWIS and the project Oracle databases. USGS personnel verified WSLH generated data for integrity and data was transferred using file transfer protocols over the World Wide Web for population of the City of Madison database. As part of the EMPACT project, the City of Madison Information Services Department was charged with creating a database and website to display beach information. The City of Madison IS department maintained a Progress 9.1 relational database management system and programming language to manage and process the data collected for this project. A custom Progress database was constructed for this project. The development software Progress WebSpeed was installed and configured for rapid delivery and displaying of website pages. The Progress AppServer software was installed and configured for use in connecting from the website to the database. The interface was available to simplify the search for relevant data. The necessary servers were used to deploy the software. Not only did use of this database system facilitate a readily useable public interface, but it also ensured the sustainability of capabilities developed during the course of this project after the grant expired. This project achieved the goal to present the City of Madison with an opportunity to develop expanded capabilities to distribute City data to the public via the web. The IS department has benefited greatly by the expertise that was gained from developing this website, and this expertise has been put to use in developing other applications for the City of Madison. 22

Community partners, such as the Friends of Lake Wingra and the University of Wisconsin Extension, who utilized project information, accessed the data via the public web page and by direct communication with MDPH personnel. The University of Wisconsin Environmental Health Office staff provided complementary microbiological data collected from the university beaches. This data was also entered data into the Progress database and made available to the public through the web site.

4.2.

INFORMATION MANAGEMENT, EQUIPMENT AND SECURITY

Equipment used in data retrieval and storage included several systems that are operational and maintained by permanent on-site personnel at each individual facility. The City of Madison used an HP 9000 Unix server running the HP-UX 11.0 operation system for administration of the Progress database. A Compaq Proliant server with Windows NT 4.0 was used to administer the Madison beach information web page. The USGS NWIS database was based on an Ingres database and served by a Sun Microsystems quad 400 server using Sun Solaris 5.6 operating system. For Oracle database administration, multiple Dell servers were in operation using the Windows NT 4.0 operating system. The WSLH used Concurrent 3200-400 computers with Concurrent OS32 R09-02 operating system to serve the Concurrent DMS database. All of the systems were maintained by full-time System Administrators who were responsible for the operation, maintenance, and security of their respective computer networks. The City of Madison incorporated a standard city-wide security system for protection against outside interference. Filters were in place on network routers to control access to selected ports and intrusion detection measures were used. Database security was ensured via the generic access privileges inside the database itself. While the general public had access to the data, modification of data was allowed only for qualified users with proper password access. The password entry system was incorporated in the general design of the web interface layout for outside users. The database itself was stored on a separate Unix server and accessed only via an IP connection and not directly client to server. Network traffic was logged on files located on all primary servers. Backups were automated and occurred on a daily basis using a standard tape rotation scheme. Servers were located in locked rooms with a keycard entry system. This room also employed environmental protection and monitoring, with automatic notification of any environmental conditions exceeding acceptable ranges. The USGS incorporated a rigid security system for protection against outside interference. Security guidelines were standard throughout the USGS and documented for each individual type of operating system in use. Network traffic was logged on files that were located on all primary servers. Filters were in place on network routers to control access to selected ports. All user access required a password for entry to the system. Only selected individuals who were trained and qualified specifically for use of each individual database (Oracle and NWIS) entered data into databases. Only personnel with password access to the system were able to retrieve data. Backups were automated and occurred on a daily basis using a standard tape rotation scheme. Some tapes were stored on-site and others were stored at separate USGS locations in Madison, WI. System administrators controlled access to all servers. The servers were located in locked rooms with a keycard entry system and stringent environmental controls. The WSLH also maintains tight security. Security guidelines have been standardized for the database system and the computers that administer the database. Network traffic was logged on files that were located on all primary servers. All user access was governed by password entry. Only selected individuals who were trained and qualified could enter data into the database. Only personnel with password access to the system could retrieve data. Backups were automated and occur on a daily basis using a standard tape rotation scheme. System administrators controlled access to all servers, and the servers were located in locked rooms with a keycard entry system and stringent environmental controls. 23

4.3.

DATA ANALYSIS

The continuously monitored water quality, meteorological, physical data and indicator organism data were used to develop a probability-based tool for real-time assessment of the risk of pathogen and highlevel indicator organism occurrence. These tools will greatly enhance the base of information used to make beach closures decisions at Madison beaches. Selected data from the continuously monitored parameters was included in the Oracle database to complement the microbiological data for use in the research component of the project and final data analysis and model development. These data along with the microbiological data were used during final data analysis and development of probabilistic models. The Oracle database was instrumental for carrying out many of the tasks during the statistical analysis.

4.3.1. Censored Data Analysis Data sets that contain censored values, or values reported “less than” or “greater than” a detection limit, pose a number of problems for reporting and statistical analysis. For a single detection limit, it is tempting to substitute a single value for each censored value. Depending on the value chosen, the results are almost certain to be biased. In this study, there were very few data values below the detection limit but censoring of the data at the upper end, where the data was reported as greater than a quantification limit was more common. Because high values are what prompt beach closings, only “greater than” values are of concern and the method for estimating their values is important, whereas setting the “less than” values equal to the detection limit should have no impact on the analysis. Consider that the data above the quantification limit belongs to a single statistical distribution, and the only reason the values aren’t known is due to limitations of the analysis. If a single value equal to the quantification limit is chosen as a substitute, the resulting analyses will be biased low. Choosing some multiple of the quantification limit will only be correct if the distribution is symmetrical above the quantification limit and happens to be centered around the chosen multiple of the quantification limit. Thus, simple substitution methods are likely to result in a bias in any statistical analysis. The analysis used in this report is patterned after the fill-in procedure for multiple detection limits discussed by Helsel and Cohn (1988). The general procedure involves ranking the uncensored data from lowest to highest, including all of the censored values. Because the censored values are greater than the quantification limit (2419 MPN per 100mL), the censored values comprise the highest ranks in the dataset. Given the assigned ranks, a cumulative probability can be determined using the Weibul position formula. For the uncensored values, a probability distribution is fit to the data. The resulting distribution can be used to estimate values for each of the censored values based on their assigned rank and resulting cumulative probability. For the E. coli and enterococci data, a log-normal distribution provided the best fit to the upper end of the uncensored data. Estimated values for all of the samples above the quantification limit were assigned by random sampling from the values determined from the fitted probability distribution.

4.3.2. Statistical Analysis The statistical analysis was conducted in two phases. In the first phase, regression models for prediction of E. coli were developed using environmental variables as the predictors (independent variables). In the second phase, regression models for pathogen levels were developed using indicator organisms (fecal coliform, E. coli, and enterococci) or environmental variables as the predictors. Because more than 80%

24

of the coliphage (USEPA Method 1602) samples were censored at the lower end, it was not included in the regression analysis. Predictive models for E. coli were considered using both logistic and traditional continuous regression models. For the logistic regression approach, E. coli values in excess of 1000 MPN/100ml were considered high, and values below 1000 MPN/100ml were considered low. This threshold value is based on the City of Madison beach closure criteria (see Section 6.3, page 42 of this document). For the traditional continuous regression models, regressions were explored using raw data and log-transformed data. Values for the three pathogens (Cryptosporidium, Giardia and Salmonella) were used together to determine if conditions existed that would warrant a beach closure. If values for any of the three pathogens were above their respective detection limit, that sample was classified as a high pathogen value (See discussion on infective doses of the pathogens in Section 6.3). If the values for all the pathogens were below their respective detection limits, that sample was classified as a low pathogen value. With data classified into two groups (low and high) it is possible to use discriminant function analysis to determine the variables or factors that best distinguish between the two groups. Numerous independent variables were chosen (Table 4-1). Table 4-1: Description of independent variables used in the discriminant function and regression Variable turb Sc Tw O2 Ht Wmax Wmin AWmax AWmin APn SRdir SRdfs SRT SRUvb Ta Pt In EI

Description Turbidity, one hour average in NTU Specific conductance, one hour average in µS/cm Water temperature, one hour average in °C Dissolved oxygen, one hour average in mg/L Significant wave height in feet Wind speed perpendicular to the beach shore in mph as measured in Middleton, WI by USGS Wind speed parallel to the beach shore in mph as measured in Middleton, WI by USGS Wind speed perpendicular to the beach shore in knots as measured in at the Dane County Regional Airport by National Weather Service Wind speed parallel to the beach shore in knots as measured in at the Dane County Regional Airport by National Weather Service Total rainfall for n-hours prior to the sample, in inches Direct solar radiation, 24 hour average in Watts/m2 Diffuse solar radiation, 24 hour average in Watts/m2 Total solar radiation, 24 hour average in Watts/m2 Ultraviolet-b solar radiation 24 hour average in milliWatts/m2 Air temperature, in °C Total rainfall, in inches Maximum n-minute rainfall intensity, in inches/hr USLE Erosivity Index

In general, logistic regression can be used to provide a model that predicts the probability of occurrence of a particular event. Typically, this involves a data set where the dependent variable has values of either 0 (no occurrence) or 1 (occurrence). The logistic regression procedure relies on the logistic transformation, given as

25

⎛ P ⎞ L = log⎜ ⎟ ⎝1− P ⎠ , where P is probability, and L is the resulting logistic transformation. Note that the logistic transformation takes a continuous probability ranging from 0 to 1 and transforms it to a continuous variable ranging from -∞ to +∞. Thus, one can write a general linear regression using the logistic function as

L = β

0

+ β1X

1

+ β

2

X

2

+ L + β

n

X

n

,

where βi are regression coefficients, and Xi are a suite of n independent variables. Retransforming the logistic function results in an expression for probability, namely

P=

1

exp[− (β 0 + β 1 X 1 + β 2 X 2 + L + β n X n )] + 1 ,

With logistic regressions, each model provides a prediction of the probability that a sample would result in a high value. In general, probability values less than 0.5 can be classified as low, and values in excess of 0.5 can be classified as high. However, in the case of pathogen data and the identification of beachclosing conditions, one is willing to accept higher errors in the identification of low values in favor of lower errors in the identification of high values. To assess the various regression models, multiple probability cutoffs (probability distinguishing between low and high values) were examined. For each potential cutoff, Bayes theorem was used to estimate the probability of correctly identifying a high event and the probability of misclassifying a low event as high (Corsi and others, 2003). The cutoff which provided a relatively high value for correctly identifying a high event while keeping the probability of misclassifying a low event to a relative minimum was chosen as the best model.

5.

QUALITY ASSURANCE/QUALITY CONTROL

The key goal of Quality Assurance was to ensure that all data are of known quality. The quality of data is generally known when all components associated with its derivation are thoroughly documented, such documentation being verifiable and defensible. Study participants followed the Quality Assurance Planning process set forth in the proposal. A Quality Assurance Project Plan (QAPP) was developed and provided a blueprint for determining data quality and details the elements of all project activities, including project management, measurement/data acquisition, assessment/oversight and data validation. The characteristics of data that measure accomplishment of a specified purpose can be expressed in terms of representativeness, comparability, precision, accuracy and completeness. Representativeness expresses the degree to which data accurately and precisely represents a characteristic of a population, parameter variations at a sampling point, a process condition, or an environmental condition. Appropriate selection of sample site and sampling procedure is critical to obtaining a sample that is representative of the environment in which it is collected. Comparability expresses the confidence with which one data set can be compared to another. Comparability of data is assured by following standard analytical procedures and calculating and reporting all data in generally accepted units. Precision is a measure of 26

reproducibility between measurements of the same property, usually under prescribed similar conditions. For this study, precision of the tests was expressed in terms of percent difference and/or standard deviation values between replicates and the ability to detect the target organism in a complex sample matrix. Accuracy is how closely the measured value compares to the true value. The accuracy of the analyses used for this study was based on statistical analysis of blank and spiked sample testing results. Completeness is a measure of the amount of valid data obtained from a measurement system compared to the amount that was expected under normal conditions. The overall QA objective for this project was to develop and implement procedures for field sampling, chain-of-custody, laboratory analysis, and reporting that would provide results that support the workplan for the project. Specific procedures were used for sampling, chain-of-custody, instrument calibration, field and laboratory analyses, reporting of data, quality control, audits, maintenance of equipment and instruments. Data verification and corrective action procedures were defined in the work plan and followed during the project. The following criteria were used to evaluate the analytical laboratory: (1) appropriate, approved, and published methods, (2) documented standard operating procedures, (3) approved quality-assurance plan, (4) types and amount of quality-control data fully documented and technically defensible, (5) scientific capability of personnel, and (6) appropriate laboratory equipment. The microbiology laboratories must follow good laboratory practices—cleanliness, safety practices, procedures for media preparation, specifications for reagent water quality—as set forth by American Public Health Association (1998). The planning process specified Data Quality Objectives (DQOs) to ensure that the type, quality, and quantity of environmental data used in decision making be appropriate for the intended beach work application. A series of planning steps included qualitative and quantitative statements derived from outputs of each step of the DQO Process that: • • •

clarify the study objective; define the most appropriate type of data to collect; and determine the most appropriate conditions from which to collect the data.

5.1.

SAMPLE/DATA COLLECTION QA/QC

Standard operating procedures and the QAPP were followed to assure that improper sample collection, sample contamination, and out-of-control analytical procedures did not cause the loss of data. A complete record of the QA plan for the State Laboratory of Hygiene and the Madison Department of Public Health laboratory was maintained in each of the participating analytic sections and is available for review as necessary. Standard USGS quality assurance/quality control (QA/QC) methods and definitions for sample collection are published in Mueller et. al., 1997. The State of Wisconsin performs laboratory audits and certification of the study laboratories. They meet the general procedural requirements for the microbial testing defined by the USEPA. To ensure collection of the highest quality data, project-specific QA/QC procedures were followed during all activities. The following are quality assurance and quality control procedures that apply to water sampling for pathogen analyses, indicator testing, water chemistry measurements and physical/meteorological characterizations of samples.

27

5.1.1. Field Data Collection QA/QC QC procedures for specific conductance, dissolved oxygen, rainfall, temperature, wind speed and direction, and chlorophyll measurements of water samples included calibrating the instruments, measuring duplicate samples and checking the reproducibility of the measurements by taking multiple readings. Instruments and equipment used to generate water quality data are calibrated according to established procedures and examined to ensure satisfactory performance. Assessment of field sampling precision and bias were made by collecting field replicates, matrix spike (MS) samples, and field and bottle blanks for laboratory analysis. If any errors were detected while proofing the field log-books, the field staff made corrections to the original entry. Field data was reported principally through the transmission of report sheets containing tabulated results of the field measurements. Measurement of Meteorological Data Tipping bucket rain gauges were calibrated for accuracy prior to field installation and were checked periodically for proper operation, debris accumulation and cleaned as necessary. Multiparameter Water Quality Measurements YSI 6000 multi-parameter water quality meters collected continuous water temperature, turbidity, specific conductance and dissolved oxygen data. USGS personnel calibrated the meters and applied corrections to compensate for drift between calibrations. Delivery to Analytic Laboratory Samples were transported in coolers on wet ice to maintain transit temperatures between 0°C and 4°C. Holding time between sampling and analysis did not exceed published standards and occured within 24 hours of collection. Date and time of sample collection and test setup and arrival temperatures were recorded by sampling staff and laboratorians alike. Continuously Monitored Data YSI 6600 series multi-parameter water quality sensors were used to continuously measure water temperature, turbidity, specific conductance and dissolved oxygen at the sites. Thermocouples were used to measure air temperature. YSI calibrations were performed on site. An extra YSI multi-parameter water quality sensor was calibrated in the U.S. Geological Survey laboratory and concurrent measurements with the extra meter and the continuous meters were made at each site and used as a basis for calibration. Corrections for drift were determined from calibration records and applied to the raw data for determination of final records. Sensor calibrations were conducted according to standardized USGS methods (Wagner, 2000). Only direct-read instruments were employed in the field. The use of specific conductance/temperature meter, turbidimeter, rain gage, anemometer, and DO probes generated direct measurements following calibration. The data were collected by data-loggers and immediately transmitted to the web page. If errors were detected, the results were reissued to the web page. Automatic Sampling Equipment Sanitation Equipment and reusable polyethylene sample bottles used for manual collection and processing of water samples were soaked in phosphate-free detergent solutions, scrubbed manually (or in automated steam washers), triple-rinsed with DI water, autoclaved and dried prior to use. For small volume indicator samples, presterilized single-use sample bottles were used. Automated samplers were programmed to run through a rinse cycle (sample tube purge and initial cycle with fresh lake water) prior to sample collection 28

to flush contaminants. Sample bottles used in automatic samplers for collection of event samples were washed in an automatic dishwasher with phosphate-free detergent at 70° C followed by a citric acid rinse followed by a triple-rinse with type II reagent water and dried prior to use. Sample splitters were cleaned between event samples by rinsing with tap water with phosphate-free detergent solutions, manual scrubbing and triple-rinsing with type II reagent water. Bottle blanks were analyzed to monitor potential contamination.

5.2.

QUALITY CONTROL CHECKS

The laboratories have a QC program to ensure the reliability and validity of the analysis performed at the laboratories. All analytical procedures are documented in writing as SOPs, and each SOP includes a QC section that addresses the minimum QC requirements and QC criteria for the procedure. The internal quality control checks might differ slightly for each individual procedure, but in general, the QC requirements include the following: • • • • • • •

Method/Preparation blanks Instrument blanks Matrix spikes Laboratory duplicates Calibration standards Standard reference materials Control charts

Any samples analyzed in nonconformance with the QC criteria were reanalyzed by the laboratory, when it was deemed necessary. The quality control acceptance criteria and spike concentrations are specified in the analytical methods.

5.2.1. Blanks Method Blanks Method blank samples were generated within the laboratory and used to assess contamination resulting from laboratory procedures. Method blanks were prepared by following the procedure step-by-step, including the addition of all reagents in the quantities specified by the method. If interferences or contamination were evident in the method blank, steps were taken to reduce or eliminate the interference(s). A method blank was run with each group of twenty or fewer samples, or as specified in the referenced method. The quality of all solutions used (including water) for preparation of culture media were analyzed according to protocols described in the Laboratory QA plan. For ‘routine’ sample analyses, the Laboratory follows the guidelines prescribed in Standard Methods for the Analysis of Water and Wastewater, 20th ed; for nonroutine tests; for nonroutine tests, laboratory personnel follow the established SOP’s documented in the analytic unit. As an example of specific QC guidelines for E. coli testing, laboratory staff demonstated proficiency in processing seeded water samples containing low levels of reference microorganisms and sterile negative control samples using the selected analytic method (enumerative ColilertTM). Hybridization assays (gene probing) were accompanied by both positive and negative controls in the form of synthetic DNA (positive control) or sterile, oxidase-treated lake water (negative control) to identify if interferences existed in sample water and also to identify carryover contamination from previous assays within the laboratory. To verify the cleanliness of the

29

media and reagents, blanks were analyzed with analytical batches. No analytes were detected from these blanks. Field Blanks Field blanks were collected to assess the quality of the data resulting from the field sampling program. Three field blanks were collected from the automated sampling stations at each site to monitor potential sample contamination through the entire sampling process, which included all equipment (automated samplers, sample collection bottles and splitters), filtering and analytical procedures. Field Blank Sampling A 4-foot extension was attached to the automatic sampler collection line at the intake point. The pump on the automatic sampler was then manually triggered to collect a subsample using the same purge and rinse cycles as a fully automatic subsample. The extension was submerged into reagent grade water when the pump was pumping forward and removed when the pump was pumping in reverse. After one10-L plastic automatic sampler bottle was filled, it was placed in a cooler, iced and transported to the USGS field office for processing. At the USGS field office, the samples were treated exactly the same as event samples; they were split into analysis containers with the churn splitter and transported to the WSLH and the MDPH laboratory for analysis. Three to five field blanks were prepared during grab sample collection at each sampling location by transferring sterile reagent water into sterile sample bottles at lakeside for processing according to the same method as field samples, i.e., bottles were rinsed, filled and iced. The blanks for the pathogen analyses were collected into 20-liter containers of ASTM Type 2 water, supplied by the laboratory, and transported along other sampling equipment during a field sampling trip. This process was used to monitor overall sample integrity and potential contamination problems of the sampling procedures. No analytical constituents were detected above detection limits in any blank samples.

5.2.2. Replicates Sources of variability and bias introduced by sample collection and environmental variable measurement affect the interpretation of concentration data. Replicate samples were analyzed to check for sampling and analytical reproducibility, as distinguished from the precision of analysis of laboratory replicates. To estimate the variability of pathogen sampling and analysis, a total of 5 duplicate sample pairs were collected for Cryptosporidium and Giardia and a total of 10 replicate samples were collected for coliphage, E. coli O157:H7 and Salmonella from automated sampling stations to evaluate precision. Automatic sampler replicates were processed during select events by filling additional bottles for analysis from the churn splitter. Replicates were delivered and analyzed at the laboratories as routine samples. Additionally, 1 to 2 replicate samples were collected for these parameters as grab samples. A total of 10 duplicate sample pairs were collected for the indicator bacteria from the automated sampling stations and 43 replicate samples were collected for the indicator bacteria as grab samples. Triplicate samples were collected for indicator bacteria once on two beaches. Replicate samples were collected sequentially at each station with minimal time between samples. Samples were processed, delivered and analyzed as routine samples with analytic variability across replicates referenced as an indication of precision within the process. The MDPH and WSLH also included laboratory duplicate analysis as part of their QA program.

30

Sample heterogeneity of bacterial distribution and small sample size collected and analyzed for indicator bacteria resulted in considerable variability among some duplicate pairs. The relative percent difference (RPD) values were calculated for the total of 49 duplicate pairs and the relative standard deviation (RSD) values were calculated for triplicates for indicator parameters from the three beaches. Six to eight percent of these values calculated for E. coli, fecal coliform, and enterococci exceeded 100 percent. Sixteen percent of the values calculated for E. coli exceeded 50 percent. Seventeen percent of the RPDs or RSDs calculated for fecal coliform exceeded 50 percent. Twenty eight percent of these values calculated for enterococci exceeded 50 percent. Typically, more homogeneous samples were obtained by the automatic samples due to the larger sample volume collected. In general, the values were higher with low bacterial counts. To estimate variability using the automated sampling station, one pair of duplicate samples was collected from each site for indicator bacteria and pathogens. A blank sample was also collected from each site to assure sterility of the sample collection.

5.2.3. Initial and Ongoing Precision and Recovery As part of the internal laboratory quality control effort, the WSLH performed Initial and Ongoing Precision and Recovery (IPR and OPR) analyses for Cryptosporidium and Giardia (EPA Method 1623) and coliphage (EPA Method 1602), including the analyses of laboratory method blanks and spike samples. Four replicate spike analyses were performed for IPR. Ongoing precision and recovery tests were performed for each sampling event. The laboratory prepared these spike samples using reagent water adding a known quantity of coliphages and inactivated cysts and oocysts and compared the results with unspiked subsamples. All IPR and OPR results were acceptable.

5.2.4. Laboratory Matrix Spike Sampling Laboratory matrix spike analyses provide information about the effects of the sample matrix on the preparation and measurement methodology. Matrix spike analysis is part of the WSLH quality assurance program as contained in the Quality Assurance Manual. Because of the variability in the performance of Method 1623, matrix spike analyses were performed to determine recovery efficiency of Cryptosporidium and Giardia from the samples that were collected from the three sampling locations. In-house matrix spikes were also performed for the male-specific coliphage (Method 1602). The laboratory identified and selected samples for matrix spike analysis among the grab samples from the three beaches. A total of six matrix spike analyses were performed for Cryptosporidium and Giardia.A total of nine matrix spike analyses were performed in duplicate for coliphage during 2003. All matrix spike analysis results were acceptable. Spike analysis data are included in the Attachemts to this report.

5.2.5. Field Matrix Spike Sampling Field matrix spike samples are quality control sample analyses that provide information about the effects of field procedures as well as the effects of the sample matrix on preparation and measurement methodology. Using automatic sampling methods, beach water was collected into a Teflon lined stainless steel churn splitter. Once the churn splitter was filled, samples were drawn off for background Cryptosporidium, Giardia and turbidity concentration analyses. The remaining volume of water in the churn was then determined and 500 heat-killed Cryptosporidium and 500 Giardia organisms were added. The churn splitter was then placed on a table in the lake where an extension was attached to the automatic sampler collection line. The churn was then used to thoroughly mix the spike. The automatic sampler was then triggered to collect a sample. The sample collection included the purge and rinse cycles which were collected in a waste container for disposal. After collection of the sample, the remaining spiked beach 31

water was added to the waste container. The sample bottle was then collected from the automatic sampler, placed in a cooler, iced and transported to the USGS field office for processing. At the USGS field office, the sample was treated exactly the same as event samples; it was split into the analysis container using the churn splitter and transported to the WSLH for analysis. Satisfactory accuracy results were achieved during all spike analyses. All laboratory reagent and matrix spike and field spike recoveries were within their respective established control limits. Spike analysis data are included in the Attachemts to this report.

5.2.6. Positive and Negative Controls Salmonella and E. coli O157:H7 Positive and negative controls were processed with each set of samples. These are cultures whose behavior on the test media is known. They are used to determine the acceptability of the media as prepared for the test being conducted. The controls are used in the laboratory to verify the identification of the constituent of concern and the ability of the method and analyst to enumerate and/or detect the organism of concern.

6.

FINDINGS

6.1.

DATA OVERVIEW AND SUMMARY

The microbiological testing and continuous monitoring resulted in a huge data base. In order to begin understanding the data it is useful to look at some summary statistics. Tables 6-1 through 6-3 present information to summarize data from each of the three beaches tested. The tables include columns for each of the microbes tested. One of the objectives of the study was to determine if changes between meteorological parameters could be linked to changes in the microbial loads, selected event periods of rainfall, high wind, wave, turbidity or high swimmer counts were assessed. To begin to understand this, the data wass first stratified by separating the regularly scheduled samples from the samples taken during events. The summary statistics in each stratification include the number of samples tested and the percent of these that tested positive. A rough idea of the levels of the organism detected is then presented by giving the minimum level detected, the maximum level detected, the median (the number where half of the results were higher and half of the results were lower), as well as the 25 and 75 percentile numbers. These summary numbers facilitate simple comparisons of the event microbe counts to the regularly scheduled samples. Scheduled samples were collected and analyzed for the indicator organisms fecal coliforms, E. coli, and enterococci five times per week and three times per week for coliphage. In addition, samples were also collected and analyzed for the pathogenic microorganisms Cryptosporidium, Giardia, and Salmonella once per week and E. coli O157:H7 three times per week. The general trend for all three beaches is that pathogens were much more likely to be detected during the events. Additionally for both Olbrich and Spring Harbor beach, the indicator counts increased substantially during events. Data from all three beaches is combined in summary form in Table 6-4.

6.2.

STATISTICAL ANALYSIS OF PATHOGEN RESULTS

Additional in-depth statistical analysis is needed to truly understand the collected data. One or more pathogens were detected in 37% of all the samples analyzed (scheduled and event samples). At individual beaches 40%, 44% and 27% of the samples were pathogen positive at Olbrich, Spring Harbor and Vilas beaches, respectively. At first glance, these high percentages of occurrence are alarming, however one needs to consider that not all pathogen occurrences are necessarily reason for concern since the level

32

detected may require very large volumes of water ingestion to reach infective doses. In this study this is especially true for Salmonella, where the levels detected were low. In addition, no information was obtained on the viability of the parasitic organisms. Review of clinical reports on patients infected with parasitic pathogens of concern did not reveal any illnesses or outbreaks related to recreational activities at the study beaches during the two swimming seasons. For the scheduled samples, one or more pathogens were detected in 28% of the samples. Of the pathogens monitored, Giardia was present most often followed by Cryptosporidium and then Salmonella (Table 64). The presence of E. coli O157:H7 was confirmed in one scheduled sample. Giardia was present more often in samples collected from Olbrich beach than those from Vilas or Spring Harbor beaches. Cryptosporidium was detected most often at Spring Harbor beach. The frequency of Salmonella’s presence was similar at Olbrich and Spring Harbor beaches but was not detected in Vilas beach scheduled samples. For event samples, one or more pathogens were detected in 56% of the samples. Of the pathogens monitored, Giardia was again present most often followed by Cryptosporidium and then Salmonella (Table 6-4). E. coli O157:H7 presence was not confirmed in any event samples. Giardia and Cryptosporidium were present more often in samples collected from Olbrich beach than those from Vilas and Spring Harbor beaches. The frequency of Salmonella presence was the same at all three beaches. Using contingency tables and a chi-square test to compare scheduled and event sample results, the occurrence of pathogens in event samples is significantly higher than seen in the scheduled samples (α=0.05) (Helsel and Hirsch, 1992). Although pathogen occurrence was greater in event samples than scheduled samples, differences at individual beaches were not necessarily consistent. The number of positive Cryptosporidium and Giardia results from samples collected at Olbrich beach was higher during event periods than from scheduled samples. The number of positive results from Spring Harbor was higher during events than from scheduled samples for Giardia but was the same for Cryptosporidium. The number of positive Cryptosporidium and Giardia results from Vilas beach was higher during event periods than from scheduled samples. Microbial indicator results for all beaches combined were generally higher in samples collected during event periods than scheduled samples except for F+ coliphage which were similar between the two types of samples. For individual beaches, indicator results are generally higher during events at Olbrich and Spring Harbor beaches, but there were no discernable differences between the two types of samples at Vilas beach. Again, the F+ coliphage results were similar between the two types of samples at each beach.

33

Giardia (per 100 L) 11 6 55% nd nd 33 67 200 Cryptosporidium (per 100 L) 11 5 45% nd nd nd 33 125

Cryptosporidium (per 100 L) 23 1 4% nd nd nd nd 267

Salmonella Quasi MPN (per 100 mL) 12 1 8% nd nd nd 0.1 1

Salmonella Quasi MPN (per 100 mL) 21 2 10% nd nd nd nd 0.2

E. coli O157:H7 presumptive (P/A per 500 mL) 10 3 30% -

E. coli O157:H7 presumptive (P/A per 500 mL) 40 14 35% -

34

Please note that per analytical method requirements, the units of volume are different for the various organisms.

# analyzed # detected % detects minimum 25th percentile median 75th percentile maximum

Event Samples

# analyzed # detected % detects minimum 25th percentile median 75th percentile maximum

Giardia (per 100 L) 23 5 22% nd nd nd nd 333

Scheduled Samples

E. coli O157:H7 confirmed (P/A per 500 mL) 3 0 0% -

E. coli O157:H7 confirmed (P/A per 500 mL) 16 0 0% -

F+ coliphage (per 100 mL) 12 2 17% nd nd nd nd 1

F+ coliphage (per 100 mL) 65 7 11% nd nd nd nd 9

Table 6-1: Summary of Olbrich Beach Sample Results

Fecal colifor m (CFU/ 100 mL) 12 12 100% 50 75 494 650 2300

Fecal colifor m (CFU/ 100 mL) 133 122 92% nd 20 70 240 1500

E. coli (MPN/100 mL) 12 12 100% 23 37 414 381 1733

E. coli (MPN/100 mL) 134 132 99% nd 13 39 130 >2419

Enterococci (MPN/100 mL) 12 12 100% 41 47 751 400 >2419

Enterococci (MPN)/100 mL 134 134 100% 1 20 53 152 >2419

Giardia (per 100 L) 12 5 42% nd nd nd 32 67 Cryptosporidium (per 100 L) 12 3 25% nd nd nd 8 400

Cryptosporidium (per 100 L) 24 6 25% nd nd nd 8 179

Salmonella Quasi MPN (per 100 mL) 11 1 9% nd nd nd nd 0.1

Salmonella Quasi MPN (per 100 mL) 23 2 13% nd nd nd nd 10

E. coli O157:H7 presumptive (P/A per 500 mL) 11 5 45% -

E. coli O157:H7 presumptive (P/A per 500 mL) 40 14 35% -

35

Please note that per analytical method requirements, the units of volume are different for the various organisms.

# analyzed # detected % detects minimum 25th percentile median 75th percentile maximum

Event Samples

# analyzed # detected % detects minimum 25th percentile median 75th percentile maximum

Giardia (per 100 L) 24 3 13% nd nd nd nd 154

Scheduled Samples

E. coli O157:H7 confirmed (P/A per 500 mL) 5 0 0% -

E. coli O157:H7 confirmed (P/A per 500 mL) 15 0 0% -

F+ coliphage (per 100 mL) 12 1 8% no nd nd nd 2

F+ coliphage (per 100 mL) 65 13 20% nd nd nd nd 5

Table 6-2: Summary of Spring Harbor Beach Sample Results

Fecal colifor m (CFU/ 100 mL) 12 12 100% 10 78 345 1160 4100

Fecal colifor m (CFU/ 100 mL) 130 116 89% nd 13 60 218 6000

E. coli (MPN/100 mL) 12 12 100% 10 102 215 1090 >2419

E. coli (MPN/100 mL) 131 130 99% nd 29 64 260 >2419

Enterococci (MPN/100 mL) 12 12 100% 12 55 209 2346 >2419

Enterococci (MPN/100 mL) 131 130 99% nd 20 55 205 >2419

Giardia (per 100 L) 11 2 18% nd nd nd nd 100 Cryptosporidium (per 100 L) 11 2 18% nd nd nd nd 100

Cryptosporidium (per 100 L) 24 2 8% nd nd nd nd 375

Salmonella Quasi MPN (per 100 mL) 11 1 9% nd nd nd nd >10

Salmonella Quasi MPN (per 100 mL) 24 0 0% nd nd nd nd nd

E. coli O157:H7 presumptive (P/A per 500 mL) 10 8 80% -

E. coli O157:H7 presumptive (P/A per 500 mL) 41 21 51% -

36

Please note that per analytical method requirements, the units of volume are different for the various organisms.

# analyzed # detected % detects minimum 25th percentile median 75th percentile maximum

Event Samples

# analyzed # detected % detects minimum 25th percentile median 75th percentile maximum

Giardia (per 100 L) 24 2 8% nd nd nd nd 194

Scheduled Samples

E. coli O157:H7 confirmed (P/A per 500 mL) 9 1 11% -

E. coli O157:H7 confirmed (P/A per 500 mL) 26 0 0% -

F+ coliphage (per 100 mL) 12 3 25% nd nd nd 0.25 13

F+ coliphage (per 100 mL) 65 7 11% nd nd nd nd 5

Table 6-3: Summary of Vilas Beach Sample Results

Fecal colifor m (CFU/ 100 mL) 11 11 100% 6 75 180 255 2200

Fecal colifor m (CFU/ 100 mL) 134 131 99% nd 60 140 315 1520

E. coli (MPN/100 mL) 12 12 100% 10 47 132 265 2419

E. coli (MPN/100 mL) 135 135 100% 6 40 108 266 >2419

Enterococci (MPN/100 mL) 12 12 100% 3 15 24 141 >2419

Enterococci (MPN/100 mL) 135 135 100% 2 16 40 83 2420

34 13 38% nd nd nd 33 200

Giardia (per 100 L) Cryptosporidium (per 100 L) 34 10 29% nd nd nd 33 400

Cryptosporidium (per 100 L) 71 9 13% nd nd nd nd 375

Salmonella Quasi MPN (per 100 mL) 34 3 9% nd nd nd nd >10

Salmonella Quasi MPN (per 100 mL) 68 4 6% nd nd nd nd 10

E. coli O157:H7 presumptive (P/A per 500 mL) 31 16 52% -

E. coli O157:H7 presumptive (P/A per 500 mL) 121 49 40% -

37

Please note that per analytical method requirements, the units of volume are different for the various organisms.

# analyzed # detected % detects minimum 25th percentile median 75th percentile maximum

Event Samples

# analyzed # detected % detects minimum 25th percentile median 75th percentile maximum

Giardia (per 100 L) 71 10 14% nd nd nd nd 333

Scheduled Samples

E. coli O157:H7 confirmed (P/A per 500 mL) 17 0 0% -

E. coli O157:H7 confirmed (P/A per 500 mL) 57 1 2% -

F+ coliphage (per 100 mL) 36 6 17% nd nd nd nd 13

F+ coliphage (per 100 mL) 195 27 14% nd nd nd nd 9

Table 6-4: Summary of All Study Beach Sample Results

Fecal colifor m (CFU/ 100 mL) 35 35 100% 6 85 180 485 4100

Fecal colifor m (CFU/ 100 mL) 397 369 93% nd 30 90 270 6000

E. coli (MPN/100 mL) 36 36 100% 10 52 132 597 >2419

E. coli (MPN/100 mL) 400 397 99% nd 23 64 215 >2419

Enterococci (MPN/100 mL) 36 36 100% 3 29 99 453 >2419

Enterococci (MPN/100 mL) 400 399 100% nd 18 48 127 >2419

6.3.

INDICATOR BACTERIA RESULTS

All indicator bacteria exhibit a great deal of temporal and spatial variability (Table 6-6 and Figures 6-1 to 6-3). Some enterococci and E. coli levels exceeded the measurement range of the test method (> 2419 MPN/100 mL). The figures do not show the censored data above that level. The maximum bacteria levels are short-lived, normally lasting less than a day at their highest levels, frequently coinciding with rainfall and high wave periods. The peaks and valleys of the bacterial populations generally follow the same pattern with differences in magnitudes. Some of the differences in fecal coliform densities may be due to Klebsiella strains that would be included in the fecal coliform results if present. The relationship of fecal coliform to E. coli and E. coli to enterococci was examined by calculating correlation coefficients (Table 6-5). Log-tranformed data were used to give less weight to individual samples that are very low or very high and give a value that is more representative of the mean. The bacterial densities were significantly correlated with each other. Relations between fecal coliform and E. coli show stronger correlations than other pairings. Of the three beaches monitored, bacteria levels at Spring Harbor Beach peaked most often at levels over the measurement range (≥2419 MPN/100 mL) of the enzymatic test system for enterococci and E. coli. followed by the levels at Olbrich Beach. Although the bacteria levels at Vilas Beach exceed the USEPA guidance density value of 235 MPN/100 mL more frequently than the levels at the other beaches, the bacteria maxima in general are lower than the maxima at Spring Harbor or Olbrich, particularly during the 2002 swimming season. A stormsewer outfall is located just east of the Spring Harbor Beach and is a likely cause of elevated bacteria levels. The near-by Starkweather Creek may contribute to the bacteria densities at Olbrich Beach. Furthermore, Olbrich and Spring Harbor beaches frequently contain considerable aquatic weeds and filamentous algae that may harbor higher levels of bacteria. The year-toyear variation was greatest at Olbrich Beach and Vilas Beaches.

38

Counts/100 mL 0

500

1000

1500

2000

2500

3000

5/28/02 6/10/02 6/17/02 6/23/02 6/26/02 7/2/02 7/8/02 7/12/02 7/19/02

39

Figure 6-1: Olbrich Indicator Bacteria 2002-2003

7/26/02 8/2/02 8/5/02 8/12/02 8/19/02 8/23/02 8/29/02 9/9/02 6/12/03 6/17/03 6/24/03 6/27/03 6/30/03 7/8/03 7/11/03 7/16/03 7/22/03 7/26/03 7/31/03 8/5/03 Fecal coliform E. coli Enterococci

8/11/03 8/14/03 8/20/03 8/24/03 8/29/03

Counts/100 mL 0

1000

2000

3000

4000

5000

6000

7000

5/28/02 6/12/02 6/19/02 6/25/02 7/2/02

40

Figure 6-2: Spring Harbor Indicator Bacteria 2002-2003

Figure 6-2. Spring Harbor Indicator Bacteria 2002-2003

7/8/02 7/15/02 7/19/02 7/26/02 8/2/02 8/7/02 8/13/02 8/19/02 8/26/02 9/1/02 6/10/03 6/17/03 6/23/03 6/26/03 6/30/03 7/8/03 7/13/03 7/18/03 7/24/03 7/31/03 8/5/03

Fecal coliform E. coli Enterococci

8/8/03 8/12/03 8/18/03 8/25/03 8/29/03

Counts/100 mL 0

500

1000

1500

2000

2500

3000

5/28/02 6/10/02 6/14/02 6/21/02 6/27/02 7/5/02 7/10/02 7/17/02 7/25/02

41

Figure 6-3: Vilas Indicator Bacteria 2002-2003

7/31/02 8/5/02 8/9/02 8/14/02 8/20/02 8/26/02 9/1/02 6/12/03 6/18/03 6/24/03 6/29/03 7/1/03 7/9/03 7/14/03 7/18/03 7/25/03 7/30/03 8/4/03 8/11/03 8/15/03 Fecal coliform E. coli Enterococci

8/22/03 8/26/03 9/2/03 9/25/03

Table 6-5: Correlations Coefficients Between Bacterial Indicators Fecal coliform Spring Harbor Vilas Olbrich 1.000 1.000 0.787* 0.713* 1.000 0.535*

Olbrich 1.000 0.617*

Fecal coliform E. coli Enterococci

E. coli Spring Harbor

Vilas

1.000 0.705*

1.000 0.558*

Olbrich n = 173 Spring Harbor n = 165 Vilas n = 165 *Significantly different from zero Correlations represent log-transformed data Table 6-6 and 6-7 present indicator bacteria summary statistics and results that exceeded beach advisories or closure criteria. As a whole, nine percent of the E. coli results were above the beach closure limit of 1000/MPN used by the City of Madison. If the guidance density value of 235/100 mL suggested by USEPA for E. coli were used as a closure criterion, the indicator results would have led to substantially more beach closures. E. coli densities among all beaches exceeded the USEPA limit in 30 percent of the samples. Enterococci densities exceeded the USEPA limit in 54 percent of the samples. The ambient Water Quality Standards established by the USEPA indicate that the single sample maximum densities should not exceed 235/100 mL for E. coli and 61/100 mL for enterococci. These values are based on epidemiological data for samples collected at chest height deep water translating into 8 illnesses per 1000 people. The City of Madison closure criterion of 1000 MPN E. coli/100 mL was established using many years of beach testing results from samples collected at depth two feet with samples withdrawn one foot from the water surface. The rationale for sample collection depth is to evaluate conditions where exposure to young children occurs. It is also well known that the bacterial densities decline with the depth of water. Table 6-6: 2002 -2003 Average Indicator Values Vilas Beach (n=165)

Average St Dev %RSD Geomean

Fecal coliform

E. coli

(CFU/100mL)

(MPN/100 mL)

253 340 135 123

255 415 163 103

Spring Harbor Beach (n=165)

Enterococci

Fecal coliform

E. coli

(MPN/100 mL)

(CFU/100mL)

(MPN/100 mL)

137 351 256 39

333 842 252 78

255 415 163 103

St. Dev – Standard deviation from the mean %RSD – Relative standard deviation

42

Olbrich Beach (n=173)

Enterococci

Fecal coliform

E. coli

Enterococci

(MPN/100 mL)

(CFU/100mL)

(MPN/100 mL)

(MPN/100 mL)

360 693 193 68

213 355 167 72

255 415 163 103

248 542 219 54

Table 6-7: Samples Exceeding Beach Closure Limits (as number and percent of all samples) Site Olbrich Spring Harbor Vilas Total

6.4.

E. coli >235 MPN/100 mL number % of total 27 20 44 50 121

E. coli >1000 MPN/100 mL number % of total 12 9

33 38 30

14 9 35

Enterococci >61 MPN/100 mL number % of total 73 55

10 7 9

79 62 214

62 47 54

PATHOGENS IN BEACH WATER AS COMPARED TO INDICATOR BACTERIA LEVELS

Another objective of the project was to determine if increases in the indicator organism levels were able to predict the presence of pathogens. While the issues of infectious dose and the effectiveness of the pathogen detection methods confound these determinations, it is the general concensus of public health practitioners that the presence of high levels of indicator organisms will predict (or indicate) the presence of pathogens. The data summarizing the predictive nature of the indicators is presented in Table 6-6. Samples were collected for the pathogenic microorganisms Cryptosporidium, Giardia, Salmonella and E. coli O157:H7. Presence of E. coli O157:H7 was confirmed in only one sample and is not included in the following analyses or discussions. A discussion of presumptive E. coli O157:H7 results are presented elsewhere in this report. During the study 108 beach water samples were analyzed for the pathogens Cryptosporidium, Giardia and Salmonella. Not all of the samples were analyzed for all of these pathogens, several were analyzed for just Salmonella or just Cryptosporidium and Giarida. In general, samples were analyzed once per week for Cryptosporidium and Giarida and three times per week for Salmonella. Forty of these samples (37%) tested positive for at least one of these pathogens and are hereafter referred to as pathogen positive samples. A quick review of table 6-8 indicates little correlation or predictive value for any of the indicators. Using a Wilcoxon rank-sum test, (Wilcoxon, 1945) there was no statistical difference (p = 0.05) between median fecal coliform, E. coli or enterococci results when the samples were pathogen positive or pathogen negative (Table 6-8). This significant finding indicates that the assumption that elevated bacterial indicator levels predict the presence of pathogens in recreational waters may be false. Table 6-8: Indicator Organism Summary Statistics when all Pathogen Results are Negative or When any Pathogen Result is Positive When all pathogen results are negative F+ coliphage (per 100 mL) # analyzed 67 # detected 8 % detects 12% minimum nd 25th percentile nd median nd 75th percentile nd maximum 13

Fecal coliform (CFU/100 mL) 67 60 90% nd 40 140 305 3000

43

E. coli (MPN/100 mL) 68 67 99% nd 29 75 289 >2419

Enterococci (MPN/100 mL) 68 68 100% nd 30 71 257 >2419

When any pathogen result is positive F+ coliphage (per 100 mL) # analyzed 40 # detected 6 % detects 15% minimum nd 25th percentile nd median nd 75th percentile nd maximum 3

Fecal coliform (CFU/100 mL) 40 40 100% 6 40 105 365 4100

E. coli (MPN/100 mL) 40 40 100% 3 29 78 237 >2419

Enterococci (MPN/100 mL) 40 40 100% 3 16 47 309 >2419

Futhermore, most of the pathogen positive samples occurred at low indicator levels (Figure 6-4). Seventy three percent of the pathogen positive samples occurred when the E. coli level was below the recommended USEPA level of 235 MPN/100 mL (Dufour, 1984 and USEPA, 1986) and 90% occurred below Madison’s current beach closure criteria of 1000 MPN/100 mL. There is no apparent difference between the indicator statistics from the pathogen positive samples and the pathogen negative samples and a high percentage of the pathogen positive samples occur at low indicator bacteria levels suggesting that the use of these organisms as a surrogate for pathogen occurrence at the three study beaches does not always accurately reflect the presence of the pathogens studied.

44

Figure 6-4: Number of pathogen positive results found in various E. coli concentration ranges

Number of pathogen positive samples in E. coli concentration ranges at three Madison beaches

Number of pathogen positive samples

14

108 pathogen samples represented, 40 positive results (37%)

12

29 pathogen positive results (73% of positive results) at E. coli less than 235 MPN/100 ml

10 8

36 pathogen positive results (90% of positive results) at E. coli less than 1000 MPN/100 ml

6 4 2 0 50

150

250

350

450

550

650

750

850

950 1750

E. coli concentration range (MPN/100ml) Note that the upper limit of th E.coli analysis method is 2419/100mL, so the values in the last range of the above histogram may be larger than they appear.

6.5.

F+ COLIPHAGE SAMPLING RESULTS

A total of 223 water samples were collected at the three beach sites for determination of F+ coliphages (those coliphages recently emanating from a warm blooded animal) between June 10, 2002 and September 25, 2003. F+ coliphages were detected in 33 of these samples that may have had multiple plaques per sample, ranging in relatively low concentrations from